Telex Corp. v. International Business MacHines Corp.

AMENDED FINDINGS OF FACT AND CONCLUSIONS OF LAW

CHRISTENSEN, Senior District Judge (Assigned).

GENERAL

Finding 1. This case involves the electronic data processing industry — an industry based upon a concept and system of reckoning (binary) as simple as turning on and off a switch; in which transmissions are timed in billionths of seconds (nano-seconds), storage capacity (memory), measured by millions of combinations of bits of information (megabytes) ; in which numerous problems involving logic or arithmetic functions are separately but simultaneously worked upon and instantly solved within a single system; in which in their own peculiar language machines communicate with one another (multiprocessing) and then in words understandable by humans may present printouts of results at the rate of as much as 2,000 lines per minute; in which devices facilitate maintenance by the detection and isolation of their own malfunctions or mistakes (diagnostic programs); upon which most other industries of the country and countless businesses, as well as science and space explorations, vitally depend; in which product and market developments seem almost kaleidoscopic when viewed from the outside; which appears unique in monopoly context by reason of its youth and apparent dynamics, but which by the same token in this ultramodern setting- may be unprecedented also because of increased inducements for,' and vulnerability to, sophisticated submarket control on the one hand, and massive industrial espionage on the other.

STATEMENT OF THE CASE — PRELIMINARY PROCEEDINGS

F2. This is an action brought by the Telex Corporation and Telex Computer Products, Inc. (“Telex”) against the International Business Machines Corporation (“IBM”) in pursuance of Section 4 of the Clayton Act ( 15 U.S.C. § 15 ) to recover treble damages for alleged violations of Sections 1 and 2 of the Sherman Act, 15 U.S.C. §§ 1 , 2, and Section 2 of the Clayton Act, 15 U.S.C. § 13 . IBM counterclaimed against Telex for alleged unfair competition, theft of trade secrets and copyright infringement in reliance upon state law and 17 U.S.C. § 101 with reference to the infringement of copyrights.

F3. Telex’s initial complaint was filed on January 21, 1972, in the United States District Court for the Northern District of Oklahoma (Action No. 72-C-18), alleging IBM’s monopolization of, and attempts to monopolize, the worldwide manufacture, distribution, sale and leasing of electronic data proc

*268

essing equipment since 1954, and seeking damages in the amount of $238,290,000, trebled, injunctive relief, attorneys’ fees, and costs. With the consent of the parties the issues and discovery for the purpose of these proceedings were limited to the United States.

F4. Concurrently with the filing of its complaint Telex moved before the Judicial Panel on Multidistrict Litigation (JPML) in the matter entitled “In re IBM Antitrust Litigation”, Docket No. 18, to transfer its case to the United States District Court for the District of Minnesota for coordinated and consolidated pre-trial proceedings with Control Data Corporation v. International Business Machines Corporation, 3-68 Civ. 312, and Greyhound Computer Corporation v. International Business Machines Corporation, 3-70 Civ. 329 (N.D.Ill. 70C 2203), both of which were then pending in that court.

1

On February 1, 1972, Telex amended its complaint to describe in more detail its monopolization claims relative to the manufacture, distribution, sale and leasing of plug compatible peripheral products which could be attached to an IBM central processing unit. On February 25, 1972, Telex’s motion to consolidate was argued before the JPML.

F5. On March 15, 1972, while its motion for consolidation was pending before the JPML, Telex filed a second complaint in the Northern District of Oklahoma (Action No. 72-C-89) alleging that IBM had violated Section 2 of the Sherman Act ( 15 U.S.C. § 2 ) by announcing its “Fixed Term Plan” in May of 1971 and its “Extended Term Plan” on March 1, 1972. Telex sought a temporary restraining order and preliminary injunction from the Oklahoma court. On April 19, 1972, the JPML issued orders transferring the Telex actions to the Minnesota court, Telex Corp. v. International Business Machines, Inc., D.C., 342 F.Supp. 200 (1972), and Honorable Philip Neville, United States District Judge for the District of Minnesota, was assigned to handle complicated discovery and other matters preliminary to trials. Under his able supervision millions of documents were discovered or exchanged and photographed, and various procedural rulings made. On June 12, 1972, Telex filed a supplement to its complaint, alleging violations of the antitrust laws by IBM in the then soon-to-be-announced IBM System 370/168 and 370/158 central processing units (CPU) with integrated CPU memory and integrated disk control circuitry and a lower priced incremental memory. Telex sought injunctive relief preventing IBM from integrating any memory or disk control circuitry into its System 370 central processing units and from lowering its prices for memory incremental to the CPU memory.

F6. On July 21, 1972, in partial response to a motion by Telex, the Minnesota court granted a temporary restraining order enjoining IBM from making any announcement of its 370/168 and 370/158 central processing units until the Minnesota court’s decision on Telex’s pending motion for preliminary injunction was entered. IBM sought relief from the Minnesota court's action in the Court of Appeals for the Eighth Circuit (Docket No. 72-1447) both by way of appeal and extraordinary writ. That court on July 28, 1972, determined that the temporary restraining order entered by the Minnesota court was tantamount to the issuance of a preliminary injunction because it exceeded the ten day limitation set forth in Fed.R.Civ.P. 65(b). The Telex Corporation v. International Business Machines Corporation, 464 F.2d

*269

1025 (8th Cir. 1972). It was ordered that the preliminary injunction be dissolved because the district court had made no findings relative to the ultimate probable success of Telex on the merits or on Telex’s claim of irreparable injury. On July 28, 1972, Telex moved for a second temporary restraining order to be limited to ten days. That motion was denied on August 1, 1972. On October 6, 1972, the Minnesota court, after the submission of affidavits, evidentiary appendices and briefs, denied Telex’s motion for a preliminary injunction as well as IBM’s motion for summary judgment. After extensive pre-trial discovery on both sides in the Minnesota proceedings, Telex finally amended its complaint on January 2, 1973, to demand damages in the amount of $416,100,000, trebled, and on January 8, 1973, filed an amended consolidated complaint.

F7. Telex moved on January 9, 1973, that its cases be remanded to the Northern District of Oklahoma for trial. On January 10, 1973, the case was assigned to this Judge for final pre-trial preparation and trial. Telex’s remand motion was granted on January 15, 1973, and as of that date, subject to the Minnesota district court’s retention of jurisdiction on certain privilege issues, which have now been finally resolved, this court obtained jurisdiction of these proceedings. On January 22, 1973, IBM answered Telex’s amended consolidated complaint and filed two counterclaims — one alleging unfair competition and theft of IBM trade secrets, and the second alleging Telex’s infringement of IBM copyrighted manuals.

F8. Pre-trial conferences were held before this court on February 20, March 30, and April 13, 1973. At the final pre-trial conference a jury, previously demanded on both complaints and the counterclaims, was waived by both sides. This court entered a final pre-trial order, based largely upon the March 30 conference, on April 12, 1973, which enumerated in detail the contentions of the parties, stipulated and disputed factual matters, the documentary evidence intended to be offered, and the witnesses to be called by the respective parties.

F9. Trial commenced on April 16, 1973, and the record was closed on May 24, 1973, after 29 days devoted to the taking of evidence. The expedition of the case consistent with full, fair and vigorous presentations was due in important measure to the ability, organizational talent, diligence and experience of counsel, together with the routine informal conferences held each morning before the convening of court among court and counsel where evidentiary problems were anticipated, presented and explored and the management of proceedings was otherwise charted from day to day. A brief post-trial conference was held on May 25, 1973. The parties submitted their separate proposed findings of fact and conclusions of law in compliance with the request of the court, and on June 18 and 19, 1973, oral arguments were had. Whereupon the case was submitted for decision and by the court was taken under advisement. On September 17, 1973, findings of fact, conclusions of law and judgment and decree were filed. On October 16-18, 1973, timely post-judgment motions for correction and amendment thereof were argued to the court and submitted for decision on supplemental briefs.

The court, now deeming itself fully advised, makes the following amended Findings of Fact in addition to the statement of the case and proceedings set out above, and the minute stipulated facts not set out, including but not limited to the (i) Stipulation of Background Facts Concerning the Electronic Data Processing Industry, the Products, the Industry, the Parties and the Issues, dated March 23, 1973, (ii) Stipulation of Fact No. 2, dated April 12, 1973, (iii) Stipulation of Fact No. 3,'dated April 7, 1973, and (iv) Defendant’s Exhibit 1662 (an exhibit compiling plaintiffs’ admissions introduced at trial), but incorporated herein by reference since there is no contest with respect to them.

*270

I

THE PARTIES

F10. International Business Machines Corporation (IBM) was incorporated on February 24, 1924, and maintains its principal place of business in a state other than the State of Oklahoma where plaintiffs are incorporated and maintain their principal places of business. Before its entry into the electronic data processing (EDP) industry, IBM manufactured punched card accounting machines and other products. In addition to its EDP business, IBM develops, manufactures and markets other business machines including copiers, dictating equipment, and electric typewriters. IBM has been deeply involved in the phenomenal growth of the electronic data processing industry since almost the beginning of the industry. IBM’s first EDP system offered for sale was the IBM 701. The first IBM 701 was completed in April of 1953 and was intended primarily for scientific work in connection with nuclear research. The first IBM computer intended for commercial work was the IBM 702, which was installed in early 1955.

Fll. The Telex Corporation was incorporated in February, 1963. It is the successor to Telex, Inc., which was incorporated in May, 1940. Since at least 1959 Telex has been manufacturing products which have been used by electronic data processing equipment manufacturers as part of their equipment.

F12. Because of the success of IBM’s System 360, certain companies such as Telex entered into the marketing of devices functionally equivalent to IBM devices. The devices marketed by Telex and others plugged into and replaced parts of the System 360. In 1966 Telex began to market replacements for the magnetic tape devices which were part of IBM’s System 360 computers. Before the receipt of a contract with DuPont to replace DuPont’s installed second-generation IBM magnetic tape devices, Telex had been conducting engineering development work to modify the Telex Model M3000 — a magnetic tape drive then being marketed by Telex to other EDP equipment manufacturers— to provide an appropriate electronic interface to attach to an IBM central processing unit (CPU). An interface is a shared boundary between electronic data processing machines, or more accurately between the channels or physical pathways connecting those machines, through which data or programs may be transmitted, received, stored or processed. After receipt of the DuPont contract, the work was completed and the equipment delivered in August, 1966. The Telex-developed machine was designed with an electronic interface to work in conjunction with an IBM CPU. It was designated as Telex’s Model 4700. In late 1966, additional 4700’s were installed at Lockheed Aircraft and at Electronic Business Service (AMI). The total engineering cost for designing the electronic interface necessary to adapt Telex’s tape drive for use with IBM equipment was $42,000.

F13. In May, 1969, Telex began to market replacements for IBM’s disk drives. Telex does not manufacture but purchases the disk drives and disk drive controllers from another company. In November, 1970, Telex announced that it would begin marketing a printer and a printer controller. The printer mechanism is not manufactured by Telex but is purchased from Control Data Corporation. Telex manufactures some of the printer electronics and the printer controller. In November, 1971, Telex announced it would offer a replacement for the memory or main storage used with certain IBM systems. In November, 1971, Telex announced a 6360 memory for attachment to IBM Systems 370/155 and 370/165. That memory was first delivered in November of 1972. Telex purchases the parts of the memories from various corporations. It assembles the parts and markets the final product. As of January 26, 1973, Telex had installed two memories. In November, 1971, Telex announced a 6345 memory for attachment to an IBM System

*271

370/145. Telex has not yet delivered any 6345 memories. Telex has recently advertised its intention to announce a memory for attachment to IBM Systems 370/168 and 370/158. Telex’s forecast for first customer delivery for these products is the fourth quarter of 1973.

F14. Telex has never announced or delivered a “communications controller” but is presently developing a plug compatible communication controller equivalent to IBM’s 3705. Telex has negotiated an agreement with Hitachi, a Japanese corporation, under which Telex will engage in a joint development effort to develop a .CPU compatible with IBM’s System 370 and competitive with System 370 Models 135 and .145. No final decision has yet been made as to the actual manufacturing and marketing of such a product.

F15. In the aspect of its business relating to the marketing of EDP products to IBM end-users, Telex in the past has had a company policy generally of following IBM’s product leadership and subordinating any technological product innovation. Telex products are designed as the functional equivalent of previously announced IBM products, except for whatever technological advances Telex is able to introduce because of the later announcements of its products. Telex’s plug compatible tape drives, disk drives and printers have had better performance in some respects than IBM’s corresponding products.

F16. Since entering the EDP indusr try and up to 1971, Telex reported a phenomenal growth in revenues. Its revenues from EDP products and services sold to customers within the United States as reported in the “census”

2

rose from $870,000 in fiscal 1967 to $56,-840,000 in fiscal 1971.

II

THE INDUSTRY

F17. The electronic data processing (EDP) industry is a young and dynamic one ranking high in importance among the industries of the nation. The first commercially built EDP system — the Univac I — was delivered in 1951 to the Bureau of the Census. The demand for EDP products and services as indicated by the revenue of companies responding to the court census has grown from $48 million in 1952 to $10.2 billion in 1970.

F18. Electronic data processing is employed by government and major producers of goods and services throughout the country to make their operations more efficient and to provide new and better products and services. The scientific community has used and is continuing increasingly to use electronic data processing extensively. The kinds and types of available products, equipment or services used are determined by the applications for which there is a need and the available resources to meet the need. Examples of electronic data processing applications are almost as numerous as business and scientific applications: The computer can keep track of enormous numbers of people who have made reservations with the various airlines and tell passenger reservation agents when planes are full. EDP systems control many manufacturing processes and almost entire factories, monitor patients with severe heart disease in hospitals, and control some navigational systems on airplanes. Such systems are used in printing newspapers, controlling

*272

traffic lights, guiding ships, navigating and controlling space missions, and even in designing other computers. Less dramatic but widespread are day by day business applications affecting the lives and fortunes of almost every individual in the country in one way or another. There has been a marked increase in the sophistication of EDP customers in the last few years. Immediate purchasers of EDP products and services are most often large institutions such as the United States government, universities, or large industrial organizations. Most of the EDP systems are installed in the five hundred largest governmental and business organizations. Many professional consultants offer assistance in the design of EDP systems and the procurement of EDP products and services, thus enabling smaller users to make more knowledgeable decisions concerning their EDP needs.

•F19. Dramatically increasing demands for EDP products and services and the needs of EDP users have resulted in a rapid growth in the number of companies which offer EDP products and services and in a variety of products and services which are offered to accomplish the data processing needs of users. Many different kinds of companies have been attracted to the EDP industry. The number of companies responding to the census and reporting EDP revenue in each year from 1952 to 1970 has grown from 13 to 1773, a growth in number of more than 136 times in eighteen years. According to the census, the number of companies which manufacture and market a complete EDP system has grown from 3 in 1952 to 96 in 1972, but only 8 or 9 of these companies are considered in the trade as principal systems manufacturers. Thesé include IBM, Univac (Sperry Rand), Burroughs, Control Data Corp., General Electric, Honeywell, RCA, XDS (Xerox), NCR, and Digital Equipment Corporation (DEC). Recently RCA and GE have gone out of the systems business. The products and systems formerly manufactured and marketed by RCA and GE remain in the market and are now being maintained, serviced and remarketed by Honeywell and Sperry Rand. As a result Honeywell and Sperry Rand probably have been strengthened and their ability to compete with IBM enhanced by their acquisition of the computer operations of RCA and GE.

F20. Spurred additionally by the su-cess of IBM’s System 360, manufacturers of certain peripheral devices began in 1966 to market to end-users products which were functionally equivalent to certain IBM System 360 devices. The equipment of these companies “replaced” IBM devices and utilized all the system’s support and services provided by IBM. in 1966 Telex, and shortly thereafter other independent manufacturers, began manufacturing and operating magnetic tape drives which were functionally equivalent to IBM magnetic tape drives and which could be “plugged” into, and thus were “plug compatible” to IBM central processing units. The end-user customer of an IBM computer system then had the option to use an IBM tape drive or to use one made by a “plug compatible manufacturer” (PCM). Peripheral equipment manufacturers have expanded their peripheral product lines, moved into the leasing of complete EDP systems, and certain of those companies, including Telex, Memorex and Mohawk Data Sciences, are now expanding, or considering expanding, into the manufacture and marketing of their own central processing units.

F21. In the mid-1960’s leasing companies began purchasing computer products from IBM, which they then leased to computer end-users. These companies purchased $2.6 billion worth of IBM’s 360 computer hardware for which IBM received its full retail price and corresponding profit. Leasing companies typically purchase an installed IBM computer system and then lease it to the existing end-user at a rate less than IBM charges for its identical products. When an initial lease ends and the machine leaves the first end-user’s shop, the leasing company owner remarkets

*273

the "used” equipment when possible. During the late 1960’s a substantial quantity of the initial leases relating to the $2.6 billion worth of IBM 360 computer equipment mentioned above were expiring and during the late 1960’s and early 1970’s leasing companies were engaged in remarketing that EDP equipment. Another type of leasing company transaction involves the purchase of plug compatible tape and disk drives from PCM’s after they are installed and on rent in an end-user’s location. This type of transaction is primarily a method of financing and the leasing company depends upon the manufacturer to market and service the product.

F22. A service bureau owns or leases computer products and/or services and then performs data processing services for customers for a fee. The customer can get data processed by this method without owning or leasing any specific EDP “hardware” or “software”. A time-sharing company is one that installs a terminal facility in the customer’s business location; the terminal is connected to the time-sharing company’s computer system via telephone communication lines. The end-user can then time-share the computer system by means of the remote terminal for a fee. A data center is an establishment having a computer installation which permits customer personnel to operate the computer equipment for a fee. Software houses prepare and market computer programs or instructions designed to cause the central processor and peripheral products to perform their required functions. Examples are instructions that will cause data from input devices to be transferred to storage devices, to be retrieved when needed, then processed in a usable form. Facilities management companies, or system engineering consultants, such as Computer Usage Corporation, provide the customer with systems engineering and design services as well as services for the actual operation of the end-user’s computer facilities.

F23. The speed, reliability and capacity of computer products have increased greatly since 1952. One of IBM's CPU’s the 370/168 (announced but as yet undelivered) when compared to the Univac I will have 700 times the storage capacity of Univac I, and it will execute additions 4,300 times faster, multiplication 3,100 times faster, and division 2,000 times faster. The data transfer rate of current tape drives is 40 times greater than that of the earliest tape drives used with the Univac I. Memory technology has increased cycle speed of main memory devices a thousandfold since 1952. Electronic circuitry improvement permits products to be made today which were difficult to conceive a few years ago. Speed, capacity and reliability have improved, while power requirements have dropped.

Ill

THE NATURE OF ELECTRONIC DATA PROCESSING

F24. Electronic data processing (EDP) is the conversion of words, letters, numbers or combinations of words,, letters and numbers, or other types of data, into electronical signals; the data is then collected, stored, sorted, analyzed, compared or computed. The “hardware” products and “software" programs that perform these functions are often referred to collectively as a computer system, or simply as a computer. Computing may involve both simple and complicated calculations, or the storing and sorting of large amounts of data. An example of the complicated calculations of computers is the work done at the Manned Spacecraft Center which links computer systems throughout the United States to computers on board spacecrafts to perform large numbers of precise, complicated calculations. An example of storing, sorting and comparing large amounts of information is an airlines passenger reservation system, or a warehouse inventory system.

*274

F25. An EDP system consists of products which perform five basic functions. These are “processing”, “storage”, “input”, “output” and “control”. Input is the entering of data into storage. The input devices convert data from an “ordinary” language form (i. e., English and numbers) to “machine” language or electronic signals which are then understandable to a computer. Output is the opposite. Output devices convert the “machine” language or electronic signals to the output form desired, such as printed or typed in humanly understandable language on paper, recorded on magnetic tape or magnetic disk, punched as a hole in a punched card, or displayed on a television-like screen. Output devices can also be used to open or close a valve,, or to transfer electrical impulses to another computer system. Storage of data is accomplished in either the main memory or some type of auxiliary storage.

F26. Main memory is the storage from which data are transferred to the processor and to which data are returned in their processed form. Auxiliary storage is the storage from which data are interchanged with the main memory for processing, temporary transfer, or more permanent storage. Auxiliary storage is usually accomplished in some one or more of the following: Large core storage (LCS), data cells, magnetic drums, magnetic disk devices, magnetic tape devices, paper tape devices and punch’ cards. The type of auxiliary storage used is dependent upon the applications and needs of the customer with reference to the stored data.

F27. The processing function is the computation or performance of logical operations. These logical operations involve additions, subtractions, and comparisons. The logic is composed of simple steps done rapidly to achieve the ultimate results. A control function enables a computer system to perform a large number of consecutive instructions. The control function can usually understand or evaluate the various operations as they are concluded and perform alternate operations without human intervention based upon such evaluation. The control function directs and coordinates the operation of the various products making up the system and can be performed by a combination of hardware, microprogramming and software. Programs are sequences of instructions which tell the various devices what to do. Programs are also referred to as software.

■ F28. A modern computer system is composed of a variety of individual devices each of which usually performs a different function that may be needed to perform a particular needed application. The user may select from various products the particular combination of individual devices and software which will solve the customer’s data handling needs, taking into consideration the economics and applications involved. The user’s choice of alternative devices may depend upon trade-offs among price, capacity, speed, flexibility, space requirements, and the number, kind and priority of applications to be performed and users to be serviced, but inherent are various limitations of function and application which as a practical matter most often dictate a particular device for a particular application.

F29. Individual input/output products include teletype machines, typewriter terminals, television-like displays which use cathode ray tubes, card punches and punched card readers, magnetic tape drives and magnetic disk drives. There, are also devices which read magnetic characters or optical characters such as those on checks, and there are devices which read coded tags on merchandise. When used in stores, these devices automatically record the sale, bill the customer, and remind the store to reorder.

F30. Printers perform an output function. Like a typewriter terminal, a printer converts electrical signals into printed characters and numbers. Printers, however, operate at much higher speeds than typewriter terminals. Mechanical impact printers operate at up to

*275

2,000 lines per minute while electrostatic printers which operate similar to copying. machines, can operate at higher rates. There are special purpose printers used to produce graphs, charts, drawings and maps. Other output devices produce microfilm. These are called computer output microfilm, or COM, devices.

F31. Products which perform a storage function include magnetic core arrays, semiconductor circuitry, magnetic tape drives, magnetic strip files, magnetic drums, and magnetic disk drives. All these devices, except the semiconductor circuitry, store data by converting electrical signals into a .magnetized recording that can be reconverted into electronic signals. A magnetic recording is permanent in that it remains when electric power is “off”, but the semiconductor circuit loses its stored data when the electric power is “off”.

F32. The CPU is generally where most of the logical functions or calculations are performed. Controllers, channels, peripheral processors and multiplexors are smaller processors designed for a particular use, and when used permit a more efficient use to be made of the central processor by speeding up interchanges of data and making preliminary or intermediate computations for relay to the CPU.

F33. EDP products are built from electronic and electromechanical components. The components include electronic circuits, devices for converting electrical impulses to magnetic, devices for converting magnetic impulses to electrical, devices for converting electrical current to mechanical movement, as well as cables, connectors, metal frames and various power and cooling elements. The most numerous physical parts of an EDP system are the electronic circuits. Electricity, as used in a computer, essentially has only two states or conditions —it is either “on” or “off” as is the case of an electric light. By combining electronic switches which are on or off, computing can be done if “on” equals 1 and “off” equals 0. Different combinations and sequences of l’s and 0’s then can be used to represent all numbers and all letters. When electronic data processing began, each electronic circuit was made up of a vacuum tube, such as is used in radio or television, plus wires and resistors. The development of transistors in the 1950’s allowed the vacuum tubes to be replaced by transistors.

F34. IBM was an early user of the transistor in its EDP systems. IBM built its own factories to make transistors. The use of transistors made possible the reduction of size, cost and power requirements of an EDP system and increased reliability and speed. This allowed the construction of EDP systems of greater capacity and operational speed and expanded the number and types of applications for which such equipment could be used. As work continued on the refinement of the transistor at places like Bell Laboratories, Texas Instruments, Motorola, Fairchild, and IBM ways were found to combine the various components making up an electronic circuit into a single chip, which is now about % inch square. This chip is called an “integrated circuit”. In the 1960’s, IBM as well as others began to build EDP systems using integrated circuits. This allowed a further reduction in size, a further increase in reliability, a further increase in speed and a further reduction in cost. Work at IBM and other places has led to the continual miniaturization of the circuits. It became possible to produce multiple circuits on a chip. This was referred to first as “medium-scale integration” and later, as the number of circuits increased, “large-scale integration”. In the latest EDP equipment, components are in use which have more than two thousand circuits on a single silicon chip % inch square. Under development in IBM and other laboratories are chips containing 16,000 circuits. Moreover, there are under development processes which, it is believed, will produce chips with 64,000 circuits or more.

*276

IV

RELEVANT MARKET

F35. In determining whether there is monopoly power to control prices or exclude competitors in any part or line of commerce, the court is required to consider a relevant market or markets within which such determination can be made. Manifestly, the electronic data processing market in general is one relevant to such an inquiry. But the fact that monopoly power may not exist on the part of any company within that general market as a whole does not end but only begins the inquiry in this case. It should also be noted that we are not primarily concerned with prior or subsequent years, but that in view of the issues of this case a determination must be made as to the relevant market or markets in the period 1969-1972, timing also being an important element here because of the youth and dynamics of the market and its various developmental stages over the years. It is recognized that a purely transitory condition could be so brief or insubstantial as to be

de minimis

or immaterial in appraising market power; but it must also be recognized that in a real sense every market condition may be temporary in the perspective of historical development, and yet the policy of the antitrust laws does not permit the unlawful application of monopoly power against competition to its damage over a substantial period even though, if competitors could hang on for a time, technological or other developments might change the competitive situation for the better.

F36. Telex asserts that in the period mentioned IBM possessed monopoly market power in the general systems (CPU) “relevant market’’, in the market for peripheral devices plug compatible with IBM CPU’s, and in the “relevant sub-markets” for magnetic tape products, direct access storage products, memory products, impact printer products, and communication controllers that were plug compatible with an IBM CPU. IBM claims that it had no monopoly power in any such general markets and that submarkets did not exist because competition in the EDP industry was primarily on a systems basis, and that the relevant market consisted of EDP systems and the products which make up such systems and the companies which provide alternatives to such systems. IBM further claims that even if one were to limit the focus to particular parts of a system, such as peripherals, the relevant market must include all peripheral products, not just those currently attached to IBM systems. It is further contended by IBM that once the decision is reached that the relevant market should include peripherals attached to competitive systems, as well as those attached to IBM systems, it does not make any difference with respect to IBM’s share whether the market is limited to such peripherals or is broadened to include all products which make up systems. IBM further claims that even if “plug compatible” tapes, disks, printers, communications controllers, and memories did constitute separate sub-markets, if the decision were made to include disks or any of the other products attached to known IBM systems, as well as those attached to other than IBM systems, IBM’s share of each of these submarkets would be well below the level that would support any inference of market power.

F37. The potential general market toward which the efforts of both companies seem directed, with the progressive broadening of Telex’s base and the technological and industrial developments in prospect, appears substantially the same, and the real issue is whether that market may be realistically subdivided in the time frame 1969-1972 to focus on and encompass only those parts of current product lines which are respectively attached to IBM systems, rather than all those products which actually have similar uses in connection with other systems ; although, with respect to the claimed

attempt

to monopolize, these distinctions may not be critical. By definition every manufacturer has 100% of its

*277

own product. Thus, where a hopeful competitor first offers a product as a substitute for the original, the originat- or typically will continue to have a large share of that product. So, likewise, in the EDP industry each manufacturer of systems normally has a large percentage of the peripheral equipment, which is part of its system. By this token a systems manufacturer has 100% of the peripherals attached to its new system until someone begins to copy some or all of these peripherals or designs others to take their place on a plug compatible basis. It is an oversimplification to say, however, that under Telex’s market definition theory, as soon as someone begins (or perhaps even plans or thinks about) copying a part of a new system, as IBM argues, the manufacturer of that system becomes a monopolist and has an obligation not to cut prices or do anything else that might reduce the profitability of the copier. The record in this case shows that peripheral devices attached to IBM equipment but manufactured or supplied by others during the relevant period have grown into, and have been recognized as, a significant, distinct and important part of the EDP industry. Again, for the particular period mentioned, we are not dealing with mere theory but with a historic, economic fact, transitory or otherwise. The question persists, however, whether such suggested subdivisions of the industry can properly be regarded as relevant markets or submarkets within which economic power can be separately appraised. A related dilemma must be avoided by at once precluding the unreasonable fragmentations of markets

3

and preventing the monopolization of separately competitive components while a whole industry is thus subverted part by part.

F38. Peripheral products constitute an important part of a data processing system, accounting for 50-75% of the price of the system. Such products are critical to the performance of the system as a whole. It cannot be gainsaid that indirectly at least and to some degree the peripheral products attached to non-IBM systems necessarily compete with and constrain IBM’s power with respect to peripherals attached to IBM systems. The quality and price/performance of the peripherals attached to a system are a substantial factor in a customer’s choice between competing systems, and if for example IBM failed to improve the price/performance of its peripherals, customers might choose systems (including peripherals) of other systems manufacturers. For example, the IBM Merlin (3330) disk drive was believed by IBM to be a critical factor to the competitive price/performance of the 370 systems 135, 145, 155, 158, 165 and 168. The 3330 was therefore announced in June of 1970 at a price/performance designed to make IBM more competitive with both systems manufacturers and peripheral equipment manufacturers. The Court finds that the document entitled “Listing of Manufacturers of Plug Compatible Products, The Products Offered and the Systems Manufacturers for Whose Systems the Products are Offered” (attached to “IBM’s Response to Some of Telex’s Proposed Findings of Fact and Conclusions of Law Relating to Telex’s Antitrust Claims” dated June 15, 1973) is a summary of the details there stated. Peripheral pricing and product announcements of one systems supplier influence subsequent peripheral pricing and product announcements of other systems suppliers, although it may be difficult to identify any given competitive price cut or product improvement as

*278

a reaction to a single competitive act. Many companies, including Telex, which manufacture or market peripheral equipment for attachment to IBM CPU’s also manufacture or market equipment for attachment to non-IBM CPU’s, but to a substantially lesser extent. Moreover, suppliers of peripherals plug compatible with non-IBM systems could in various instances shift to the production of IBM plug compatible peripherals, and vice versa, should the economic rewards in the realities of the market become sufficiently attractive and if predatory practices of others did not dissuade them. In the absence of defensive tactics on the part of manufacturers of CPU’s, the cost of developing an interface for a peripheral device would generally be about the same regardless of the system to which it would be attached, and such cost has not constituted a substantial portion of the development cost of the peripheral device.

F39. In this extraordinary industry dominated as it has been by IBM’s influence, neither theoretical relationships nor technological similarities supply the full answer to the relevant market problem. In the realities of the marketplace, as recognized and acted upon by IBM as well as by the plaintiffs and their customers, it must be determined (a) whether plaintiffs’ concept of relevant markets keyed peculiarly to devices plug compatible with IBM CPU’s is sound, and (b) whether there is sufficient demand or supply interchangeability, sub-stitutability or flexibility as to render indistinct or ineffectual the lines dividing the submarkets relied upon by plaintiffs as among themselves or as between them and general EDP systems. A differentiation between the IBM plug compatible peripheral market in general and submarkets involving particular types of such peripherals seems not so critical, since it appears likely that IBM’s market power would not significantly vary as between them.

F40. IBM and other systems manufacturers design, develop, manufacture and market system solutions to data processing problems on a systems basis primarily, although with respect to particular applications the suitability of particular peripheral equipment may be emphasized. In designing a system, IBM and other systems manufacturers must design the boxes comprising the system, the configuration of boxes in the system to provide the best solution to a particular set of requirements, the system software essential or helpful for the operation of the hardware generally, and the particular applications software to perform the customer’s special applications. The reliability and predictability of the system involve the hardware, software and the personnel maintaining and operating the system. There are significant expenses involved in designing systems so that the various boxes can be integrated into different configurations and combinations. There are a number of other systems development costs not easily identifiable because of difficulty in segregating an engineer’s time between developing a particular unit and working on its integration into a system. Similarly, systems marketing costs are hard to define because of the difficulty of separating the time a salesman spends configuring a system from his other activities. Particularly with respect to new customers, systems manufacturers offer a substantial amount of EDP education, which is essential in order to market systems. Systems development and marketing costs are allocated across all of IBM’s products and are included in the pricing of those products. Some customers can perform completely their own systems integration work in view of the level of sophistication among them, but others rely upon IBM or other systems suppliers to do this work and to provide systems control programs. Telex and other peripheral manufacturers do not incur substantial systems development and systems marketing costs in connection with their plug compatible business, nor could they market their products for attachment to IBM systems without IBM system software to which is devoted about 30% of

*279

IBM’s annual development cost. Thus there are practical and logical difficulties in serving the peripheral market from the systems market.

F41. It is true also as a generalization that to a substantial degree each of the different functions of a system can be and is performed by a variety of devices and that users not infrequently can choose among different devices which make up an EDP system on the basis of priee/performance and the particular applications desired. This interchangeability, however, is between particular peripheral devices for particular applications, and in and of itself does not render particular devices a necessary part of a systems market. Rather, it raises the question whether all or a portion of the peripheral devices are a part of the market for peripheral devices. Merely because there are alternate ways of storing data in an EDP system, each of which competes to a degree with others in various applications, does not mean that it is appropriate to consider storage products such as tapes, disks and memories and their substitutes as a part of the systems market rather than part of a peripheral device market.

F42. Devices which perform a storage or memory function include core arrays, semiconductor circuitry, magnetic tape drives, magnetic strip files, magnetic drums, and magnetic disk drives. Each of these devices has a different operational speed and a different cost and, depending upon the needs and budget of the user, each can be used in structuring the computer in different ways to: a limited extent. Constrained by particular applications, needs and objectives, these devices compete with one another in a limited sense and in some applications users can employ different devices interchangeably. An EDP user might “trade off”, for example, the higher performance of memory for the lower price of disks in certain applications, whereas for other applications disks and tapes could perform similar functions and be used interchangeably. A user -might trade off the higher performance of magnetic disks or drums for the lower price of magnetic tapes on some applications, and in general, but still in a limited sense, users may have price/performance alternatives or trade-offs among disk drives, memory, tape drives, the tape library, the vault, the disk pack, etc. in configuring any total EDP system. In a limited sense, too, certain storage devices such as memory are interchangeable with the CPU itself, users choosing between larger or faster CPU’s with relatively small amounts of memory and smaller or slower CPU’s with relatively large amounts of memory; and in certain instances CPU’s function as peripheral devices, and peripheral devices or parts of peripheral devices have similarity to CPU’s. Terminals which perform input and output functions also have processing functions, storage functions and control functions. Intelligent terminals perform processing functions otherwise performed by communications controllers or central processing units. Most magnetic core storage, magnetic tape drives, magnetic disk drives and magnetic drums contain some processing control functions. Printers, like Telex’s 5848, can also perform processing storage functions with a controller and magnetic tape drives..

F43. The users also choose between terminals, printers and computer output microfilm devices as various means of accomplishing an output function depending upon the various needs and applications involved. Special kinds of printers, called plotters, can be used to produce graphs, charts or drawings, and even maps. Other types of printers called computer output microfilm (COM) devices produce microfilm. A teletype is one kind of slow speed printer. Other kinds of printers include drum printers, which can be slow or high speed (from 300 to over 1000 lines per minute), chain printers and train printers. Less expensive slow speed printers are alternatives to more expensive higher speed printers. Terminals are also used to perform output func

*280

tions. Electrical signals are converted into words and numbers on a display screen or typed on a roll of paper. Many terminals also have processing functions, storage functions and control functions. Computer output mireofilm may be a direct competitor to printers because of relative hardware costs as well as cost of paper versus cost of microfilm, and some customers have replaced or are replacing printers with COM equipment on a price/performance basis.

F44. An essential element of any electronic data processing system is the control function. In large portion the control function is performed by software or programming. The cost of developing operating system software is substantial and competition in the supply of better operating systems necessarily affects the price a manufacturer can charge for its EDP systems. But it is true only in the superficial sense that software can be used as a direct substitute for hardware, although cost of certain hardware or the extent of its necessity may be affected by the software.

F45. Some suppliers of peripheral devices can and do become suppliers of systems and the suppliers of full systems can and do supply peripheral devices plug compatible to the CPU’s of other manufacturers in some instances. Suppliers of peripheral devices, including Telex, either have planned or are considering movement to full systems. Texas Instruments, which began as a supplier of components for EDP and other electronic purposes, now markets the world’s fastest CPU, and Memorex, which began as a supplier of peripheral devices, announced two full EDP systems in 1972. Suppliers of full systems, including IBM, can and in some instances do provide peripheral devices for use with the systems of other manufacturers either directly to end-users or to other system manufacturers. Manufacturers of CPU’s and peripherals use to a substantial degree the same technology, making it technologically practical given time, funds and personnel, to switch from one to the other. In the long range this potential “supply substituta-bility” has had and will have substantial effect upon the development of the market and upon trends of competition, but during the period with which we are concerned supply substitutability was a minimal factor in the marketplace as a constraint upon pricing. It was a fact of economic life in the industry that new technological developments and new entries into the market were continuing, but the primary factor which governed the pricing of peripherals for entrance into the peripheral market was the demand elasticity or the substitutability of immediately available products in connection with the needs and applications of users.

F46. Computer equipment is different than used automobiles because when properly maintained such equipment generally performs as well today as it did when new, subject to repair and subject to obsolescence through technological advances. Various end-users view leasing companies as a competitive alternative and in many instances may substitute leasing company equipment for installed IBM equipment. Telex itself leases full systems to users, including IBM CPU’s, and purchases certain peripheral equipment from other manufacturers and remarkets it to end-users. Service bureaus, time-sharing companies and data centers are also used by customers to a degree as alternatives to acquiring new or additional EDP systems. An EDP user may obtain his own equipment, may have his data processing done by establishments such as service bureaus, data centers and time sharing companies, or he may purchase time from another user. Some EDP end-users consider service bureaus, data centers and time purchased from other users as practical alternatives to acquiring new equipment. But with respect to peripheral equipment to be added to or integrated with IBM CPU systems, these alternatives have not provided substantial constraints on IBM’s product and pricing decisions.

*281

F47. Systems manufacturers offer central processing units and peripheral products which are electronically compatible to each other. The peripheral products designed to be compatible with one manufacturer’s central processing unit are not interchangeable or attachable to the central processing units of another manufacturer without modification of their interfaces. As a practical matter, there is no direct or box for box competition between IBM’s peripherals and the peripherals of other systems manufacturers, and in order to replace IBM peripherals with the peripherals of another system manufacturer, the user must first replace his IBM central processing unit. The only box for box peripheral competition of any substantiality has been and is between IBM and the plug compatible manufacturers (PCM’s). IBM’s Systems competitors were not directly affected by IBM’s pricing and product actions for peripherals and made no competitive price responses to IBM’s 2319A and B and Fixed Term Plan (FTP) price reductions for its peripheral products. After FTP, IBM’s Systems competitors were not mentioned in any of IBM’s FTP tracking documents as having cut or reduced their price for any of their products. Time sharing companies, service bureaus, and data centers, were not directly affected by IBM’s price and product actions for peripherals, and after 2319A and B and FTP made little if any competitive pricing responses to IBM's peripheral price reductions.

F48. IBM markets its product by both lease and sale. All sales to end-users or leasing companies are at IBM’s full retail price. IBM determines its retail sale prices by establishing a monthly rental multiple for the product that is equivalent to the number of months of rent that IBM reasonably expects to receive for that product. The rental-sale multiple for each machine may be different — based on the estimate of product life. When IBM is paid the full economic value of a product sold, it expects the product to be used for its full product life. Peripheral products separately leased by other companies or by IBM are not numerous, and leasing companies most generally lease systems or a combination of peripherals and CPU’s. There is a substantial amount of IBM equipment owned by leasing companies. The pricing of leasing companies is constrained or affected by IBM’s pricing policies, which may neutralize to an extent the competitive effect of leasing company activities as to IBM pricing.

F49. The court has not been unmindful of these and other circumstances and arguments pressed upon it by IBM in attempted demonstration that since its predatory acts or market power have not been proved in respect to the EDP industry or the systems market as a whole, it cannot be vulnerable to a charge of monopoly by reason of the interrelationship among components of the industry. Some practical considerations among other more imponderable ones militate against such a theory: (1) The pattern for a divide and conquer strategy of monopoly which its acceptance would permit and foster, and (2) in the realities of the market and of competitive conduct, neither IBM, its competitors nor the public have experienced difficulty in subdividing the EDP industry into markets roughly equivalent to the classification contended for by plaintiffs. IBM recognized as early as 1964 that a' separate and distinct market for input/output peripheral products that were plug compatible to IBM central processing units was developing. For several years IBM studied possible market and product actions which would minimize potential entry of new competitors into that market. Plug compatible manufacturers have been defined in IBM internal documents as “those manufacturers which merely have to plug into IBM hardware to be operable”.

•F50. In late 1969 “peripherals” were designated as a “key corporate strategic issue” — (“KCSI”)—by IBM’s management committee. The key peripherals issue was limited to selected competitive compatible products which replaced IBM

*282

products in an IBM computer system. In IBM’s internal processing and study of this issue “competitive compatible products” were described as “system attached input/output and memory products” including “magnetic tape drives and control units — direct access storage products and control units- — -impact printers and control units.” Memory products were both “main” and “large capacity storage”. Excluded were central processing units, consoles, paper tape products, communication products and control units, information display products and control units, RPQ’s and non-standard products, and non-system attached products. The objective of designating peripherals as a KCSI was to assess the factors affecting both current and future competitive compatible peripheral products; to review IBM strategies, policies and practices so as to identify exposure areas; and to recommend actions to reduce or eliminate such exposures. These plug compatible products were treated by IBM for competitive studies, strategies and other purposes as separate economic entities. And, particularly, IBM recognized for said purposes central processing units, memory products, consoles, -paper tape products and control units, communication controllers and related communication products, magnetic tape products and control units, direct access disk drives and subsystems, and impact printers and control units, as separate economic entities. This separate consideration and treatment no doubt is ascribable in part to convenience of record keeping, comparison of data, effectiveness of evaluation and such factors, but with reference to memory products, magnetic tape products and control units, direct access disk drives and subsystems and impact printers and control units, in view of the competition of other marketers furnishing devices plug compatible to IBM machines, such suppliers, IBM and the industry in general came to regard these lines especially as representing separate economic entities as a result, and for the purpose, of actual competition in the marketplace.

F51. IBM’s 2319A-B product and marketing actions hereinafter discussed affected and were intended to affect directly only one type of product, its 2314 type equipment, and were particularly intended to reduce profits for Telex and Memorex on this type of product. The only IBM products forecasted by it to be protected by IBM’s Fixed Term Plan (FTP) was IBM’s tape, disk, and printer products. The only competitive products forecasted by IBM to be affected by FTP were plug compatible manufacturers’ tape, disk and printer products. When, as here, predatory action is selective and focused, and its anticompetitive effects are similarly shunted away from a more general market, corresponding submarkets should be more readily recognized. IBM has made the persuasive argument that a market concept based on the idea that every manufacturer has a monopoly in each of the components of its product is too sweeping, and necessarily flawed, the flaw being “the disregard of economic forces operating in the markets where manufacturers compete.” With such a generalization there can be large agreement. But the critical flaw in application to the circumstances of this case, it seems to me, would be created by ignoring the separate market and submarkets within which IBM .waged its predatory competitive battles and which became and were thereby made separate competitive entities in the marketplace and within which monopoly power existed and was exercised.

F52. The court finds that the peripheral devices plug compatible with the CPU’s of IBM may be considered the relevant market for the purposes of this case, and that relevant submarkets existed for plug compatible tapes, disks, memories and printers with their respective controllers, and communications controllers.

F53. CPU’s are not reasonably in-cludable within this market and these

*283

submarkets, nor are software as such, but the peripheral equipment plug compatible to IBM CPU’s which are separately leased by leasing companies to end-users are. Alternate sources of computer time such as service bureaus, time-sharing companies, data centers, users selling excess time and the like are not reasonably includable in the relevant market or submarkets with which we are concerned in this case, since their competitive relationships are tangential and indirect, and do not supply a real or substantial competitive force in the relevant markets mentioned. It is true that a large part of the competition in the industry takes place on a systems basis, but the relationship of this competition to the relevant markets with which we are concerned again is tangential and practically indiscernible. Certainly in another context the competition between systems manufacturers would constitute, pr be a part of, a relevant market, but such relevant market is not material under the facts of this case since the competition involved here is not between systems manufacturers but between IBM and plug compatible manufacturers and suppliers.

F54. Nor do the alternate ways of storing data in an EDP system justify the commingling or combination of sub-markets.' The evidence indicates that objectives and applications are the controlling factors in the use of alternatives and not necessarily price. While in special iristances price may have affected particular selections as between alternatives, the applications and objectives of an operation have dictated not only the selection but the price/performance ratio itself in most instances. It is more theoretical than real to say that interchangeability of use or demand as between tapes and disk storage, for instance, precludes the consideration of these submarkets separately. Nor do the trade-offs possible as between the higher performance of memory for the lower price of disks avoid these practical consequences. It is true that if memory were less expensive the user might extend the use of memories for the storage function rather than to utilize disks to the extent he does. This does not obviate the competitive reality that as between disk drives and memories a valid submarket boundary line exists. The reality of this situation appears to be that despite some theoretical interchangeability, a rise in the price of one storage device will cause a substantial number of customers to turn to similar devices less expensive rather than to use fewer of such devices and more of other types of devices. It is true, also, that to some extent certain storage devices, such as memories, are interchangeable with CPU’s themselves, and that users can and do choose between larger or faster CPU’s with relatively small amounts of memory and smaller or slower CPU’s with relatively large amounts of memory. In the realities of the marketplace this, however, has not critically affected the competition between suppliers of memory or disk products compatible with IBM CPU’s, nor does the circumstance of theoretical interchangeability mean that both CPU’s and memory belong in the same relevant market.

F55. The foregoing determinations have been made, it is believed, with due regard for the authorities concerning interchangeability of use, cross-elasticity of demand and supply substi-tutability, and defendant’s arguments based thereon. Mere theoretical cross-elasticity without substantial impact in the marketplace in relationship to demand/price has not been deemed determinative. For every product economic substitutes exist. To be included in the same market it is not sufficient that a few customers would shift from one product if its price, relative to the price of another, were raised. On the other hand, it of course is not necessary that products be identical. “Supply substitut-ability” may not be disregarded. Manufacturers who have existing technological capabilities or tooling to supply reasonably interchangeable products may effectively restrain the power of those in the market to raise prices, but the

*284

evaluation of whether this is so, again, is dependent not upon mere theory blit upon the reality if any of the effect of the potential in the marketplace. While the potential need not necessarily be an immediate one, it must not be so remote as to have no actual influence on the competitive situation. A relevant market cannot be enlarged by theoretical speculation as to future market conditions or potential substitutability having no substantial effect upon competition during a period in question. I find that neither cross-elasticity or interchangeability of use or demand, nor substituta-bility of supply, critically militates against the relevant market and submar-ket definitions within which IBM’s market power will now be assessed.

V

MARKET POWER

F56. Monopoly power is the economic ability to charge unreasonably high prices and to exclude competition. Proof of the actual use of such power for these purposes is not essential to a finding of its existence but would be an important factor in any assessment of market power. The strength of competitors is relevant to an assessment of market power. Monopoly power presupposes the power to control what happens in a relevant market. Ease of entry may be an indication of lack of' market power on the part of an alleged monopolist. Difficulty in entering, weakness of competing companies and dependence of competitors upon dominant forces in the market are among indicia of market control on the part of an alleged monopolist. Necessity of competitors to react to price changes by the alleged monopolist, particularly above or below a scale based upon self-determined reasonable cost and profit may be important. If the percentage of a relevant market controlled by an alleged monopolist is high an inference of market power may be drawn. Where its control is moderate no inference of market control may be permissible. In case of a medium range, it may be impossible to infer or to rule out monopoly, so that factors other than market percentage must be looked to primarily. Where there is direct credible evidence of market domination or predatory practices which are productive of control in a particular relevant market, inferences need not be depended upon but this more direct evidence may be determinative. Other factors to be considered are any necessity on the part of an alleged monopolist to meet competition in technology and pricing, the equality of performance in the industry and its comparative youth, growth and dynamics or change. Claimed necessity of responding to competitive influences beyond the control of the alleged monopolist may be only its excuse for anticom-petitive conduct for the purpose of maintaining or extending monopoly power or to surmount threatened competition, and monopoly is possible in a young, dynamic and complex industry as well as in an old or static one, and may be even more feasible in special cases through masking of selective market strategies in the overall technological developments. Sophistication of users or competitors may discourage monopoly but equal or greater sophistication on the part of an alleged monopolist may be a counter-balancing factor, and industry dynamics may continue in evidence through technological momentum beyond the inception of monopoly. While these and other criteria and their limitations have been considered, it is recognized that the question of monopoly control is one of fact to be determined on the whole record and not susceptible of being resolved by any mechanical applications.

F57. By their arguments and proposed findings, plaintiffs would have the court find that in the period 1969 to 1972 IBM possessed monopoly market power in a general systems relevant market as well as in the more limited relevant plug compatible market and the submarkets for magnetic tape products, direct access storage products, memory products, impact printer products, and

*285

communication controllers that are plug compatible with an IBM central processing unit. Little or no evidence was introduced in these cases that IBM evidenced an intent to monopolize, or directed efforts toward monopolization of the EDP systems market in general, except through its more focused conduct. Presumably plaintiffs now do not wish to rest their case entirely upon their plug compatible market theory, or at least consider that the dominant position of IBM in the general market supports, or lends substance, to its claims that IBM monopolized or attempted to monopolize the plug compatible market or submarkets. On the other hand, the position of IBM in the general market lends increased force to its arguments concerning its declining market share, the dynamics of the industry, the strength of competitors, and other factors tending to negate a monopoly position. While it is believed that the position of IBM in the general EDP industry, particularly with reference to general systems, is relevant, the evidence is insufficient for a finding, and it is unnecessary in the court’s view to find in order properly to resolve this case, that IBM during the relevant period monopolized or attempted to monopolize the general systems market.

F58. There is no question but that IBM occupied, and continues to occupy, an important position in the systems market. In 1970 revenues reported from electronic data processing products and services according to the census, IBM was the leading company in the industry, with almost $3.5 billion of revenue. AT&T which is not a systems manufacturer was next largest in terms of EDP revenue, $759,435,000. Of the next three largest companies, Univac (Sperry Rand), Honeywell and Control Data, none had EDP revenues in excess of $460 million. Internal IBM documents containing measurements of IBM’s share of the domestic market for systems and peripherals place IBM’s market share progressively decreasing from 75.9% in December, 1964, to 73.-3% in September, 1968, and IBM’s market share of central processing units (CPU’s) progressively decreasing from 68.6% in 1964, to 64.4% in 1968.

F59. Defining the market broadly, as IBM claims it should be, competitors include many large diversified companies with important skills and substantial financial resources, and many competitors are strong, independent and growing. Entry into such a broad market has not proved difficult for many companies. Between 1952 and 1970 the number of competitors in the EDP industry multiplied more than 136 times, from 13 to 1773, according to the census. The total United States EDP revenue has increased about 212 times from $48 million in 1952 to $10.2 billion in 1970. Many companies, in addition to IBM, have shown spectacular growth, although none to the extent IBM has. The number of companies which manufactured systems increased from 3 in 1952 to 96 in 1972. IBM’s technology, and its organization and diligence in advancing it, have been of high quality, contributing in a substantial degree to IBM’s general success in the industry. There is little or no indication in the evidence introduced in this case that IBM adopted specific programs to throttle or impede general systems competition or that it sought to implement any predatory intent with respect to the EDP industry as a whole, as distinguished from efforts directed specifically against the marketers of peripheral equipment plug compatible to its CPU’s.

F60. Generally speaking, EDP customers have been furnished with progressively better products at progressively lower prices. Memory capacity has increased by a factor of approximately 700 from the Univac I to the IBM 370/168. Performance of the CPU as measured in the execution of additions, multiplications and divisions per second had increased by a factor of over 4300, 3100, and 2000, respectively, from the Univac I to the IBM 370/168. The fourth generation 370/168 costs only a little less than seven times as much as

*286

the first generation Univac I. The performance of tape drives as measured by the transfer of characters per second has increased by a factor of 46 from the tape drive used with Univac I to the tape drive used with the 370/168 for an approximately 1% price increase. There has been a 16 fold increase in storage capacity as well. From the first to the fourth generation, there has been a 13 times performance improvement in printers for less than 3 times the price increase. There has been a 37 times performance improvement and a 20% drop in price in disk files from the first to the fourth generation. The requirements of electronic data processing users, and the profusion of companies attempting to fill those needs, have led to a marked increase in the performance of products and significant decrease in the cost per unit of computing. Broadly defined the EDP industry appears competitive and dynamic.

F61. IBM’s market share of the EDP industry as a whole or the general systems (CPU) market does not of itself justify an inference of monopoly power in the market as so broadly defined, or at least as to this plaintiffs have not discharged their burden of proof to show monopoly power as a part of their monopoly complaint: According to the census, IBM’s 1970 share of reported EDP revenue for hardware and leasing companies over $5 million, 42.3%; its 1970 share of reported EDP revenue for hardware companies over $5 million, 44.-9%; its 1970 share of reported EDP product revenue, 44.6%. IBM’s share of the value of 1971 shipments of “electronic computers and peripheral equipment, except parts”, according to the U. S. Bureau of the Census, was 36.7%. IBM’s share of the value of 1971 shipments of “Electronic Computers, Digital, General Purpose”, according to the U.S. Bureau of the Census, was 40.9%. IBM’s share of the value of 1971 shipments of “direct access storage units such as magnetic tapes and drum, magnetic cord and bulk core memory”, according to the U.S. Bureau of the Census, was 30.4%. Its share of the value of 1971 shipments of “serial access auxiliary storage units such as magnetic tape units”, according to the U.S. Bureau of the Census, was 46.6%; its share of the value of 1971 shipments of printers, according to the U.S. Bureau of the Census, was 38.3%. These shares have been declining. IBM’s share of reported EDP revenue has declined from 64.1% in 1952, to 35.1% in 1970, and there have been comparable declines in others of the categories above-mentioned.

F62. The figures on market share particularly with reference to plug compatible peripherals are not readily available from published sources, nor can they be extrapolated or inferred from census data dealing with peripheral products in general, some of which have been cited above. But the defendant, in the processing of its marketing strategy and planning, developed an organization and system well designed to segregate these data, since its studies were directed specifically to the narrower markets. Accordingly, while the court has considered the general data available, and inferences reasonably to be drawn therefrom, it has seemed fair and appropriate to afford considerable weight to data available from defendant’s studies, and it has.

F63. The inception of a related market was natural because of the dominance IBM products commanded in the marketplace and the feasibility of furnishing functional equivalents to some of these products which could be rendered plug compatible1 with IBM CPU’s. The devices marketed by Telex and others plugged into and replaced parts of the System ■ 360 family, and ultimately parts of the 370 family were replaced. Notwithstanding some difficulties of rendering interfaces compatible, and developing the personnel and technology to design and manufacture or otherwise secure equivalent devices, entry was initially easy for peripheral equipment manufacturers because they could choose to copy only proven successful products.

*287

Moreover, they could utilize in many instances systems hardware provided by the system manufacturer and typically would sell only after all systems engineering, systems marketing, side preparation and systems installation work had been completed. The number of companies supplying peripheral equipment plug compatible with IBM systems grew from 2 or 3 in 1966 to some 100 today. A number of these handled only one type of device and few if any had the variety of peripheral devices plug compatible to IBM CPU’s that Telex had. The relatively' large number of companies in time engaged in the plug compatible business did not represent a corresponding dispersal of the business, since the major share during the developing years was concentrated within a matter of a dozen plug compatible manufacturers (PCM),

F64. Of course, with respect to the IBM peripherals later replaced by plug compatible devices of other manufacturers, IBM initially had 100% of the market. But as the plug compatible business developed on the part of other manufacturers or suppliers, IBM’s market share was substantially eroded, and in due course it became a concerned competitor for peripheral devices to be attached to its own systems. Starting in 1968 there was a very rapid growth in the quantity of equipment shipped by peripheral equipment manufacturers which was plug compatible with IBM CPU’s. This occurred first with tape drives in 1968 and then with disk drives in 1969 and this plug compatible growth continued into and perhaps through 1970, and became in itself an important and recognized market which increasingly was enlisting new participants and inviting plans for further expansion.

F65. The increase in plug compatible business and the decrease in IBM’s share of the plug compatible market must be evaluated in light of the foregoing circumstances and the fact that it was not until 1970 that IBM’s strategic and tactical responses to the inroads of the plug compatible manufacturers became really effective. The apparent vitality of the plug compatible market and the increase in the number of companies engaging therein accordingly cannot be considered as necessarily negating the monopoly power or predatory intent of IBM, since as far as the evidence discloses IBM did not really begin the exercise of whatever market power it had as against plug compatible equipment manufacturers until the period 1969-1970. It also is relevant to note that any increase in competitive entries into the market in 1972 and 1973 could well have been affected by the institution of the present litigation in early 1972, and, for prospective entrants, at least some possibility which pending litigation to obtain remedial action may have held out.

F66. The testimony of Bonham and the Bonham charts indicate in accordance with the census that in 1970 IBM received revenues of $1,137,819,000, from its plug compatible peripheral products, and all other manufacturers of IMB plug compatible products together received slightly in excess of $100 million in revenues; that in 1970 IBM had 90% of the revenues from tapes attached to IBM CPU’s and PCM’s had 10% of the revenues; that in 1970 IBM had 68% of revenues received from disk drives attached to IBM CPU’s and PCM’s had 32%; that in 1970 IBM received 99.6% of all revenues for memory products attached to IBM CPU’s and that PCM’s had the remaining .4% of memory revenues; that in 1970 IBM received 92.3% of revenues from communication controllers attached to IBM CPU’s and that the PCM’s had 7.7% of such revenues.

F67. IBM’s internal documents indicated that in 1970 IBM had 80% of tapes (units) attached to IBM CPU’s, and that PCM’s had 13%; that in 1970 IBM had 94% of disk drives (units) attached to IBM CPU’s, and that PCM’s had 6% ; that in 1970-71 IBM had 99% of the market for impact printers attached to IBM CPU’s; that Telex was an entrant in the high speed impact plug compatible printer market with a potential market share of 8.5% for PCM’s,

*288

and 91.5% for IBM by 1978; that Telex and a number of other companies were potential entrants into the plug compatible memory market, and that the total number of units of magnetic tape devices (240X-2420-3420 type) and disk drive products (2311-2314/2319’s-3330 type) attached to IBM central processing units as of December, 1970, June, 1971, and December, 1971, June, 1972, and December, 1972 (both the PCM’s and IBM’s share including all devices marketed by them whether leased or sold) were as follows:

The following four charts accurately reflect the information set forth. In the following four charts the figures following “user owned” and “leasing company” refer to devices initially manufactured by IBM. In the preceding chart the devices listed under “IBM” include IBM manufactured devices whether owned by IBM, owned by end-users or owned by leasing companies:

*289

*290

F68. The Bonham charts on which some of plaintiffs’ contentions with reference to market share are based cannot be taken at face value and must be evaluated with reference to related circumstances as shown by the evidence. For example, only revenues from the principal or dominant PCM’s in the market are included in some of the charts; revenues from leasing companies which acquired IBM plug compatible memory devices and disk devices from IBM and marketed them in competition with IBM were excluded and the figures were not updated to indicate a probable continuing decline of certain IBM market shares during the period 1971-72. Weighing these and other factors, and particularly testing the charts as against the internal documents of ■ IBM with reference to its shares of the relevant submarkets, it is fair to say that they support a finding that IBM’s share of the relevant submarkets or the combined submarkets comprehended in the general market classification “peripheral equipment plug compatible to IBM”, is such as to permit an inference of monopoly power on the part of IBM, and the court so finds.

F69. IBM’s internal documents generated in connection with competitive studies looking toward management decisions to meet the competition of plug compatible manufacturers have been deprecatingly referred to in argument by the defendant as being the products of non-management employees, or as grossly underestimating competition. It seems appropriate to note here, however, and for later reference in connection with the predatory conduct of IBM, that most of the studies were made by highly trained and qualified IBM personnel acting within an organization justly noted for its perception and responsiveness to market conditions, and with technological standards and aids likely superior to most great companies of the United States. There were some uncertainties in the precise sources of some data coming to the attention of management, and some combinations of documents in the evidence did not represent the precise form in which they were submitted to management. Yet the inputs into management with reference to competitive and market situations, the testing and processing by management of these im-puts and its rather consistent acceptance of their bases in any management outputs, indicate to me, by and large, in view of all the circumstantial evidence in the case that IBM’s internal documents represent significant evidence not only as to market shares but as to the intent and purpose of the defendant. They were prepared for submission to

*291

management in the course of the business of a highly competent and effective organization at a time when we can assume that their litigious significance was not distorting. They were represented by forecasts, chart presentations, tables, and similar documents and were generated with specific reference to the competitive situation involving the plug compatible market at the time, and their relationship to top management decision under all the circumstances may be readily perceived in most critical instances.

F70. On the basis of all of this testimony with reference to market shares, inferences that may be reasonably drawable therefrom, direct testimony as to IBM’s domination of the relevant markets and submarkets, and the effect upon competitors in this market and submarkets of its predatory competitive practices, during the period 1969 to 1972, inclusive, I find IBM possessed monopoly market power in the relevant market of peripheral equipment plug compatible to IBM CPU’s and in the relevant submarkets for magnetic tape products, direct access storage products, memory products, impact printer products and communication controllers plug compatible with IBM central processing units. The court further finds in this connection that this general market and the submarkets specified above comport to the competitive realities of the period ; that by reason of the peculiar development of the EDP industry with the historical domination of IBM in the general systems market, with the market of plug compatible peripherals growing up, in historical and developmental fact, as a separate competitive entity along with its submarkets, the separate reality of this market along with its submarkets cannot be dismissed or ignored as a mere example of a manufacturer necessarily having a monopoly of its own product. The court further finds that whether the submarkets hereinabove defined be considered separately and severally or whether they be combined into the market for peripheral EDP equipment plug compatible with IBM CPU’s, monopoly power during all periods material herein was possessed and exercised by the defendant IBM.

VI

IBM’S PRACTICES AND INTENT

F71. IBM’s top management became concerned in the summer of 1969 that IBM forecasts with regard to plug compatible competition were understated. This concern was intensified in January of 1970 when IBM learned that the Bureau of the Budget intended to encourage federal agencies to use equivalent lower cost peripheral equipment compatible with CPU’s supplied by IBM and by other systems manufacturers and suggested the utilization of standard interfaces.

F72. By 1970, some plug compatible devices offered in the tape and disk areas by PCM’s were, in fact, functionally superior to, and were regarded by IBM as superior to, their corresponding IBM products. The Telex 4700, 4800 and 6420 model' tape drives were, in fact, superior and were regarded by IBM as superior products to the IBM equivalent tape drives Models 729, 2401 and 2420. Telex’s 5311 and 5314 disk drives were in fact superior, and were regarded by IBM as superior to IBM’s equivalent 2311 and 2314 disk drives. Product superiority was achieved by Telex because these products were first delivered substantially later than the IBM products. The Telex 5314 and the Telex 5328 which is the controller for the Telex 5314 were purchased from Itel and were developed for Itel by a group of 12 former IBM personnel. The following table shows when IBM first an-

*292

nouneed and delivered the products and when Telex first announced and delivered the products it intended to replace the IBM product:

*293

F73. In response to the increasing competition IBM was receiving from plug compatible manufacturers marketing some functionally superior peripheral devices in the tape and disk areas, IBM’s Management Committee, in February of 1970, designated peripherals as a “Key Corporate Strategic Issue” (KCSI). To be so designated was a management device to deal with an issue that had broad implications and required the attention of its top corporate management.

F74. After the designation of peripherals as KCSI a task force was formed in March of 1970 to be headed by H. E. Cooley, Vice President of the Systems Development Division. The Peripheral Task Force or Cooley Task Force, as it became known met regularly both in formal and informal meetings from the middle of March of 1970 until its report to the Management Committee of IBM on July 31, 1970. The objective of this task force was to examine the competitive threat to IBM of plug compatible suppliers. A Telex trial witness, Richard Whitcomb, who was IBM’s manager of I/O Systems Marketing from the fall of 1968 to the summer of 1971, participated in the work of the Peripheral or Cooley Task Force on behalf of the Data Processing Division. A purpose of the Peripheral Task Force was to study and recommend plans and product strategies to impede the growth of IBM's plug compatible competition. The Peripheral Task Force made in-depth analyses of various plans and strategies each having as a significant purpose the containment and retardation of the growth of IBM plug compatible competitors. The task force made in-depth assessments of the status of plug compatible competition and analyzed the viability of particular plug compatible competitors, including Telex.

F75. In the summer of 1970, IBM’s top management believed that plug compatible competition was one of IBM’s major business problems. And the “Mallard” project was the first concrete response to this problem. The Mallard disk file was announced by IBM as the 2319A disk storage facility for the Model 145 System 370 on September 23, 1970. The 2319A was a reworked 2313 disk drive with one of the spindles removed. IBM removed one of the four disk drives from the 2313 box (IBM’s four spindle 2314 type disk drive box) and put in some of the control function electronics from the 2314 controller. The control function on the 2319A for the 145 was handled by integrated file adapter (IFA) that was placed underneath the covers of IBM’s System 370/145, together with the portion moved into the disk file cabinet. The user had the option of using the 2314 subsystem attached directly to the channel instead of the internal IFA. The software programming support for the operating system was identical in each configuration.

F76. The 145 end-user who elected to use a 2314 controller, a 2312 and a 2318 for a three spindle configuration, instead of the IFA and 2319A, was charged the higher 2314 prices for the same identical function. In short, if the user did not choose the IFA and 2319 he received no price reduction for the disk drives utilized on the 145. The end-user who selected the IFA/2319 disk drive subsystem for the 145 saved $1,325 per month on a three spindle configuration. The $333 per spindle price represented a $103 per month reduction per spindle below IBM’s price per spindle for its 2313 four drive configuration. IBM’s monthly rental per spindle for the 2319A was $100 per month lower than Telex’s then current price per spindle for Telex’s equivalent drive; and, was less than the price then being offered by IBM’s plug compatible competitors. The monthly rental adopted by IBM for the 2319A was the lowest rental profile considered by its management prior to the announcement. The price level announced was forecast by IBM to have maximum impact on IBM’s plug compatible competitors. Prior to the 145 announcement, IBM considered raising the rental price of the 145 CPU to offset the reve

*294

nue reduction that would result from the reduction associated with the IFA/2319 when compared to the 2314 control unit with the 2312/2318 attached.

F77. The 2319A disk subsystem did not substantially increase the performance of the 2314 subsystem. The end-user customer received similar functional performance by utilizing a three spindle 2319 box with the IFA as he did by utilizing a 2314 subsystem. The data rate, the access rate, and the data capacity per spindle were the same. IBM’s price cuts for the 2319A and IFA were not justified upon the basis of reduced manufacturing costs.

F78. IBM may have reduced its cost somewhat through reuse of 2314’s which were being returned to IBM because of plug compatible competition, but it is clear from the evidence that any decreased cost was of minor importance or influence in the Mallard plan and that price reduction independent of cost on limited products in competition with plug compatible suppliers was the primary purpose of the response. IBM camouflaged the 2319A price cut as a “new” product for the purpose of avoiding a general price reduction to all its installed 2314 subsystems which would have reduced IBM’s revenue stream of $514 million a year on its installed disk base by approximately $120 million per year.

F79. The 2319A price cut was designed by IBM specifically to contain plug compatible competition. It originated in the Cooley or Peripheral Task Force and was approved by top management. Its primary purpose was to maintain control of the plug compatible disk market for IBM. It was introduced by IBM with the specific purpose and intent of suppressing plug compatible disk competition. IBM admits, indeed argues, that its action was a competitive response necessitated by the inroads of plug compatible competition and that it in fact did not succeed in maintaining IBM’s market share. But IBM already possessed a dominant market share, and continues to do so. Notwithstanding lawful acquisition theretofore, its intent to maintain its monopoly by unlawful predatory conduct cannot be equated reasonably with an ordinary competitive response.

F80. IBM, in October, 1970, organized a second peripheral task force to analyze plug compatible competitors in the disk drive area. The scope of the task force study included analyzing of the marketing, management, maintenance, production and engineering capability of IBM’s plug compatible competitors. The. group was directed to study and estimate the announcement and first customer shipment dates on PCM’s 3330 equivalents and make a cash flow analysis, including financing arrangements, of PCM’s, to make an estimate of the PCM’s 2314 manufacturing cost and to determine “how long can OEM PC suppliers go on 2314 prices?” This group’s report concerning Telex concluded that Telex was viable, that its management was competent and aggressive and that it had a strategy of marketing a full line of high volume IBM plug compatible peripherals, that its in-house engineering capability was good, but that its manufacturing costs were 10% to 15% above IBM's. The Telex analysis concluded that Telex’s cash flow was inadequate to permit Telex to finance its own lease base and that Telex’s key exposure was “impact by IBM — shortens product life”.

F81. Prior to a further announcement involving the 2319 IBM had concluded that Memorex and Telex were the two most significant plug compatible competitors for 2314 type disk drives. Telex witness Whitcomb in his role as IBM’s I/O Marketing Manager attended a presentation by Telex’s deposition witness, Fassig, on Memorex and Telex in the fall of 1970. Fassig’s presentation was an analysis of the impact on Memo-rex and Telex of various price cuts by-IBM on 2314 drives, the corresponding price reactions that Memorex and Telex would be forced to make, and the effect upon their viability. Fassig’s analysis demonstrated that as IBM would cut the

*295

price in the 2314 area, and Memorex and Telex would respond, there would be a very serious impact on the profits and revenues of both Memorex and Telex. By October 20, 1970, IBM’s Management Review Committee was considering extending the 2319 program to the System 360 and the elimination of IBM’s extra use charge on disk drives. On December 10, 1970, the Management Review Committee approved the “2319B” announcement.

F82. IBM announced the 2319B on December 14, 1970. The 2319B was a single box containing three 2314 disk drives. The 2319 did not attach to an IFA but attached to IBM’s 2314 control unit. In conjunction with the 2314 control unit the 2319 could be utilized on all IBM Systems 360 and 370 computers to form a disk subsystem of three, six or nine 2314 drives (8 drives plus a spare). The 2319A announcement only permitted the use of one 2319A box with the IFA. The 2319B announcement permitted the use of an additional 2319B box with the 2319A box and the IFA, thus giving the 145 IFA user the option of using the IFA and two 2319 boxes with an additional 2314 box to make up eight drives, or using the 2314/2319B nine drive subsystem.

F83. IBM’s rental price on the 2314 control unit utilized with the 2319B remained at $1,480 per month. IBM’s rental price on the 2319B was set as $1,000 per month, or $333 per 2314 drive • — the same price that had been set by IBM on the 2319A announcement. The monthly rental price on the 2319B represented a substantial price cut for 2314 drives. In a 2319 subsystem consisting of three disk storage units, the 2319 monthly rental price represented a price cut of over $1,000 per month for a 2314 subsystem. In addition, the 2319B announcement eliminated IBM’s extra use charge on 2319A, 2319B and 3330 disk drives. The elimination of the extra use charge represented an additional and substantial price reduction. The 2319B announcement was purely a price cut. The 2319 did not even purport to represent any increase in performance in a 2314 subsystem. This price cut was to a point below the prices IBM’s plug compatible competitors were charging for their plug compatible equivalent 2314 drives. IBM’s price cut on the 2319B announcement cut the price of the nine 2314 drives used in a 2314/2319 subsystem approximately $700 below the average price of IBM’s plug compatible competition and $800 below Telex’s price.

F84. Just after IBM made its 2319B announcement IBM had an installed base of 47,051 lease and purchase 2314 spindles, as compared to an installed base of 2,639 2314 equivalent type spindles for all of its plug compatible competition together, and further, IBM had over 94% of all disk drives installed with IBM CPU’s. IBM’s forecasts and analyses with respect to the adoption of the 2319B program considered IBM’s competition with plug compatible manufacturers. None of the IBM forecasts and analyses with regard to the adoption of the 2319B program was expressed or geared in any way toward competition between IBM and systems manufacturers, leasing companies, software and consulting organizations or service organizations. The 2319B was designed by IBM as a predatory action contrived to maintain its 94% control of the plug compatible disk market.

F85. IBM’s 2319B price cut substantially impacted its plug compatible competitors’ revenues and profits by reason of responsive price reductions which IBM had anticipated would have to result, but it did not succeed in fully protecting IBM’s installed base. As Mr. Evans, then President of IBM’s Systems Development Division, testified: “For a month or two after the 2319 was announced, the plug compatible sales leveled off; but then, as I recall, there were pricing actions by the competition, and that curve turned right back upward.” Not only did IBM plug compatible competitors lower their prices but the plug compatible competitors had complete modularity (1 drive per box)

*296

on their 2314 equivalent drives and by reducing their price per drive they were able to sell between the configurations of IBM’s three, six and nine 2319 drive subsystem configuration. (1, 2, 4, 5, 7 and 8 drive configurations.) Mr. Fin-nell reported to the Management Review Committee in January, 1971, with regard to IBM’s 2319B and 3420 pricing actions: “OEM reaction to our recent tape and disk pricing action were as expected or lower. We are continuing to update our 1971 forecasts— raises the question of are you really ahead or are you back to where you started before you adjusted your own prices.”

F86. In December, 1970, IBM announced a price reduction for all its disk devices, including the 2314, 2319 and 3330, by eliminating all additional use charges. Immediately after IBM’s announcement of the 2319B, Telex negotiated the 28% price reduction from its supplier ISS for the 2314-type devices it was buying. Telex, other peripheral equipment manufacturers, and some leasing companies dropped their prices for the 2314-type devices to levels substantially under those of IBM. After the decrease in Telex disk device prices the order rate of Telex disk devices again increased significantly. In the period from November, 1970, to December 31, 1972, Telex shipped 1074 more 2314-type disk drives and 191 more 2314-type disk controllers than it had forecasted in November, 1970, that it would ship.

F87. Telex claims that as a follow-up predatory action IBM announced its 3420 or “Aspen” tape device; that Aspen was not planned by IBM until after PCM’s started gaining substantial shares of the installed base from magnetic tape products, and that Aspen was a price manipulation which was conceived to reduce the number of drives marketed by plug compatible manufacturers without reducing IBM's revenues from its installed 2420 tape drives. While there is some evidence to support such a theory, the court considers it less than preponderant and finds that the announcement and marketing of Aspen was not predatory and did not represent in and of itself an unlawful attempt to monopolize. The Aspen development began in approximately 1966, and finally incorporated significant technological innovations not found in prior tape devices. The price of the IBM Aspen was based on lower cost when compared to the 2420 and included reasonable profit. Telex announced its equivalent 2420 products almost two years after the announcement of the 2420 Model 5 and has offered insufficient proof as to how in any event it was damaged by the 2420 Model 5. There was no Aspen issue included in the final pre-trial order, but the court has made the foregoing findings against the possibility that it may be contended that such an issue with the acquiescence of the parties was actually tried and considered during the trial.

F88. Telex contends that IBM’s announcement of its Fixed Term (leasing) Plan (FTP) in May of 1971 was a predatory act and that similarly IBM’s announcement of its Extended Term Plan (ETP) in March of 1972 was a predatory act. Telex also claims that FTP and ETP constituted illegal restraints of trade because they locked out competitors; and a further contention is made that IBM’s announcement of a CPU price increase on July 28, 1971, was undertaken specifically for the purpose of recouping the losses occasioned by the introduction of FTP. IBM denies these contentions and alleges that it announced its leasing plans in response to similar plans offered by virtually all IBM’s competitors, recognizing that without some form of long term lease it would suffer serious and continuing loss of business to systems manufacturers, leasing companies and peripheral equipment manufacturers, pointing out that it now has three long term lease plans under which it offers most of its EDP equipment. Again, there seems little question but that in a different context, or directed to general competition, the

*297

leasing plans adopted by IBM might be unexceptional or entirely justified. The question remains, however, whether in the setting of IBM’s dominant position in the plug compatible submarkets and in view of the evidence as to its specifically directed intent and concern with reference to the plug compatible competition in those markets, the two leasing plans above-mentioned can be sustained as against Telex’s attack. In view of the fact that most of IBM’s systems manufacturer, leasing company and peripheral equipment manufacturer competitors were offering long term leases by the Spring of 1971 (Finding F100), IBM expected to, and was likely to, continue to lose substantial systems and peripheral business unless some plan was adopted.

F89. IBM’s 2819B announcement failed to retain IBM’s high share of the plug compatible disk market and failed to contain the growth of IBM’s plug compatible competition during the first quarter of 1971. The latter continued to make strong advances with its installations in the 2314 disk area. By February 12, IBM’s plug compatible competitors had installed 3,006 2314 equivalent spindles; by March 15, 3,491; and by April 9, 4,614. IBM’s plug to plug peripheral competition commenced to proliferate from tapes and disks to printers in the last quarter of 1970. By September, 1970, IBM anticipated that Telex would have a plug to plug compatible printer, and revised its printer forecast. Telex was regarded by IBM as the leading competitor in the plug compatible peripheral marketing area because of its broader product lines, having tapes, disks and printer, and IBM suspected that Telex would soon offer a memory device.

F89a. In 1970 and 1971 IBM experienced the effects of a nationwide recession combined with inflation, which caused a substantial increase in the level of returns and discontinuances, of its EDP equipment including peripheral equipment. IBM at that time offered equipment only on short-term leases or for sale; its rental customers could effectively return their equipment to IBM on 30 days’ notice. As a result of the economy, many of IBM’s rental customers took advantage of this privilege and returned a significant amount of equipment to IBM. IBM’s experience was not shared by its leasing company, systems manufacturer or peripheral equipment manufacturer competitors, since their equipment was generally leased for terms of one, two or more years, with termination charges or other costs in the event of cancellation. Another factor affecting IBM’s business in this period was the increasingly lower rental prices charged by leasing companies and peripheral equipment manufacturers for equipment similar to IBM’s. As a consequence of these factors, IBM’s sales force in 1970 achieved only 50% of its selling objective. In 1971, IBM experienced the worst sales record year in its history for EDP equipment.

F90. Mr. Whitcomb prepared an overview study of IBM’s plug compatible competition late in the first quarter of 1971. This overview study was presented to the President of IBM, IBM’s Data Processing Division, and IBM's Management Review Committee. It concluded (a) the plug compatible phenomenon was accelerating in volume and scope; (b) the plug compatible competition presented a serious threat to IBM’s potential growth since the exposed peripherals represented 63% of IBM’s installed lease base; (c) defending against plug compatible competition was difficult because of their pricing and performance advantages; (d) IBM should try to combat peripheral competition by frequent advances in technology utilizing “midlife kickers” and pricing which would take advantage of IBM’s short-lived product lead in the peripheral area including the consideration of long term uses; 13% of IBM’s systems were “contaminated” with plug compatible tape or disk equipment and PCM penetration would increase with package selling. Specifically, the Whitcomb study found that IBM would lose 19% of the plug

*298

compatible tape market by 1976 and IBM’s planned tape program was inadequate and IBM would lose 28.7% of the plug compatible disk market by 1976 and IBM’s planned disk program was inadequate. The Whitcomb study also found that IBM should be concerned about the memory and printer areas. While it is clear that IBM expected plug compatible competition to increase in volume and scope by the end of the first quarter 1971, it is also clear that IBM even by its worst case forecast only anticipated loss in the neighborhood of 20 to 26% of tape and disk markets to all of its compatible competition by 1976.

F91. Even after the 2319 price cuts, IBM on in-depth study considered Telex a “viable” competitor that could “manage impressive earnings. . . .’’In April, 1971, IBM’s Management Review Committee concluded that its control of plug compatible disk and tape drives was being eroded and the printers and memories would be next. IBM determined to deal more effectively with its plug compatible competition. At a Management Review Committee meeting on April 23, 1971, IBM’s chief executive officer, Mr. T. J. Watson, Jr., formulated IBM’s basic policy approach. Mr. Watson informed the Data Processing Group that he wanted “a clear understanding that the company swallow whatever financial pills required now and get ready for the future . . . irrespective of financial considerations of one of two years — must return this business to a growth posture and operate accordingly.” Mr. Watson stressed the need for IBM “to make the hard decisions today so that the same problems don’t have to be faced again and again down the road.” IBM’s Management Review Committee appointed a task force to develop a new peripheral strategy and specific action programs to deal with plug compatible competition. The task force, although not specifically given that name, was sometimes known within IBM’s organization as the “Blue Ribbon Task Force”.

F92. On May 6, 1971, this task force made a report to IBM’s Management Review Committee (MRC). It recommended drastic tape and disk price cuts to contain IBM’s plug compatible competition. Specifically, the task force recommended that IBM reduce its price by 50% on 2314 and 2420 disk and tape drives, by 20% on 3330 disk drives and by 15% on 3420 tape drives. The MRC rejected these recommendations and directed the task force to develop a strategy for memories and printers and to rework “the possibility of a long-term leasing approach as suggested by FTC (Frank T. Cary).” In the ensuing three weeks this task force made a number of iterative reports to IBM’s Management Review Committee on long-term leasing of specific peripheral products. The MRC gave final approval to IBM’s Fixed Term Plan (FTP) on May 25, 1971.

F93. The task force’s presentations and recommendations to IBM’s MRC during May of 1971, indicated that some disks, tape and printers were going to be included and some omitted from FTP, and that there would be an omission of card I/O devices and system 3 products. Most of its iterative forecasts were in terms of the impact that IBM action would have on IBM’s plug compatible competition. The inclusion of the 1403 Nl and 3411 printers was recommended because of plug compatible competition expected from Telex. The task force determined that the Fixed Term Plan leasing would cost IBM millions of dollars in revenues and profits during the first two years, and projecting that on disk drives IBM would lose $13,200,000 in 1971, and $20,300,000 in 1972; on tape drives $6,500,000 in 1971 and $5,300,000 in 1972; on printers $11,800,000 in 1971, and $18,500,000 in 1972. In short, the task force determined that IBM would sustain revenues reduction of more than $75 million in 1971 and 1972 by putting tape disks and printers under FTP leases. Notwithstanding these projected losses, it was thought that FTP would be very profitable to IBM in the long run because losses from plug com

*299

patible competition would be decreased and it would have more units out in the field for longer periods of time.

F94. On May 27, 1971, IBM announced FTP. One and two year leases on IBM disk, tape and printer peripheral products (except those excluded) were provided for, with an 8% monthly rental discount for one year leases and a 16% monthly rental discount for two year leases. IBM also eliminated its extra use charges on products leased under such leases. Punitive penalties for cancellation of a lease by a customer were included. The penalty for a two year lease terminated during the first twelve months was five times the monthly rental charge. The penalty for a one year lease cancellation, or a cancellation of a two year lease during the second year, was two and one-half times the monthly rental charge. IBM also reduced the purchase price by 15% on the products covered.

F95. IBM’s price cuts under the Fixed Term Plan were even greater than the apparent 8% to 16%. The elimination of IBM’s extra use reduced IBM’s monthly rental charges on disk products covered by the two year Plan by 31%, and its tape products by 20%. On printers the reduction was about 30-35%. The price cuts in some instances put IBM prices below those of its plug compatible competitors.

F95a. Between 1968 and 1972, Telex had a number of products which competed with IBM products. There were many price changes and price variations during this period. All of those Telex products were at all points in time listed at lower prices than comparable IBM products except on four isolated occasions. Telex’s prices were generally higher than the prices of other plug compatible manufacturers. In addition, Telex and the other plug compatible manufacturers generally reduced below list the prices they actually charged through various forms of price concessions.

F96. The benefits anticipated by IBM in connection with the adoption of FTP revolved around the suppression of IBM’s plug compatible competition. Indeed, the very creation of the task force was occasioned by plug compatible competition. Pricing presentations of the task force to the Management Review Committee were importantly concerned with comparisons of plug compatible prices and projections. Defendant’s officers at the trial expressed the view that FTP was simply to render the company “more competitive” and to obtain more business by meeting the competitive efforts on a basis similar to that of plug compatible suppliers. It is the court’s view that such justification, which could be convincing under different circumstances, is overpowered by IBM’s monopoly position in the particular markets involved and the rather clear indication that its action was directed not at competition in an appropriate co'mpetitive sense but at competitors and their viability as such. The products specified by FTP were those peripheral products on which IBM was receiving, or on which it anticipated that it would receive, substantial plug compatible competition. The statement at the trial by Mr. Carey, Chairman of the Board, President and Chief Executive Officer of IBM, that tapes and disks were covered because “we, obviously, had to reduce our prices on them or go out of business and so they were very logical candidates for the Fixed Term Lease Plan”, aside from its character of confession and only attempted avoidance, was overstated factually. IBM’s plug compatible competition in the disk tape area did not threaten to drive IBM out of the business in those markets. In June of 1971, IBM’s plug compatible competitors had only 14.5% of the plug compatible disk market and 13.7% of the plug compatible tape market, and IBM’s worst case forecast, that is, “if IBM did nothing”, predicted that IBM’s plug compatible competition in total would capture 28.7% of the 2314/2319 plug compatible disk installations, 48% of the 3330 plug compatible disk installations, and 19% of the plug compatible

*300

tape market by 1976. Nor is Mr. Carey’s explanation that FTP was merely an experiment borne out fully by the record; it was directed specifically at markets in which plug compatible competition was of special concern to IBM and the selection of these areas was not on the basis of a random experiment to ascertain the effect of the plan, but to accomplish results with respect to these markets forecast in advance by its experts. The plan was extended beyond disks and tapes to printers from an apparent desire to lock up that market before Telex could start deliveries. On May 6, 1971, the Management Review Committee directed the task force to prepare a long term lease approach for disks and tapes “plus a strategy for memories and printers”. Telex was known also to be interested in memories.

F97. IBM did not place its CPU’s under the Fixed Term Plan. In fact, IBM raised its prices on its CPU’s and 360 memories to offset its peripheral price cut within two months after reducing its prices on its disk, tape and printer products under the FTP. Long term leasing represented a substantial change in IBM policy not only in the swing from short or open-ended leases, but in previously contemplated general pricing policy. Mr. Emery, a member of the task force wrote: “It was pointed out in most instances that any policy change which we now advocate for peripherals would have to be applicable to processors as there is no justification for different treatment.” One of the studies had concluded that long term lease plans “must apply across the board — peripherals not different”. IBM’s Management Review Committee at one point instructed the Cooley Task Force not to consider long term lease plans “since there does not appear to be a way of limiting such a plan specifically to the peripheral marketplace”.

F98. By January, 1971, IBM had determined that it needed a price increase on products not covered by FTP. The Management Review Committee was actively considering raising IBM’s prices in March, and on March 30, 1971, the Data Processing Group made a recommendation to IBM’s Management Review Committee ■ for a price increase on CPU’s. On April 7, 1971, Mr. Learson wrote Mr. Bietzel stating: “We believe at the moment that we should postpone any pricing action for another month .” In late June, IBM rejected the idea of placing CPU’s and memories under a Fixed Term Plan on the ground “this would prematurely erode the FTP concept to the entire product line, and, in addition, would be ineffective unless accompanied by some degree of pricing action.” IBM not only increased its prices on CPU’s and memories in July, 1971, but it was IBM’s, estimate that those price increases would offset IBM’s price decreases on disk, tape and printer products placed under the Fixed Term Plan. On August 6, 1971, Mr. Powell of IBM wrote: “I can support the position that the net effects of the FTP and price change will probably be a wash insofar as business volumes are concerned ! . The net effect of the FTP and price changes will not significantly increase [the customers’] total cost and no system decreases were forecast.”

F99. In March 1972, IBM announced its “Extended Term Plan” (ETP), a variation of the Fixed Term Plan having no substantially separate or different economic impact of consequence in this case. ETP also was optional, IBM customers having the opportunity to rent IBM equipment under the 30 day lease contract formerly utilized by IBM if they were willing to forego the price reductions provided in FTP. The defendant has now three long term lease plans pursuant to which it leases central processing units, tape drives and tape drive controllers, disk drives and disk drive controllers, printers, communications controllers, consoles, channels, and other products. In March of 1973, IBM announced a term lease plan which offers a four year lease on System 370 virtual storage processors.

F100. Surface justification for IBM’s turning to fixed term plans does

*301

not insulate its conduct in monopoly context from serious question. Since the mid-1950’s IBM and others have offered their customers the opportunity either to purchase computers or to lease them on short term leases basically cancelable on 30 days’ notice. Until the early 1960’s many customers were hesitant about making commitments longer than for a month or so at a time because of the difficulty of evaluating the rapid changes in EDP technology, which situation changed with increasing sophistication of customers and more general acquaintanceship with industrial developments and prospects. For a number of years prior to IBM’s announcement of the Fixed Term Plan in 1971, many of its competitors had offered lower prices on long term leases, with cancellation penalty clauses or with no cancellation option at all. By 1971, most of IBM’s competitors, including systems and peripheral competitors and leasing companies, were offering users long term lease options. IBM’s studies indicated that a long term lease plan on peripheral products, among other things, would reduce IBM’s costs through decrease in “churning” of IBM’s leased equipment at the same time and for similar reasons that its competitive position in relation to PCM’s would be enhanced. But preponderant evidence demonstrates that IBM’s fixed term plan was generated and implemented at the time it was with the primary intent and purpose of suppressing plug compatible competition and to maintain its monopoly power in the plug compatible disk, tape and printer markets and the general plug compatible market for peripheral devices.

F101. With reference to the FTP, as in the instance of the 2319 and memories, the intent and purpose of IBM in taking competitive action or reaction becomes important in view of its dominant position in the markets. Some of the evidence is equivocal. One of the difficulties lies in the inadequacy of the minutes of The Management Review Committee to clearly indicate the reasons for approving the actions complained of here. There was an abundance, or perhaps IBM would now think an overabundance, of documentation in lower echelon views, studies, computations, projections, forecasts and recommendations leading up to the action of the Management Review Committee, the top executive authority in IBM. In retrospect, and in view of the absence of full documentation at the top, inferences are arguable and have been argued to the effect that the predatory intent clearly indicated by task force or other processors of problems, and their related data, were disregarded or rej'ected by top management. But I have felt constrained to reject this bland construction by consideration of the record as a whole, and by the very organizational framework within which the record demonstrates such decisions and intents were initiated, formulated and pursued. In the first place I am doubtful that the intent of subordinate agents must be entirely disregarded in determining intent of a corporation represented, as all corporations are, by top management. Especially is this so in the case of IBM, which the evidence makes clear was finely tuned, organized and managed to reflect to top management the composite of a sophisticated, widespread and coordinated employee organization for the purpose of management decisions. In the absence of some clear record or indication to the contrary, it reasonably may be inferred that top management in adopting recommended actions or modifications not inconsistent with the data and recommendations submitted did not entirely reject their rationale and reasoning. Moreover, there is considerable direct evidence on vital points to indicate that top management itself did in fact subscribe to the anticompetitive views and objectives of lower echelons, the numerosity and pervasive nature of which preclude their disregard.

F102. One of many examples of the relationship of corporate investigation and processing to corporate decision, the comprehensive and systematic studies constituting input into top management,

*302

and the likely dependability of those studies, is furnished by plaintiffs’ Exhibit 107R, somewhat randomly selected. This exhibit is headed “Key Corporate Strategic Issue . . . Peripherals”. The objectives were listed as “A. To assess all pertinent factors affecting current and future impact of competitive compatible products on IBM’s worldwide business. B. To review current IBM strategies, policies and practices and to identify and prioritize exposure areas in relation to A above. C. To recommend actions required to produce an optimum IBM strategy for peripheral products.” After listing “External Factors” such as itemized “Federal Government Environment and Influence”, “Marketing Environment”, “Maintenance Environment”, “Manufacturing Environment”, “Pricing Environment”, “Engineering Environment”, and “Market Potential and Trends”, and after listing “Internal Factors” including “Ease of Competitive Interface” and “Legal Considerations”, the “Scope of Issue” is defined. “The scope of this issue”, says the memorandum, “shall be selected competitive compatible products which replace IBM products in an IBM computer system.” To obviate any uncertainty, “Competitive compatible products” are further defined as “system attached input/output and memory products.” In further refinement it is stated that “Input/output products shall include . . . magnetic tape drives and control units, direct access storage products and control units, impact printers and control units, card readers, card punches, reader/punches and control units.” “Memory products shall include main (buss connected) and large capacity storage (plug connected)”. “Excluded products” were “Central Processing Units (CPU)-eonsoles-paper tape prod-uets-communication products and control units-R.P.Q’s and non-standard prod-uets-non system attached products.” 'A comprehensive methodology utilizing line and staff resources, calling for “action program(s) for each pertinent factor”, and for the presentation of “program recommendations to MC” (Management Committee). Specific assignments were then made to various staffs, including “Marketing” and “Legal”, with an indication that additional line organization was “to be designated by DP Group and WTC.” Each assessment was to include,

inter alia,

“IBM Strategy/Policies/Practices Relating to this Factor”, “Identification and Priority of Major Exposure Areas” and “Recommendations for IBM Actions.”

F103. IBM’s Fixed Term Plan effectively suppressed the growth of plug compatible competition in the plug compatible disk, tape and printer markets and effectively contained IBM’s competition in the plug compatible market for peripheral devices. IBM’s plug compatible competitors’ share of the plug compatible disk market never exceeded 17.-6% after June, 1971. IBM’s plug compatible competitors’ share of the plug compatible tape market never exceeded 15% after June, 1971. Immediately upon the announcement of IBM’s Fixed Term Plan IBM instituted an in-depth tracking procedure to determine the effectiveness of IBM’s Fixed' Term Plan. On June 18, Mr. Rodgers reported to Mr. Learson that IBM had already signed up 16% of its entire installed disk, tape and printer base on the Fixed Term Plan. On June 24, Mr. Justice reported to Mr. Hume that Fixed Term Plans then covered 19% of IBM’s installed base on disks, tapes and printers, and by July 22, 1971, 40% of IBM’s disk, tape and printer installed base was covered by FTP. Mr. Learson wrote, “They are reporting a 40% coverage on FTP. Is this high or low with reference to what we expect. I consider this very important.” And the answer came back, “Our objective was to hit 40% coverage on files, tapes and printers combined by 12/31/71.”

F104. IBM’s plug compatible competitors as expected by IBM made competitive price reactions to the Fixed Term Plan. By June 17, Telex became the fourth plug compatible competitor to announce a price reduction; but even with

*303

such price reductions the plug compatible competitors could not successfully compete with the Fixed Term Plan. Mr. Rodgers estimated that the Fixed Term Plan cut IBM’s plug compatible competitors’ order rate by 50%. IBM’s tracking showed that 90% of its new disk and tape products, specifically its Merlin and Aspen products, were being installed under F.TP. This tracking was in accordance with IBM’s estimate made in June, 1971, that on the Merlin FTP acceptance would be 95% of the rental base and; in fact, IBM’s acceptances have been at the 95% level anticipated by IBM. IBM in its internal documents described the competitive constraints that its FTP imposed upon plug compatible competition, particularly plug compatible competition for the new 3330 disk and 3420 tape products: “One competitor has already announced a PC 3330/3830 and others can be expected to follow soon. . . The competitor will offer long term leases similar to IBM’s with the base rental initially 10% below ours and declining 5% per year. The competitor will face a new environment, however, in that the bulk of his early installations will represent conversions from PC or IBM 2314’s rather than plug for plug replacements of installed 3330’s. This will be due to the user’s reluctance to break the IBM contract due to the penalty payment required. As a result, the competitor will face harder selling and harder installation since he has not yet shown the capability to provide sys-terms, conversion, and application support. . . While the PC competitors will make a strong effort, it is assumed that near-term 3330 erosion will be contained until the FTP contracts approach maturity. By that time, Winchester, Iceberg, the 3330A/B and the 333M will all be available as customer options and should hold the market for IBM. . . The 3330/3830 FTP will receive wide acceptance in the marketplace. It is estimated that 95%. of the rental base during 1971 and 1972 will be under FTP . It is further assumed that the savings engendered by the FTP will increase the migration rate from the IBM and PC 2314-type products into the 3330/3830.”

FI 05. By the end of 1971 the effect of FTP on plug compatible competitors was measured by IBM. DP Commercial analysis reported in December, 1971, “Since the announcement of FTP, there has been a 62% decrease in PCM tape monthly sale rate.” In disk drives Commercial Analysis reported, “In 4.5 months since IBM’s FTP announcement, the PCM monthly sales rate is down to 475 spindles per month, off 48% compared to 905 per month during the first 5 months of 1971.”

F106. The plaintiffs claim that the internal Integrated File Adapter for 3330 disk drives used with the 370/135 coupled with a price substantially less than the external 3830 Mod. 2 equivalent, constituted a discriminatory physical and economic tie between the IFA and the 370/135 central processing unit; and that the internal ISC with the rental price substantially less than the external stand-alone equivalent constituted a discriminatory physical and economic tie of the ISC to the 370/145, 158 and 168 central processing units. These integrated controllers consist of control electronics contained within the same boxes or frames as the CPU’s or other EDP equipment and using a part of the CPU resources to perform their functions. IBM offered customers the option of acquiring integrated controllers at prices lower than stand-alone controllers and there is no question that prices were substantially lower for the integration. Unlike the situation which existed in respect to the Mallard, the fixed plan, leases and IBM’s actions directed to memories hereinafter considered, the integrations mentioned are not shown to have been dictated by specific predatory objectives on the part of IBM. While some question is raised in the evidence as to the economic justification for the extent of the price reductions, and some justifiable suspicion may exist as to predatory intent, a finding that such intent was a significant motivation for the

*304

integration is not deemed warranted by the evidence in view of the preponderant showing that these integrations represented a legitimate technological and performance advance consistent with trends in the industry and at significant decreased cost.

F107. While cost and performance justifications may have existed to an extent, it is found that IBM lowered the price of its FET monolithic memory products and raised prices on its CPU with the primary purpose of creating barriers to entry for potential plug compatible memory competitors. The 370/156 and 165 were introduced in June, 1970, with magnetic core memories which were contained in boxes external to the CPU, but which were cable-connected to the CPU. The main memory for the 155 and 165 consisted of the 3360 processor storage together with a high speed monolithic (bipolar technology) buffer storage which made up a two level “heirarchical” type memory. IBM was greatly concerned with the high market penetration which independent manufacturers of plug compatible memories for System 370 threatened, it being estimated by it that such penetration might amount by 1976 to as much as 23%. Pending the availability of improved technology, IBM’s Management Review Committee explored and adopted a memory strategy which repriced memory by reducing prices and by at least partly offsetting this reduction by an increase in CPU prices. Its studies indicated that plug compatible memory companies could become viable competitors in supplying memory for IBM CPU’s by offering their products at $6,000 per megabyte if that price was under IBM’s prices; that is, viability and entry would depend upon the slope or pricing level of IBM’s FET monolithic memories.

F108. IBM formed another task force in March, 1971, which was charged with the mission of developing a “memory strategy” which would optimize profit and revenue for IBM and also control the market penetration that was fore-casted for plug compatible memory products. The work of this memory task force included an attempt to fix a price for IBM’s monolithic FET memories that would influence potential plug compatible competitors to stay out of the market. The IBM Management Review Committee set the monthly rental price for IBM’s FET monolithic memory at $5,200 per month per megabyte, which was less than the amount reporting experts had indicated a potential competitor would be required to charge in order to enter the market and be profitable and viable. The monthly rental price for the 158 CPU was raised from the $20,600 charged for a 155 CPU to $30,700 for the 158 CPU to offset the decrease in price for the FET memory, the percentage increase in price being higher than the percentage improvement in performance. The monthly rental price for the 168 CPU was raised from the $36,400 charged for a 165 CPU to $48,600 charged for a 168 CPU to offset the decrease in price for the FET memory, the percentage increase of the 168 CPU when compared to the 165 being higher than the percentage improvement in performance. Neither the design of the 370/158 and 370/168 nor the price of $5,200 per megabyte per month prevented Telex from planning competing memory products and in March, 1973, it announced that it would market memory components for 158 and 168 systems. Control Data, Itel, Ampex and Intel have announced memories for IBM’s 370/158 and 370/168 systems at prices substantially below IBM’s price.

F109. The same task force studied a proposal that “concurrent with the introduction of FET memories, the minimum entry sizes of the 145, 155, 165 and 195 be raised.” A chart considered by it showed that a “minimum” memory of one-fourth megabyte on the 145 would “protect” 71% of the 145 memory from exposure to competition, a “minimum” memory of one-half megabyte on the 155 would “protect” -42% of the 155 memory; and, a “minimum” memory of one megabyte on the 165 would “protect”

*305

37% of the 165 memory from competition. Plug compatible memory vendors were forecast to receive a $35 million share of the market for memories on System 360, $623 million share for memories on System 370 if IBM did not take preventive action. One of the financial analysts, Hochfeld, in March, 1971, raised the question with his superiors of the legality of increasing memory minimums and the danger of a civil damage suit. He testified at the trial that based upon the studies and analysis that he had done it was his opinion that when IBM dropped its price, of FET monolithic memories for the 158 and 168 to $5,200 per megabyte it made it impossible for anybody to enter the field and be viable competing with IBM on the 158 and 168 memories.

F110. Beyond the matter of pricing, plaintiffs contend that IBM “. unlawfully bundled” a minimum FET monolithic memory “under the covers” with its 370 CPU’s so as to protect a substantial portion of the memory from exposure to plug compatible competition. There is further indication of an anti-competitive design in the investigation of proposals for minimum memories in IBM CPU’s, but, again, it appears that there are so many practical and technical justifications for the integration of memory as to raise substantial question concerning the validity of plaintiffs’ contention on this point. Memory is an essential component of the central processing unit since all processing units must have storage to operate. The integration of memory generally tends to reduce cost and improve performance. If memory is packaged separately in a separate box it does require additional frames and covers; it requires its own separate cooling system, power supply, accoustic baffles and electrostatic shielding. It also requires extra cables. The speeds and circuits used in CPU’s have increased dramatically in recent years; to best take advantage of this faster technology it seems desirable, all other things being equal, that memory be integrated into the CPU’s so that the wire lengths between logic and memory elements be reduced as much as feasible. As a result of cost and performance advantages integration is accepted as an industry standard and an objective of good engineering design. Historically most CPU’s designed by IBM and various other EDP suppliers have included a minimum main memory integrated within the CPU. Historically, also, whenever a main memory has been integrated within a product it has been included in the price of that product and not separately priced. Reduction in the size of memory components has made it practical to integrate more memory with the logic elements of the CPU.

Fill. The preponderance of the evidence establishes in the case of IBM’s 370/158 and 370/168 CPU’s that IBM’s integration plan was adapted primarily to achieve the cost and performance improvements made possible by reduction in size of the memory component. Other companies are still able to attach their memories to IBM 370/168 and 370/158 Systems and Telex has recently announced its intention to do so. The IBM 370/155 and 370/165 primarily use magnetic cores for memory, and because of its size such magnetic core memory is sold in a separate box. At the time the 370/155 and 370/165 were announced, advances in semi-conductor technology were rapidly obsoleting core memory technology. The main memories of the IBM 370/158 and 370/168 CPU’s announced by IBM on August 2, 1972, are made up of FET semi-conductor circuitry. The trend toward miniaturization of computer circuitry no doubt will continue in the future to permit further integration. Telex itself anticipates that memory chips containing 8,000 circuits will be commercially available by the time it delivers its replacement memories for 370/158 and 370/168 Systems. Using an 8,000 circuit chip, 4 megabytes of storage can be housed in a space approximately 21 inches by 15 inches by 6 inches. Chips containing 64,000 circuits are presently under development at a number of laboratories. IBM antici

*306

pates chips containing 256,000 bits by the 1980’s. In the court’s view it would not be a proper application of the antitrust laws under the circumstances shown by the record to preclude or discourage the utilization of advancing technology by this type of integration. The testimony, and particularly Hoch-feld’s, affords some indication that one of the motivations for the “bundling” of the minimum. memory by IBM was to reduce exposure to plug compatible competition. Yet, dominant justifications on other grounds lead the court to believe that this was not unlawful in and of itself, however significant some of the related testimony may be in indication of a predatory intent primarily directed in actuality to other areas as herein found.

Filia. There was no evidence that IBM reduced prices below cost and a reasonable profit. Indeed, when announced the profitability of the 2319 disk storage units, the 370/158 and 168 CPU’s and CPU memory elements were anticipated to be in excess of 20%. Likewise, at the announcement of FTP it was anticipated that the profitability of the products to which it applied would be at least 20%. Those profit margins in part, of course, would have been achieved by obtaining leases of products which would have otherwise been made by Telex and other PCM’s. Those price reductions are found to be predatory.

F112. IBM’s growth and success in the industry have been due in substantial measure to its skill, industry and foresight. It has tended to set the standard for quality in the EDP industry for products and services. It has met notably favorable response in the market, and has been deeply involved in the phenomenal growth of the industry since almost its beginning. Each succeeding generation of IBM products has represented some technological improvement over the preceding generation and has involved development of new processes, storage devices, input/output devices and software. In the approximately twenty years that the EDP industry has been in existence IBM has introduced more than 600 products. Some of these products include major technological innovations. By virtue of its own research and development, IBM has obtained more than 10,000 patents which are freely licensed. I therefore cannot fully agree with Telex’s contention that “IBM did not gain, nor has it maintained its position in the industry through skill, industry and foresight”. No doubt it gained a dominant position in the industry through a praiseworthy degree of these qualities. Whether there was anticompetitive conduct that went along with them in recent years prior to 1969, the record does not disclose. The real problem here is notwithstanding this, whether IBM has maintained its monopoly position, or attempted to do so, by unlawful conduct since 1969. In the respects determined here in the critical period at least it must be recognized that its diligence and foresight have included the competitive studies and the anti-competitive objectives and intent heretofore found, and that particularly as applied to this case have included an attempt to substantially constrain or destroy its plug compatible peripheral competition by predatory pricing actions and by market strategy bearing no relationship to technological skill, industry, appropriate foresight or customer benefit. With such intent and objectives manifest with respect to its plug compatible competition, it is understandable that the defendant should be particularly opposed to the recognition, as essentially separate entities or markets, of what was initially a part and parcel of its own internal business over which it exercised legitimately a 100% control — the peripherals attached to its own developing system. But we find unconvincing the idea that separate markets or submarkets actually recognized by IBM itself in this dynamic and amazing industry could not have been developed eventually from IBM’s prior lawful domination of it; or that the objectives and planning of such a pres

*307

ently dominating force against the competition of the peripherals could somehow be deemed dissipated among lower echelons of this great organization and not considered to be reflected in

the

competitive actions of top management, or that, if reflected, should be held innocuous or futile,, or at all events lawful, as competitive weapons.

VII

IMPACT AND DAMAGES FOR ANTITRUST VIOLATIONS

F113. Telex now claims that its damages resulting from unlawful predatory acts of IBM total $361.3 million, and that these damages are comprised of $257.7 million in deprivation of market share, $92.3 million for lost rental profits and $11.3 million for lost sales profits.

F114. The record leaves little room to question that the acts, conduct and intent on the part of the defendant found herein to have been in violation of Section 2 of the Sherman Act proximately caused substantial impact and damage to the business of the plaintiffs, and the court so finds. Aside from its quantification beyond substantiality, such impact is to be found in the circumstantial evidence as a whole, the direct evidence by way of opinion and judgment of Telex officers and witnesses, some evidence elicited by the cross-examination of the defendant’s witnesses, the admitted fact that such acts were competitive responses on the part of IBM to the inroads the plug compatible competition was threatening and making against its market position, by reasonable inference in view of IBM’s market domination and organizational effectiveness that these responses must have advanced their purposes to some appreciable degree, and from a number of statistical indications to be referred to in some detail in connection with a determination of the amount of damages to which plaintiffs are entitled.

F115. As to any specific amounts of damages awardable in this case the evidence is less clear, and as justification for the sums asked for quite unsatisfactory and insufficient. There is evidence tending to show that Telex had a taxable income of $12,462,000 for fiscal year 1971; for fiscal year 1972, after the effects of the 2319 announcement were ex-pectable, Telex’s loss was ($913,000), and for fiscal 1973, after 2319 and the FTP influence, its pre-tax loss was estimated to be approximately ($7 million). Its gross receipts of $77,.843,000 in fiscal year 1971 declined to $56,076,000 in fiscal 1972. At the end of fiscal year 1971 (after 2319 but before FTP) the market value of Telex’s stock was $19 a share or $197,999,000 for all shares; two years later, after FTP, it was $4.60 per share or $48,143,500 for all shares, a loss in value of $149,855,500. According to Telex’s November, 1970, forecast and product plan, a profit of $33,837,000 would have been produced in fiscal years 1972 and 1973. Telex’s initial November, 1970, forecast showed a projection of 8,910 units,, and Telex actually shipped 4,517 units through March 31, 1973. Following the 2319A and B announcements and FTP, Telex had been able to make third party sales of $30 million from January 1, 1971 to March, 1973, while in 1969 and 1970 almost $120 million in Telex tapes and disks were sold to third parties. Prices at which equipment could be sold have eroded some 35% as compared to the 1969 and 1970 prices. Marketing expenses increased. Backlog and order rates were reduced. Recruitment of adequate personnel became more difficult as uncertainty as to Telex’s future viability increased. “Front end” expense has been increased by inadequate concentration of products and services. Competition among Telex and other plug compatible manufacturers for remaining business has intensified. Telex’s ability to secure financing has been impaired, and Telex has had to pay more for financing. There is some evidence that Telex’s problems are not unique in the plug compatible market; that some companies following 2319 and FTP failed in or abandoned their plug compatible business. There was a “plateau” that existed in Telex’s installations growth from about November,

*308

1971, to about July, 1972. From September, 1969, to November, 1971, Telex’s domestic installation or “population” of peripheral devices increased from 1,000 units to 8,000 units. From about November, 1971, to about July,, 1972, the last month included in the Telex chart reflecting such plateau, Telex’s installations neither substantially increased nor decreased.

F116. There was evidence developed by IBM at the trial that Telex’s forecasts were untried and unreliable upon which to base damage and impact claims and that in any event, the recorded business experiences of Telex were explainable by various factors and influences unconnected with IBM’s business practices. It dismisses Telex’s income tax returns as unreliable for the purpose of indicating damages, pointing out that in 1972 Telex changed for tax purposes from the accrual method of accounting for certain revenues to the installment method, thereby slowing down the tax reporting of revenues on certain leasing company transactions. The tax return handling of depreciation of Telex’s retained equipment and interest expense is also attacked. The necessity of a detailed and complex analysis including a reclassification of all items of expense and revenues is emphasized by IBM and in the absence of such a study and restatement it is asserted that the income tax returns have little or no value for comparative purposes or even as statements of actual income. The significance of stock price changes is also dismissed by IBM, it being claimed thát there is no evidence relating the fluctuations or declines in the market prices for Telex stock to any or all of the IBM actions of which Telex makes complaint, and that such stock fluctuations are assignable in whole or in part to adverse publicity over Telex’s accounting methods and other circumstances over which IBM had no control. Finally, IBM demonstrated with some persuasion that Telex’s difficulties at least in part stemmed from its own problems of management, testing, service, organization and personnel. Defendant’s charts 143, 144 and 145 indicate that while Telex disk drive spindles and tape drives installed on CPU’s manufactured by IBM leveled off markedly in number for the period 1971-1972, as Telex showed, similar non-Telex installations continued to appreciably advance. In the latter connection it is notable, however, that about the time the acts complained of by Telex were becoming effective to the extent they did, there-was a discernible diminution in the rate of increase of even these non-Telex installations. Telex’s response to the claimed adverse effect of its management and other internal problems is that they are more or less normal to the industry and that most of them have been intensified to critical stages because of IBM’s predatory actions. Telex says that the February 7,. 1973, product forecast from which IBM concludes that “Telex has turned the corner” is only a manufacturing capability forecast in support of which at the time of the trial it was not assured that adequate financing and third party sales would be available that the order rate is nowhere near the forecast,'that since the first of the year there has been a significant reduction in the level of product shipments and new production units shipped for this fiscal year are 50% lower than forecast.

F117. My task would be simpler if as to each element of claimed damage clear and unhampered causal lines could be discerned, leading to IBM’s predatory acts without passing through or commingling with the literally hundreds of other circumstances which may have influenced the figures. But in cases like this, if not in every complex case, it is humanly impossible to trace, find, and specify in detail and quantify in effect the numerous circumstances which cause or contribute to financial consequences. By such a process the determination of damages by court or jury could be bogged down in almost any case or rendered more inaccurate than a considered judgment appraisal of the combined effect of all actionable elej

*309

ments duly considered by an informed fact finder after elimination of the influence of extraneous causes. The record fragmenting of judgment might be either a mere exercise in futility or a mechanical allocation of the result of the aggregate judgment at best. It is the damage that must be quantified rather than the respective weights or contributions of the unlawful causes so long as each has substantial effect upon the damages suffered by the injured party so as to constitute their proximate cause. Notwithstanding the difficulty involved, I have found that there is reasonable basis in the evidence to fairly approximate the damages to which plaintiffs are entitled as proximately caused by the unlawful acts and conduct of the defendant.

FI 18. The largest component of the damages claimed by plaintiffs relates to the deprivation of market share, based on the difference between Telex’s forecast of November, 1970, and Telex’s forecast of January 12, 1972. Product rental prices in effect prior to 2319 were used, product life was based upon IBM estimates, and Telex’s usual product and marketing expenses were used. Telex’s November, 1970, forecast was prepared prior to the predatory actions of IBM except for the 2319A announcement. The number of units was smaller than those forecast for Telex by IBM’s April 16, 1971, internal forecast, which took into consideration the impact of IBM’s 3420 tape price cut and IBM’s 2319 disk drive pricing cut. Telex’s January 12,, 1972, forecast reflected anticipated product shipments greater than Telex’s actual business turned out to be over the same period. The difference between the Telex November, 1970, and January 12, 1972, forecasts was less than IBM’s internal documents indicated that it expected Telex to receive prior to IBM’s Fixed Term Plan announcement. IBM’s internal documents indicated that IBM calculated its increased profits that would result from adoption of the Fixed Term Plan leases for tapes, disks and printers to be $466 million. Using the latter assumption and considering that in 1970 Telex was installing approximately 63% of the non-IBM plug compatible tape drives, 31% of such disk drives and 100% of such plug compatible impact printers, a calculated loss of market share from FTP would be $218.-67 million. In support of its loss of market claim Telex also cites income tax records and stock prices as indicated above, but relies primarily upon a comparison of documents described as the November, 1970, forecast and a January 12, 1972, forecast.

F119. There are circumstances relating to these forecasts which preclude their acceptance at face value notwithstanding some supportive evidence of other types. The handwritten documents assembled as the November, 1970, forecast were prepared by different people at different times. No part of the document is actually dated and the text of the initial pages indicates a date of preparation after February 18, 1971, even though Telex prepared an entirely new and substantially reduced forecast in February, 1971. Neither the November,, 1970, forecast, nor the January 12, 1972, forecast is supported by the usual written assumptions which the evidence shows were utilized in their preparation. Forecast assumptions were “very informal and very unstructured” in the words of one of plaintiffs’ witnesses. The Telex calculation is based upon the highest forecast it made for selected products compared with the lowest forecast made for those products and the forecasts had no substantial history of reliability or accuracy. Under such circumstances, while these forecasts have been considered to be good faith evaluations in the course of business operations and for business rather than litigation purposes and thus entitled to consideration, calculations based upon them must be weighed with due regard for their limitations and other evidence. Moreover, plaintiffs’ calculations have assumed that variation in units forecast was caused solely by the IBM actions complained of, or to be found as unlawful by

*310

the court, without giving weight to the effect of established internal difficulties within Telex over which IBM had no control and the claimed unlawful integration action that has now been found to have been nonactionable.

F120. These and other limitations in the sufficiency of plaintiffs’ proof to support their claim for deprivation of market share in full leave the question whether any such claim is thereby defeated entirely or, if not, what amount of damages has been established by way of fair approximation on the evidence in view of the fact that IBM’s predatory actions have deprived the plaintiffs of the opportunity of positive proof of what their experience would have been in the absence of such action. There must also be evaluated in any award the question of to what extent the uncertainty of proof relates to IBM’s predatory action and to what extent, if any, it relates to any inexcusable failure on the part of the plaintiffs to submit evidence reasonably available to it. As to the latter problem, if trial counsel is not to be hindsighted, I am inclined to believe that generally speaking the plaintiffs must be considered to have submitted the best proof of which the nature and complexity of the case reasonably were susceptible. There is no evidence that there were any better forecasts available nor any indication that those used were contrived particularly for the purpose of prospective litigation. The opinion evidence was based on plaintiffs’ theory of liability and as it turns out it would have been more applicable had it been based upon the court’s findings after trial of what conduct was lawful or unlawful rather than upon pre-trial expectations or hopes concerning bases of liability. Beyond such forecasts and opinions, it is difficult to see how loss of market share could be established with any specificity, or how elements of loss of market share could be traced and evaluated much more in detail with reference to various factors that may have entered in by way of possible influence. To attempt by expert testimony to evaluate and weigh each individual factor and to eátegorize and evaluate the relative influence of each alleged predatory act against the possibility that the court might not sustain contentions as to some of them might so complicate a trial, extend the evidence and compromise the trial position of a party from the inception as fairly to be considered impractical and unjustified. While the question is not free from doubt, it appears to the court that any uncertainty concerning the amount of the loss of market share is not so ascribable to fault on the part of plaintiffs as to deprive them of the benefit of the rule that where the existence of impact and damages have been shown by a preponderance of the evidence, reasonable approximations of the extent of damage based upon the reasonably available evidence is not to be rejected.

F121. It has been found that sufficient evidence was introduced to show preponderantly that plaintiffs suffered substantial damages in an ascertainable approximate amount from the unlawful acts of the defendant in deprivation of the market share that it would have enjoyed had the unlawful acts of the defendant not been committed, and that taking into consideration the strengths and weaknesses of the plaintiffs’ proof as to damages in view of the whole record,, eliminating the results of internal, collateral or other considerations over which IBM had no control and unrelated to its unlawful acts, and excluding damages for lost rental profits and lost sales profits hereinafter to be separately considered, such element of damage reasonably awardable herein amounts to $70 million, subject to the adjustments made in F124 with respect to the net antitrust damages to be awarded in favor of Telex.

F122. It has been found that sufficient evidence has been introduced to show preponderantly that plaintiffs suffered, from lost rental profits, substantial damages in an ascertainable approximate amount from the unlawful' acts of defendant, and that taking into consideration the strengths and weaknesses of plaintiffs’ proof in view of the whole record, and eliminating the results

*311

of circumstances over which IBM had no control and which were unrelated to its unlawful acts, such element of damage reasonably awardable herein amounts to $39 million. This sum represents the difference in profits from Telex’s installed units based upon rental charges actually received and that would have been received in view of defendant’s unlawful acts, as compared to rental profits that would have been received on rental prices in effect prior to IBM’s 2319 and FTP announcements, with elimination of factors not attributable to IBM. Approximately $20 million of this total amount is made up of past and future claimed lost rentals on all Telex shipments prior to April 1, 1972. The other $19 million is lost rentals on shipments reasonably to be anticipated as taking place after March 31, 1972. These findings are subject to the adjustments made in F124 with respect to the net antitrust damage to be awarded in favor of Telex.

F123. Telex’s claim of $11.3 million damages for “lost sale profits” from leasing company transactions is based on a reduction in the price and sale “multiples” as a result of IBM’s pricing actions. Telex claims $8.

This text is long and has been trimmed here. Open the source document for the complete record.