Full opinion text
MEMORANDUM O’NEILL, District Judge. AMI brought this action asserting federal and state law claims against IBM, which filed several counterclaims. The parties agreed to try the liability aspect of AMI’s claim under Section 1 of the Sherman Act separately. AMI’s claim has two parts: AMI alleges that IBM’s net pricing policy constitutes an unlawful tying arrangement and that IBM’s Installation and Warranty Service Charge constitutes an unreasonable restraint of trade. The issues were tried before me non-jury; this memorandum constitutes my findings of fact and conclusions of law. See Fed.R.Civ.P. 52(a). Jurisdiction over the Section 1 claims is based on 28 U.S.C. § 1337. Suit is brought pursuant to Section 4 of the Clayton Act, 15 U.S.C. § 15. NET PRICING An IBM 308X net priced upgrade is an MES (Miscellaneous Equipment Specification) or feature (a specific set of computer products) that is installed in an IBM model 308X computer by removing parts from the computer and replacing them with different parts. Ritchie, Tr. 1076-77; Rizzo, Tr. 1188-90. Generally, the purpose of any up grade (308X upgrades included) is to enhance the performance of the computer, often to increase the capabilities of a used computer to match the performance level of a newer model. Upgrades include model upgrades (MIPS upgrades), increases in memory capacity (memory upgrades), and increases in the number of computer channels (channel upgrades). See Ritchie, Tr. 1077; Bigando, Tr. 1232; Lynn, Dep.Tr. 62. IBM 308X net priced upgrade contracts provide that the installation and removal of parte is to be performed by IBM employees, and that the removed parte become the property of IBM and are returned to the company. Ritchie, Tr. 1086-87; Rizzo, Tr. 1188-90; Levin, Tr. 78-80, 744; PX 139; PX 140; PX 138. IBM issues its net priced upgrade customers a credit for the parte removed by IBM engineers during the installation of a net-priced MES. E.g., Levin, Tr. 78; PX 457; PX 145. A customer who purchases a net priced upgrade from IBM is not charged separately for the labor associated with performing the upgrade. See DX 829. The customer receives a single price quotation for the final, installed product. IBM’s 308X product line consists of 14 models with a performance power range of 3 to 30 million instructions per second (MIPS), and a price range of $960,000 (3083E) to $6,300,000 (3084Q). See DX 1799. Upgrades are net priced only if they involve the removal and return of parts (such as TCMs) from the upgraded computer. The majority of MIPS upgrades, see Levin, Tr. 78, and numerous memory upgrades are net priced by IBM. Upgrades which do not require the removal of TCMs or other parts (such as the 3083J to 3081K model upgrade) are not net priced and are optionally available from IBM on an SWRPQ basis; ie., without IBM’s labor included. An upgrade purchased on an SWRPQ basis may be installed by third parties, such as AMI, or by IBM, if the customer chooses and pays for the service. All 308X channel upgrades and many memory upgrades are available on an SWRPQ basis. AMI asserts that IBM’s net pricing of 308X upgrades constitutes an unlawful tying arrangement under Section 1. As ultimately defined by AMI, the alleged tying product consists of the IBM parts necessary to fabricate and install the equivalent of 308X net priced upgrades, and the tied product consists of the engineering services involved in the fabrication and installation of such upgrades. AMI’s Post-Trial Reply Memorandum of Law (hereinafter “A.R.M.”) at 9. I. A tie exists when a seller refuses to sell a product (the tying product) alone and insists that any buyer who wants it must also purchase another product (the tied product). See L. Sullivan, Handbook of the Law of Antitrust § 150, at 431 (1977). “[T]he Sherman Act does not prohibit ‘tying’, it prohibits ‘contracts] ... in restraint of trade’.” Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2, 21 n. 34, 104 S.Ct. 1551, 1563 n. 34, 80 L.Ed.2d 2 (1984). Only a tying arrangement which imposes an unreasonable restraint of trade is unlawful. Certain tying arrangements are per se unlawful under the antitrust laws; that is, they are deemed unreasonable as a matter of law and “no specific showing of unreasonable competitive effect is required.” Fortner Enterprises v. United States Steel, 394 U.S. 495, 498, 89 S.Ct. 1252, 1256, 22 L.Ed.2d 495 (1969) (“Fortner I”); see generally Northern Pacific Railway Co. v. United States, 356 U.S. 1, 5, 78 S.Ct. 514, 518, 2 L.Ed.2d 545 (1958) (respecting per se unlawful arrangements in general: “[T]here are certain agreements or practices which because of their pernicious effect on competition and lack of any redeeming virtue are conclusively presumed to be unreasonable and therefore illegal without elaborate inquiry as to the precise harm they have caused or the business excuse for their use.”). A tying arrangement which does not warrant per se condemnation may be found to violate the Sherman Act under a rule of reason analysis. Hyde, 466 U.S. at 17-18, 104 S.Ct. at 1560-61; Bogus v. American Speech & Hearing Ass’n, 582 F.2d 277, 287 (3d Cir.1978). Per se rules and the rule of reason are mechanisms “employed ‘to form a judgment about the competitive significance of the restraint.’ ” NCAA 468 U.S. at 103,104 S.Ct. at 2961 (quoting National Soc’y of Prof. Engineers v. United States, 435 U.S. 679, 692, 98 S.Ct. 1355, 1365, 55 L.Ed.2d 637 (1978)). II. AMI contends that IBM’s net pricing of 308X upgrades is per se unlawful. A.M. at 15. To establish a per se tying violation, plaintiff must show: 1) the existence of two separate and distinct products or services and an agreement conditioning the sale of the tying product to the purchase of the tied product; 2) that the seller possesses sufficient economic power with respect to the tying product to restrain free competition appreciably in the market for the tied product; and 3) that the seller has thereby foreclosed a “not insubstantial” amount of interstate commerce in the tied product. See, e.g., Hyde, 466 U.S. 2, 104 S.Ct. 1551; Fortner I, 394 U.S. 495, 89 S.Ct. 1252; Columbia Pictures Indus, v. Redd Horne, Inc., 749 F.2d 154 (3d Cir.1984); Bogus, 582 F.2d 277 (3d Cir.1978). The Supreme Court’s decision in Hyde, 466 U.S. 2, 104 S.Ct. 1551, suggests that an inquiry into the legality of a tying arrangement under the per se rule requires an elaborate economic analysis. I conclude that IBM’s net pricing policy does not warrant per se condemnation. AMI has failed to establish that IBM possesses the requisite market power. AMI has also failed to prove that it, or anyone, has been foreclosed from a viable business opportunity as a result of IBM’s conduct. My finding that AMI has not been deprived of a viable business opportunity implicates several of the concerns underlying the Sherman Act and the per se rule against tying in particular. A. Application of the per se rule against tying requires proof that the seller possesses economic power in the market for the tying product. The degree or form of economic power which must be established has been described in various ways. The Supreme Court, however, has consistently held that proof of the existence of monopoly power is not required. In Times-Picayune Publishing Co. v. United States, 345 U.S. 594, 611, 73 S.Ct. 872, 882, 97 L.Ed. 1277 (1953), in an attempt to define the requisite economic power, the Court referred to “the wielding of monopolistic leverage”, or the exploitation by a seller of “his dominant position in one market to expand his empire into the next.” This language was “construed” in Northern Pacific as requiring only “sufficient economic power to impose an appreciable restraint on free competition in the tied product.” 356 U.S. at 11, 78 S.Ct. at 521. “[T]he source from which the power is derived and whether the power takes the form of a monopoly or not” were said to be irrelevant. Id. In Fortner I, the Court explained that "economic power over the tying product can be sufficient even though the power falls far short of dominance and even though the power exists only with respect to some of the buyers in the market.” 394 U.S. at 502-03, 89 S.Ct. at 1258. For purposes of assessing economic power, the Court stated that the proper focus is “whether the seller has the power to raise prices, or impose other burdensome terms such as a tie-in, with respect to any appreciable number of buyers within the market.” 394 U.S. at 504, 89 S.Ct. at 1259. In United States Steel Corp. v. Fortner Enterprises, Inc., 429 U.S. 610, 620, 97 S.Ct. 861, 867, 51 L.Ed.2d 80 (1977) (“Fortner II”), the Court repeated that there is no requirement “that the defendant have a monopoly or even a dominant position throughout the market for a tying product.” Instead, the focus of attention is on “whether the seller has the power, within the market of the tying product, to raise prices or to require purchasers to accept burdensome terms that could not be exacted in a completely competitive market.” Id. (footnote omitted). “In short,” said the Court, the issue is “whether the seller has some advantage not shared by his competitors in the market for the tying product.” Id. In its most recent tying decision, the Court emphasized the importance of proof that the seller has “some special ability— usually called 'market power’ — to force a purchaser to do something that he would not do in a competitive market.” Hyde, 466 U.S. at 13-14, 104 S.Ct. at 1558-59. The focus is on the seller’s power to “force” or “coerce”: [T]he essential characteristic of an invalid tying arrangement lies in the seller’s exploitation of its control over the tying product to force the buyer into the purchase of a tied product that the buyer either did not want at all, or might have preferred to purchase elsewhere on different terms. Id. at 12, 104 S.Ct. at 1558. Proof of a seller’s power to force involves proof that the seller enjoys “significant market power”, id. at 26, 104 S.Ct. at 1566, or a “dominant market position.” Id. at 27, 104 S.Ct. at 1566. 1. AMI relies in part on evidence of IBM’s market share to support its contention that IBM possesses sufficient economic power to compel application of the per se rule. Market share is calculated by reference to a relevant product and geographic market. Briefly stated, “[t]he outer boundaries of a product market are determined by the reasonable interchangeability of use or the cross — elasticity of demand between the product itself and substitutes for it.” Brown Shoe Co. v. United States, 370 U.S. 294, 325, 82 S.Ct. 1502, 1523-24, 8 L.Ed.2d 510 (1962). “[T]he purpose of market definition is to determine whether market power exists”; therefore, “market definitions must account for all factors affecting the ability of the [defendant] to raise prices or restrict competition.” Weiss v. York Hosp., 745 F.2d 786, 826 (3d Cir.1984), cert. denied, 470 U.S. 1060, 105 S.Ct. 1777, 84 L.Ed.2d 836 (1985). For purposes of determining IBM’s economic power with respect to the tying product, AMI defines the relevant market as the market for large scale main frame computers. E.g. A.P.F. at If 251. According to AMI, a large scale main frame computer is one that is, at the time of its introduction into the market place, among the largest in memory capacity, the fastest in computing speed, and the most expensive of computers available. Levin, Tr. 752. Large scale mainframes, says AMI, typically are priced in excess of $1.75 million. Levin, Tr. 753. IBM’s revenues from the sale of new large scale mainframes (including MES sales) from 1981 to 1985 exceeded $16 billion (1981: $1.918 billion; 1982: $3.07 billion; 1983: $3.589 billion; 1984 $3.807 billion; 1985: $3.666 billion). PX 482; PX 15; PX 48. Based on AMI’s definition of what constitute non-IBM large scale main frames and its calculation of the revenues received by firm’s selling such products, IBM’s share of revenues of all sales of new large scale mainframes was 70.2% in 1981, 78.0% in 1982, 79.0% in 1983, 78.6% in 1894, and 75.9% in 1985. PX 482; PX 15, PX 48. Also based on AMI’s definition of what constitute non-IBM large scale main frames, IBM’s share of total shipments of large scale main frames (presumably shipment of MESs is included in IBM figures) was 61.0% in 1981, 70.1% in 1982, 76.0% in 1983, 74.5% in 1984, and 65.0% in 1985. PX 483; PX 15, PX 48. Standing alone, AMI’s market share evidence tends to show that IBM enjoys substantial economic power. However, AMI’s definitions of large scale mainframes and the relevant market are flawed in several respects and tend to overstate IBM’s market share and power. AMI does not expressly include upgrades of large scale main frame computers in its relevant market definition, but does include IBM’s revenues from the sale of 308X upgrades in its calculation of IBM’s share of the purported relevant market. If it is AMI’s contention that upgrades do not belong in the relevant market, then AMI’s calculation of IBM’s share of the relevant market must be reduced accordingly. However, upgrades of large-scale main frame computers belong in the relevant market. Professor Richard C. Levin, AMI’s expert economist, included parts and labor required to upgrade large scale main frame computers in the relevant market consisting of large scale main frame computers. Levin, Tr. 750-51. According to Prof. Levin, the parts and services required for an upgrade, and the upgrade that they together produce, can under certain circumstances be a substitute for a new machine; and “the prices of upgrades are constrained to some extent by the ... prices of computers themselves.” Id. If upgrades were excluded from the relevant market definition, AMI would have to be excluded from the market, since AMI does not sell large scale mainframes. See Levin, Tr. 792. At trial, Prof. Levin acknowledged that the International Data Corporation report on which AMI relied to determine what computers are large scale main frame computers, see Levin, Tr. 783, omits “scores, indeed hundreds of computers available in the marketplace today which have as much or more MIP capacity” as some IBM computers which appear in the survey. Levin, Tr. 786-87; 789. AMI’s reliance on IBM documents (PX 15; PX 48) also raises doubts concerning the reliability of AMI’s market share data, since it is unclear that the documents were intended fully to reflect IBM’s competition from other manufacturers of large scale mainframe computers. To the extent that AMI’s market share evidence reflects the IDC’s classification scheme, such evidence tends to overstate IBM’s share of the relevant market because it fails to include non-IBM revenues from products which may fairly be described as large scale mainframe computers. Even if AMI’s market share evidence properly accounted for all sales of new large scale mainframes, a relevant market consisting solely of large scale main frame computers (and upgrades) is unduly narrow for purposes of assessing IBM’s economic power and tends to overstate it. Substitutes exist for every product and a relevant market definition “cannot meaningfully encompass that infinite range;” therefore, “[t]he circle must be drawn narrowly to exclude any other product to which, within reasonable variations in price, only a limited number of buyers will turn.” Times Picayune, 345 U.S. at 612 n. 31, 73 S.Ct. at 882 n. 31 (including general and classified newspaper advertising and excluding advertising placed in other communications media in New Orleans). However, it is also improper “to require that products be fungible to be considered in the relevant market.” United States v. E.I. duPont deNemours & Co., 351 U.S. 377, 394, 76 S.Ct. 994, 1006, 100 L.Ed. 1264 (1956) (including cellophane and all flexible wrapping materials in same relevant market in Section 2 case). The relevant market definition must include all “commodities reasonably interchangeable by consumers for the same purposes....” Id. at 395, 76 S.Ct. at 1007. It must include all producers which “have the ability — actual or potential —to take significant amounts of business away from each other.” Smithkline Corp. v. Eli Lilly & Co., 575 F.2d 1056, 1063 (3d Cir.), cert. denied, 439 U.S. 838, 99 S.Ct. 123, 58 L.Ed.2d 134 (1978). Prof. Levin substantially agreed with these statements of the considerations pertinent to market definition, see Levin, Tr. 171, and acknowledged that products may be in the same relevant market when they are alternatives only “under certain circumstances”, or when the pricing of one product constrains the pricing of another only “to some extent.” Id. at 750-51. To support its relevant market definition, AMI relies heavily on IBM documents which AMI contends demonstrate that IBM recognizes a separate market for large scale main frame computers. See A.M. at 76-77. IBM “Significant Win/Loss Reports” (SWLRs) were monthly reports prepared by IBM’s marketing divisions and furnished to IBM’s Chairman and members of IBM’s Management Committee. Armstrong, Tr. 909-910; Levin, Tr. 756, 800. The SWLRs stated, for each reported competitive situation, the IBM product offering, the competitive product offering(s) faced by IBM, and whether IBM won or lost the sale. Levin, Tr. 756. Prof. Levin examined a sample of SWLRs covering the periods December 1980 through September 1981, April and June 1983, December 1984, January 1985, February 1985, May 1985, and June 1985. Id. at 759. For each reported situation in which the IBM product offering was a large scale mainframe (a 3033 product, a 308X product, or MES upgrade of such a product), Prof. Levin classified the reported competitive alternative as a large scale or mid-scale computer, based on the International Data Corporation’s “Large Scale Computer Census” classification scheme. Id at 757-59. In 281 out of 288 (or 97.6%) reported cases in which the IBM offering was a large scale mainframe, the competitive alternative was also a large scale mainframe or an upgrade to a large scale mainframe. Id; PX 480. The competitive alternatives included leasing company placements of new or used large scale mainframes, and leasing company offerings of upgrades or reconfigurations of large scale mainframes. The usefulness of IBM’s SWLRs and Prof. Levin’s analysis of them to support AMI’s relevant market definition is limited by several considerations. IBM regarded SWLRs as poor and unrepresentative indicators of actual market activity, and discontinued their use. See Armstrong, Tr. 910-13; see also Quinlan, Tr. 1050-53, 1057-59; DX 2201. Nevertheless, assuming that SWLRs accurately reflect actual market activity, they show a large number of situations in which non-IBM large scale main frame computers (included by AMI in its relevant market definition) competed with IBM computer offerings which AMI does not include in its market definition. AMI’s reliance on only those instances in which the large scale mainframe offering was an IBM computer, rather than a non-IBM computer, is inconsistent with the relevant market definition it advances. Finally, it is worth remembering that “it is not the perceptions of manufacturers but those of consumers which are most salient in the determination of market boundaries.” Columbia Metal Culvert Co. v. Kaiser Aluminum & Chemical Corp., 579 F.2d 20, 30 (3d Cir.) cert. denied, 439 U.S. 876, 99 S.Ct. 214, 58 L.Ed.2d 190 (1978) (footnote omitted) (Section 2). The evidence compels the conclusion that AMI’s asserted market definition is unten-ably narrow. AMI improperly excludes from its market definition a host of products which provide reasonable alternatives to purchases of large scale mainframe computers and upgrades and constrain IBM’s ability to raise the prices of these products or exclude competition. The result is that IBM’s market power is significantly less than AMI contends, even before evidence relevant to assessing IBM’s economic power other than demand substitutability evidence is considered. a. AMI’s proposed market definition fails to account for the competition IBM faces from companies offering large scale main frame computers for lease. AMI acknowledges the presence of such competition, see A.P.F. at ¶ 21 and the evidence cited therein, but fails to incorporate it into the relevant market definition. An economically meaningful market definition must include all the products and services that actually or potentially compete with IBM products in question or constrain IBM’s pricing and product decisions. Levin, Tr. 171. Leasing companies, such as Comdisco and CMI, purchase computer equipment from manufacturers and lease it to users. From a consumer’s standpoint, they are an alternative source of computer equipment. They compete with IBM. See Lewis, Tr. 452-3; LaRocca, Tr. 1209; Phillips, Tr. 1586; Pontikes, Dep. Tr. 16; Walker, Dep. Tr. 19-20. Leasing companies own approximately 40 percent of all large scale mainframe computers, as defined by AMI. Lev-in, Tr. 764-65, 858-9. Prof. Levin testified that IBM’s share of the market would be reduced by an amount he was unable to determine if leasing companies were taken into account in AMI’s market definition. See Levin, Tr. 837-38, 764, 804. If leasing company transactions involving computers comparable and in many cases identical to the large scale mainframes marketed by IBM are included in the relevant market, and the market is measured on a “transaction basis”, IBM’s share of the market, according to Prof. Almarin Phillips, who testified for IBM as an expert economist, drops to 34.4 percent. Phillips, Tr. 1600-11; DX 1923. Prof. Phillips testified that such a share would not reflect “overwhelming” activity in the market on IBM’s part. Id. at 1611. AMI relies on United States v. Aluminum Company of America, 148 F.2d 416 (2d Cir.1945) to support its exclusion of leasing companies from the relevant market. See A.R.M. at 79-86. In Alcoa, upon consideration of the defendant company’s market position over a period of years and based on the conclusion that the production of “secondary” ingot was as much within Alcoa’s control as the production of the “virgin” ingot from which it was derived, the Court decided to disregard “secondary” competition in its calculation of Alcoa’s market share. 148 F.2d at 425. Unlike the situation in Alcoa, there is evidence in this case that leasing companies, which also lease non-IBM equipment, effectively compete with IBM and constrain IBM’s ability to set prices or exclude competition in the market for new large scale main frame computers. See e.g., A.P.F. 1129 and the evidence cited therein. Leasing companies, therefore, properly belong in the relevant market applicable to this litigation. See ILC Peripherals Leasing Corp. v. IBM Corp., 458 F.Supp. 423, 428 (N.D.Cal.1978), aff'd sub nom. Memorex Corp. v. IBM Corp., 636 F.2d 1188 (9th Cir.1980), cert. denied, 452 U.S. 972, 101 S.Ct. 3126, 69 L.Ed.2d 983 (1981) (“Memorex II”). b. As shown below, AMI also improperly excluded from its relevant market definition other products and services which offer potential buyers reasonable and viable alternatives to the purchase (or leasing) of large scale mainframe computers or upgrades to such computers. Their availability in the marketplace constrains IBM’s power to command noncompetitively high prices and restricts IBM’s ability to impose unwanted or burdensome terms on an appreciable number of buyers interested in acquiring computer systems or enhancing the performance of existing computer systems. These alternatives also pressure IBM to keep up with the market's growing technological demands. See e.g., Phillips, Tr. 1585. Thus, AMI incorrectly excluded many smaller capacity computers manufactured both by companies which do and companies which do not also manufacture large scale mainframe computers as defined by AMI. See Levin, Tr. 796-800. AMI excludes these products from the relevant market in part because the IDC report on which Prof. Levin relied did not classify them as large scale mainframes, and in part because AMI contends that there is only limited evidence of substitutability between these smaller computers and large scale mainframes. Id., at 759-60, 782. For purposes of this action, the reliability of the IDC report is suspect. See supra, p. 271. The evidence shows sufficient competition between the smaller computers and large scale mainframes to require their inclusion in the relevant market. At trial, marketing executives and users of complex computer systems acknowledged a trend toward “distributed data processing”, in which smaller capacity computers are used in conjunction with or in place of large scale mainframes. Mr. Quinlan, President of IBM’s North Central Marketing Division, discussed the increasing phenomenon, perhaps beginning in the 1970’s, of “hierarchal systems” and “distributed systems” involving combinations of large computers, smaller computers, and terminals. Quinlan, Tr. 1043-44. Such systems provide alternatives to the acquisition of large scale main frames, as the situation involving J.C. Penney illustrates. Id. He also referred to instances in which the smaller computers (which, in fact, tend to be quite large) are “clustered” to look like and perform functions comparable to those performed by IBM’s large scale mainframes. He referred to numerous instances (e.g., involving DuPont Research and Carrier Corporation) in which the smaller computers (excluded from AMDs market) competed successfully against IBM large scale mainframes. Id. at 1050-53,1058-59; see DX 2201. Mr. Armstrong, IBM’s Senior Vice President and Director General of IBM Europe, gave several examples of such competition, including the New York Stock Exchange’s replacement of its IBM System 360 Model 50 computer (included in AMI’s market) with over 200 Tandem “minicomputers” (excluded from AMI’s market). Armstrong, Tr. 897-98. An executive of Core States Financial Corporation testified that his company moved away from using IBM large scale mainframes for at least one application, and opted for a system of multiple computers manufactured by Tam-den and not included in AMI’s market. Shah, Tr. 1516-20. A Hewlett Packard executive stated that his company’s HP 3000 product line of computers (excluded from AMI’s market) competes with IBM 308X computers in part because of a trend toward decentralized processing. Ikezoye, Dep. Tr. 4-6. An NCR official expressed a similar view. Nielson, Dep. Tr. 4-5; see English, Tr. 1557, 1565 (illustration of trend toward decentralized processing, involving Federal Mogul Corporation); DX 1786. A study entitled “Long Range Computing Strategy Report” prepared for Kent State University by Peat, Marwick, Mitchell and Company in 1984 discussed the “trend away from older, large single processing systems ... towards systems based on large number of small processors with local intelligence, linked by communications.” DX 360, pp. 4-5, 49-52. Kent State’s Director of Informations Systems testified that Kent State ultimately replaced a Burroughs model 6812 computer (included in AMI’s market) with an IBM 3081D, but not without seriously considering purchasing a system of clustered DEC VAX 11/785 computers (excluded from AMI’s market). McKinley, Tr. 1487-90, 1494-90. There were many instances in which systems of smaller computers provided alternatives to large scale mainframe computers; the instances involved wide-ranging computer applications in different commercial settings. Such competition cannot be considered marginal. IBM considered the competition significant. In 1975, it reviewed one hundred of IBM’s largest accounts and concluded that a significant number of customers, “almost two-thirds, were concluding to off-load their systems, their large systems to competitive distributed processing systems,” Armstrong, Tr. 887-94; DX 1632, none of which are included in AMI’s market. Prof. Levin’s assertion that such competition is limited and therefore excludable from the market is unsupported by the record. The same is true with respect to fault tolerant processing, in which smaller capacity computers are used essentially in parallel to ensure continuous processing capability for applications frequently performed by large scale mainframes. AMI excludes fault tolerant computers from its relevant market on the theory that they occupy a “specialized market nitch.” See A.P.F. at 11269; Levin, Tr. 802. The weight of the evidence, however, was to the contrary and compels their inclusion in the market. See Shah, Tr. 1517-21, 1524, 1530; Quinlan, Tr. 1055-56, 1059-63. c. AMI’s relevant market also improperly fails to include peripheral products such as storage devices, input/output devices, disk drives or disk tapes. Storage devices store data for later access by the central processing unit (CPU). Input devices feed data to the CPU; output devices receive or accept data from the CPU. Disk drives and desk tapes may perform all three functions. AMI acknowledges that “[e]xternal data storage devices are in a limited sense technological substitutes for the internal memory of a large scale mainframe.” A.P.F. at 11278. AMI excludes storage devices from the relevant market in part because it contends that they are not economic substitutes in the sense that their prices constrain the pricing power of a hypothetical monopolist of large scale mainframes. Levin, Tr. 761-62. AMI takes substantially the same position with respect to computer software, the programs that enable computer systems to function and perform a variety of tasks, A.P.F. at ¶ 279, which AMI excludes from the market. AMI’s assertions are inconsistent with the evidence. Mr. Beckwith, IBM’s Director of Service, testified that IBM always considers the peripheral products offered by competitors when IBM arrives at a “competitive price” for its computers. Beckwith, Tr. 1279-83. Input/output devices particularly are among the various “components” IBM examines to determine the price of its products. Id. Competitive software offerings also influenced IBM pricing decisions, Beckwith, Tr. 1279-83, and at least one computer user testified concerning the “trade offs” between software and hardware. McKinley, Tr. 1493. There is a more fundamental reason for including peripheral products and software (systems software and application software) in the relevant market than the fact that they constrain IBM’s pricing decisions with respect to large scale mainframes. As discussed below, peripheral products and software provided significant and reasonable alternatives to a wide variety of upgrades and modifications of large scale mainframes. See, e.g., Staire, Dep. Tr. 19-20; Armstrong, Tr. 876-78. Other such alternatives which AMI’s market definition also failed to take into account include smaller computers used to offload applications from large scale mainframes (including computers smaller than those previously discussed); devices for adding memory, speed, or capacity to processors; and computer service bureaus, which are companies providing data processing services to others. Further, AMI’s market definition, to the extent that it includes upgrades at all, appears limited to MIPS upgrades. A.P.F. at 11291. Such a limitation also is unwarranted and inconsistent with the evidence. According to Charles I. Staire, Consultant to B.F. Goodrich, there are “many different ways,” to enhance the performance capabilities of a computer system. Staire, Dep. Tr. 3-4, 19-20. Mr. Armstrong testified that a buyer interested in increasing the capacity of a computer system will consider “all of [the] elements that make up the computer systems in order to determine what is the most efficient way, most economical way to solve his problem.” Armstrong, Tr. 876-78. Among the possibilities are increases in the computer’s memory capacity and channel speed, and improvements in the computer software or peripheral devices like disk drives or disk tapes. Id. Mr. McKinley of Kent State added disk storage and main memory components as an alternative to upgrading the CPU of a 3081D computer. McKinley, Tr. 1500-01. Bipin Shah of Core States Financial Corporation postponed an upgrade or the purchase of a larger computer by improving “inefficiencies in the software,” in communications, and in disk drive utilization. Shah, Tr. 1523. Thomas English of Federal Mogul testified that his company has “upgraded the CPUs from one model to the next, but usually before [the company] get[s] to that it’s a memory upgrade, it is adding channels, it is adding other I/O [input/output] devices, disk, tape, so forth.” English, Tr. 1565-66. Gary Smith of CMI testified that alternatives to upgrading a computer, in the sense of adding a MIPS upgrade, include replacing the computer, adding a second computer, adding additional memory, adding additional I/O devices, and off-loading work to other computer systems including those operated by service bureaus. Smith, Tr. 605-06. Robert Van Hellemont of Thomson McKinnon Securities stated that, “[r]ather than upgrading,” a user can “add increased memory,” add “increased discs or tape storage,” offload “application to service [bureaus]”, or offload “applications onto some number of other smaller computer systems.” Van Hellemont, Dep. Tr. 7, 76-79. Numerous other witnesses testified similarly. Harry Myland, one of the founders of AMI, acknowledged that a user “has a variety of options” to upgrading a CPU, including “memory upgrades or feature designs,” “channel adds,” and “internal speed improvements.” Myland, Dep. Tr. 62, 187. Such components, enhancements, and available services permit users at least to postpone buying upgrades of CPUs or larger (new or used) computers for a significant period of time. Phillips, Tr. 1587. A market definition which ignores such competition is unrealistic. d. A box swap, which typically refers to the purchase of a new machine and the sale of an old one, is the “functional” and physical equivalent of an upgrade. See Levin, Tr. 170-72, 766. Box swaps can also be performed with used machines; for example, with machines owned by leasing companies. Allen, Dep. Tr. 59-60; Myland, Dep. Tr. 189-90; Van Hellemont, Dep. Tr. 76-77. Users seek box swaps or upgrades at roughly the same time in the life cycles of computers actually in place. See Lewis, Tr. 1718-19, 1727-28. Although AMI has shown that a box swap is not an alternative to a 3081K to 8084Q upgrade, box swaps are viable alternatives to most MIPs upgrades and should therefore be included in the relevant market definition. e. Upgrades using parts provided by the customer, or by a leasing company from a machine of one of its lessees, also provide computer users with important alternatives to new upgrades or new machines. See Levin, Tr. 744-45. At his deposition, Mr. Allen acknowledged that the existence of 308X donor parts impacted on a user’s decision to purchase new upgrades from IBM. Allen, Dep. Tr. 964. As with box swaps, upgrades using parts taken from other machines may yield physically and functionally identical results to the purchases of new upgrades. Levin, Tr. 171, 182-83. The viability of used-parts upgrades as an alternative to upgrades performed with new parts may be limited, in practical terms, by the relative scarcity of necessary used parts, especially those needed for MIPs upgrades. Levin, Tr. 110-12; Allen, Tr. 339-40. However, the evidence does not support their total exclusion. See, e.g., Levin, Tr. 182-83. For example, between 1983 and 1986, AMI performed 237 (out of 244) 308X memory upgrades or downgrades, and 137 (out of 138) 308X channel upgrades or downgrades, with parts removed from and reinstalled on existing equipment. See DX 1363, pp. 8792-94, 9276-78; 8977; DX 1894D; DX 1894 B; DX 1894C; DX 1363, pp. 8543-44; DX 1895C. CMI also goes into the “user community” to purchase “memory, features, machines, and upgrades”. Loria, Dep. Tr. 228-29; see also Nolin, Dep. Tr. 85-86 (AMI); LaRocca, Tr. 1165-67 (UCS). f. IBM’s market share declines steeply from the figure suggested by AMI if the relevant market definition is expanded, as the evidence requires, to .include (at least to some extent) leasing companies, smaller capacity computers, peripheral products (storage devices, input-output devices, etc.), memory upgrades, channel upgrades, software, and service bureaus. Prof. Phillips, IBM’s market expert, while not attempting to measure IBM’s share of a strictly defined market, see Phillips, Tr. 1594-95, established that IBM accounted for: 20.3 percent of the “value of shipments of [electronic data processing] products worldwide from U.S. plants plus software and service revenues”; 30.8 percent of 1984-1985 “shipments of general purpose digital computers with the unit value” exceeding $250,000 (recorded at FOB platform; the actual installed retail price of such computers exceeded $1 million); and 32.2 percent of the “data processing revenues from the top one hundred firms who [had] data processing revenue” for 1984 and 1985. Phillips, Tr. 1596-1600; DX 1749, DX 1751; DX 1753. AMI did not attempt to measure IBM’s share of any market other than the untenably narrow relevant market it advanced at trial. Although market share is but one measure of a firm’s economic power and should be considered in light of other market factors, it seems clear that market shares in the range suggested by Prof. Phillips do not support a finding that a company possesses sufficient economic power to apply the per se rule against tying arrangements. E.g., Hyde, 466 U.S. at 26-27 & n. 43, 104 S.Ct. at 1565-66 & n. 43 (30 percent share in market for tying product not enough to constitute market power); Times Picayune, 345 U.S. at 611-13, 73 S.Ct. at 881-83 (30 to 40 percent not enough); Ball Memorial Hosp., Inc. v. Mutual Hospital Ins. Inc., 784 F.2d 1325, 1330-31, 1334-37 (7th Cir.1986) (over 50 percent not enough). 2. The analysis so far has focused on what may be described as demand substitutability or cross-elasticity. A thorough analysis of a firm’s economic power also takes into account such considerations as: supply substitutability; other aspects of market structure, general market performance; and the market performance of the defendant company. An examination of important market and industry characteristics further supports the conclusion that IBM lacks the requisite economic power. The absence of significant entry barriers in a given market or industry suggests that firms operating in the market or industry lack market power, even when they enjoy appreciable market shares. See, e.g. United States v. Waste Management, Inc., 743 F.2d 976, 983-84 (2d Cir.1984); Ball Memorial Hosp., 784 F.2d at 1335. Conversely, “[t]he greater the barriers faced by a new entrant, the more probable it is that control of a particular market share would enable defendant to exercise monopoly power.” In re IBM Peripheral, 481 F.Supp. at 976 (footnote omitted). The evidence established substantial entry and growth of competition in the data processing industry embraced by the correctly defined relevant market. A number of companies other than IBM manufacture a wide range of computers having the processing power of IBM large scale mainframe computers. Several of these (including Digital Equipment Corporation, Data General, Hewlett Packard, Tandem, and NCR) were admittedly excluded from the report upon which Prof. Levin relied to determine what constitute large scale mainframes. See Levin, Tr. 786-87, 789; 796-800. AMI has not offered evidence of the market shares any of these individual competitors enjoy. The number of computer leasing companies competing with IBM in the market for large scale mainframes has grown during recent years. Kenneth Pontikes, a founder of Comdisco, one of the world’s largest leasing companies, Robertson, Tr. 707, testified that there were over 500 leasing companies in operation in 1986, Pontikes, Dep. Tr. 17, up from only about a “handful” in 1970. See DX 2168, p. 7. The comparable growth of the Computer Dealers and Lessors Association supports Mr. Pontikes’ assertion. See DX 2168, p. 7; LaRocca, Tr. 1207. Comdisco enjoyed considerable growth between fiscal years 1982 and 1986, when it reported a net profit of $79 million on revenues totaling $902 million (Comdisco reported net profits of $29 million on revenues of $415 million for fiscal year 1982). Pontikes, Dep. Tr. 6-7. CMI, another major leasing company, reported revenues of $140 million for 1982, and $284.7 million for 1985. DX 1073, p., 4; DX 2168, p. 21; Smith, Dep. Tr. 187-88; DX 1071 p. 6. The number of companies, like AMI, engaged in the business of converting and modifying computer equipment also has grown in recent years. AMI itself illustrates the relative ease of entry into the field. It was formed in 1973 with an initial investment of $1,500. Allen, Dep. Tr. 216-18; Ade, Dep. Tr. 92; Myland, Dep. Tr. 4, 63. AMI reported earnings of $15 million for the period 1973 to 1981, and over $53 million in the time period covered by the complaint. See DX 1821A. In 1984, AMI employed approximately 30 “field engineers”, DX 1922, and declined an acquisition offer by CMI for $20-$30 million. Myland, Dep. Tr. 440-46. UCS, which engaged in the conversion and modification of computer equipment, was founded in 1972 by five persons and an initial capitalization of a “couple [of] thousand dollars.” La-Rocca, Tr. 1161-64, 1168-75. In 1985, TRW, Inc. acquired UCS for $7.5 million up front plus a contingency payment estimated in excess of $3 million over a three-year period. Id. at 1177-79. UCS’s business experienced its greatest growth and success during the time period covered by AMI’s complaint. Id.; see also Harnett, Tr. 1750-51. 3. The characteristics of the data processing industry and IBM’s market performance also undermine AMPs contention that IBM possesses the economic power necessary to impose per se liability. Technological progress or innovation in an industry generally reflects the presence of economic competition in the market, and is generally inconsistent with the wielding of undue economic power by a firm or group of firms in the industry. See, e.g., Greyhound Computer Corp. v. IBM Corp., 559 F.2d 488, 497 (9th Cir.1977), cert. denied, 434 U.S. 1040, 98 S.Ct. 782, 54 L.Ed.2d 790 (1978). The data processing industry has enjoyed impressive technological growth in recent years. “[Ejnormous technological change in computer hardware has occurred in recent years ... and customers are now able to obtain substantially increased computer power at less cost than was ever possible before.” DX 360, p. 10; see also, DX 380, p. 21. In a 1986 report, Burroughs, a company which competes with IBM, see A.P.F. at 11283, characterized the “information systems industry” as “highly competitive” and observed that competition in recent years has “intensified as the members of the industry, including many new competitors, have increased the number of products offered.” PX 188, p. 5. Control Data Corporation, which also competes with IBM, see A.P.F. at if 281, reported in 1986 that “[t]he market for general purpose electronic digital computer systems [was] highly competitive,” PX 245, p. 3, and that “[r]apid technological change in the computer industry [had] intensified in recent years as a result of shortened product life cycles and increasing foreign and domestic competition.” Id. at 5. IBM has contributed greatly to the rapid rate of technological innovation in the industry. Prof. Levin acknowledged that “IBM’s success in the data processing industry is attributable in part to its role as one of the leading innovators in [the] economy.” Levin, Tr. 866. IBM’s development of the TCM technology central to this litigation provides a good illustration. IBM spent over a billion dollars and over one thousand man-years of labor to develop the TCM and TCM board. Toole, Tr. 1008. As Prof. William J. Baumol, who testified as an expert economist for IBM, said, IBM’s development of the TCM technology was “forced by market pressures, which were demanding quicker, easier, simpler, more reliable installation, which competitors were approximating at the very same time and which IBM was forced to do.” Baumol, Tr. 1658-59. IBM’s price/performance improvement record also reflects the presence of strong competitive forces in the relevant market. In 1952, IBM announced its 701 computer, which had a price-performance measurement of .02 MIPs per $1,000,000.- In 1981, IBM announced its 3081K, which employed TCM technology and had a price/performance ratio of 3.1 MIPS per $1,000,000 — an increase of over 1500 percent. Granito, Tr. 939-40; DX 1793; see also DX 2021. The 3081K computer represented an increase of more than 120 percent over IBM’s 303U computer, announced four years earlier. See DX 1793. Further, IBM’s computers were priced with sensitivity toward the “competitive environment” in which the company operated. Ross, Tr. 1364, 1390-91 Prof. Levin acknowledged that one measure of market power is a firm’s ability “to move the price [of its products] around without constraint from products outside the market.” Levin, Tr. 781-82. IBM lacked such power. 4. AMI’s contention that IBM possesses sufficient power to justify per se condemnation of IBM’s net pricing policy relies heavily on the asserted “uniqueness” of the TCM technology incorporated into the 308X (and 3090) product line. Even absent a showing of market dominance, the requisite economic power may, under special circumstances, be inferred from the desirability or uniqueness of the tying product. See, e.g., United States v. Loew’s, 371 U.S. 38, 45, 83 S.Ct. 97, 102, 9 L.Ed.2d 11 (1962). IBM obtained a patent for its TCM technology in 1976. PX 598; Granito, Tr. 978-79. TCMs are the “building blocks” or “brains” of the 308X and 3090 computers. Pontikes, Dep. Tr. 29-30; Allen, Dep. Tr. 373-75. They are the critical components of the computer hardware needed to perform 308X upgrades (especially MIPs upgrades); that is, they are the critical components of the alleged tying product. According to AMI, “[t]he uniqueness and sufficiency of economic power of the tying product, parts (TCMs) for net priced 308X MIPS upgrades, are independently established by the existence of patented TCM technology.” A.M. at 40. Contrary to AMI’s assertion, neither the uniqueness of a product nor the fact that it is patented alone establishes the requisite economic power. Hyde, 466 U.S. at 37 n. 7, 104 S.Ct. at 1571 n. 7 (O’Connor, J., concurring) (“A common misconception has been that a patent or copyright, a high market share, or a unique product that competitors are not able to offer suffices to demonstrate market power.”) At least one Court has rejected “any absolute presumption of market power for copyright or patented product_” A.I. Root Co. v. Computer Dynamics, Inc., 806 F.2d 673, 676 (6th Cir.1986). But see Digidyne Corp. v. Data General Corp., 734 F.2d 1336, 1340 n. 4 (9th Cir.1984) (“[W]hen the tying product is patented or copyrighted ... sufficiency of economic power is presumed.”), cert. denied, 473 U.S. 908, 105 S.Ct. 3534, 87 L.Ed.2d 657 (1985). “[A] patent holder has no market power in any relevant sense if there are close substitutes for the patented product.” Hyde, 466 U.S. at 37 n. 7, 104 S.Ct. at 1571 n. 7 (O’Connor, J., concurring); see also Northern Pacific, 356 U.S. at 10 n. 8, 78 S.Ct. at 520 n. 8 (“Of course it is common knowledge that a patent does not always confer a monopoly over a particular commodity.”) Similarly, one Court has held that “factual uniqueness” is insufficient as a matter of law to establish market power. Will, 776 F.2d at 672; accord, Fortner I, 394 U.S. at 505 n. 2, 89 S.Ct. at 1259 n. 2 (“Uniqueness confers economic power only when other competitors are in some way prevented from offering the distinctive product themselves.”); see also Fortner II, 429 U.S. at 620-21, 97 S.Ct. at 867-68 (the kind of uniqueness considered relevant in tying product cases requires proof that the seller possesses some special “advantage not shared by his competitors”). If competitors can design and offer a package similar to defendant’s product for a similar cost, there is no barrier and no market power, and hence no special advantage for purposes of establishing “uniqueness”. Will, 776 F.2d at 673. IBM freely licenses its TCM patent. See Granito, Tr. 987. Further, the great majority of TCM’s are owned and controlled by the leasing companies, which compete with IBM and which own 90 percent of all 308X machines sold by IBM. See DX 1847; PX 25. AMI and other companies engaged in the business of upgrading IBM 308X computers use TCMs owned by the leasing companies to perform conversion and modifications. Companies such as Amdahl, Burroughs, Cray, CDC, and others, which AMI acknowledges compete with IBM in its narrowly defined market for large scale mainframe computers, see, e.g., A.P.F. at ¶¶ 273-64, presumably manufacture computer parts comparable to IBM’s TCMs in terms of price and processing power. Even if no such comparable products exist, AMI has not shown that IBM’s competitors are somehow prevented from developing products to rival the TCM. I conclude that the factual uniqueness of IBM’s TCMs, which admittedly represent a major technological advance and reflect the increasing demand for technological innovation in the data processing industry, does not warrant an inference that IBM possesses sufficient economic power to impose per se liability. The inference of market power is especially unwarranted where IBM’s share of the relevant market is not exceptionally high and other market factors suggest the presence of economic competition in the field. B. “[Wjithin [a] broad market, well-defined submarkets may exist which, in themselves, constitute product markets for antitrust purposes.” Brown Shoe Co. v. United States, 370 U.S. 294, 325, 82 S.Ct. 1502, 1524, 8 L.Ed.2d 510 (1962). AMI also seeks to establish the requisite economic power by reference to a relevant submarket comprised of the parts and services required for the conversion and upgrade of large-scale mainframe computers. A.P.F. at ¶ 285. The boundaries of a relevant sub-market “may be determined by examining such practical indicia as industry or public recognition of the submarket as a separate economic entity, the product’s peculiar characteristics and uses, unique production facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors.” Brown Shoe, 370 U.S. at 325, 82 S.Ct. at 1524. The evidence precludes a finding that parts and services required for the conversion and upgrade of large mainframes constitute a submarket distinct from the market consisting of large scale mainframe computers. There was considerable evidence of competition between, and cross-elasticity of demand for, upgrades and a variety of products, including the large scale mainframes themselves. See supra pp. 271-277. For example, Mr. Chisholm of AMI acknowledged that, “[a]s a general matter, a new box is an alternative to an upgrade....” Chisholm, Dep. Tr. 34. Upgrades and large scale mainframes are reasonable alternatives serving similar data processing needs. The price-sensitivity evidence is especially important. Upgrade prices constrain the prices of new large scale processors; IBM set its net upgrade prices to equal the price differentials of its 308X computers so that customers would not buy one product at the exclusion of the other. See e.g., supra note 15; DX 1861; see also Levin, Tr. 163-64. Finally, IBM upgrades and mainframes are built at the same manufacturing plants and are generally installed by comparably skilled engineers. AMI’s evidence of IBM’s economic power within the alleged submarket also is inadequate. AMI asserts that IBM possesses power in the relevant submarket because IBM performed 2,070 308X MIPs upgrades (of which 1,874 were net-priced), while AMI performed only 16 (of which 11 were equivalent to IBM net-priced upgrades). A.P.F. at 11299; see also PX 465; Levin, Tr. 110-11. Were there proof of the existence of a relevant submarket consisting solely of upgrades and conversions, or especially of a submarket comprised exclusively of upgrades and conversions of IBM computers, then such evidence might support the inference of power AMI seeks to draw. However, AMI has failed to establish the existence of the prerequisite submarket. ^ C. There are separate grounds for concluding that per se condemnation of net pricing is unwarranted. The amount and value of labor involved in performing 308X MIPS upgrades is insignificant compared to the value of the requisite upgrade parts. As a result, AMI has not been foreclosed from a viable business opportunity by IBM’s policy. 1. A major design objective for the TCM and TCM board technology (the TCMs are plugged into the TCM board) was to enable alterations to 380X computers to be performed relatively quickly and with little labor. Granito, Tr. 925-27, 950, 979-81. There is ample evidence that IBM attained the desired result. The TCM technology allowed 308X upgrades to be performed considerably faster, with fewer parts involved, and with less risk of malfunctions than was possible with previous technologies, including the technology embodied in IBM’s 303X product line. See Granito, Tr. 951-57; Brong, Dep. Tr. 80; Broomall, Dep. Tr. 68; Morrow, Dep. Tr. 163-65. For example, the labor required to add 1.1 MIPS to a 3033S processor (to convert it to a 3033N) was up to 20 times greater than the labor required to add 2 MIPS to a 3083E processor (to convert it to a 3083B). See Chisholm, Dep. Tr. 62-68; DX 1799. The resulting and dramatic decline in the amount and value of labor required to perform 308X upgrades has redefined the niche in the computer market occupied by AMI. This decline has eliminated business opportunities once available. By itself, such a result does not offend the antitrust laws. The antitrust laws were enacted for “the protection of competition, not competitors.” Brown Shoe, 370 U.S. at 320, 82 S.Ct. at 1521. The value of the labor content of 308X MIPS upgrades is de minimus for antitrust purposes. IBM’s Director of Finance, Alan Ross, prepared a chart setting forth, among other things, IBM’s calculations of the value of the labor content of all 308X MIPS upgrades performed during the years relevant to the complaint. See DX 1805A. Mr. Ross used 1985 MES prices and the 1985 hourly rate for field installation labor. Based on Mr. Ross’s calculations, the value of the labor content of the upgrades averaged 1.2 percent of the net MES price. See DX 3001. With respect to all but the D to K upgrade, the value of the labor content equalled or was less than 1.2 percent of the net price of the MES. Id. At his deposition, Larry Allen, AMI’s President, stated that if IBM sold the parts required for an E to B upgrade for $417,-600 (i.e., at a price reflecting a reduction of $2,400 from the net MES price of $420,000, to account for the value of IBM labor not provided), it would not present AMI with a viable opportunity to compete with IBM. Allen, Dep. Tr. 938 — 39; see also Allen, Tr. 371. AMI would not be able to compete with IBM if the value of the labor content of the E to B upgrade were .6 percent of the net MES price for the upgrade. See also Allen, Dep. Tr. 968-69 (B to K upgrade). Nor would AMI, according to Mr. Allen, be able to compete with IBM if the value of the labor content of a 308X MIPS upgrade equalled .9 percent of the net upgrade price. Allen, Dep. Tr. 968 (B to J upgrade); Allen, Tr. 372. Contrary to his deposition testimony, Mr. Allen stated at trial that the same parts prices and labor content figures would, in fact, present AMI with a viable business opportunity. See Allen, Tr. 369-372. I regard Mr. Allen’s trial testimony on this subject as hyperbolic. I credit his deposition testimony, and not his trial testimony. At trial, Mr. Allen also testified that AMI would be presented with a viable business opportunity if it were able to buy 308X MIPS upgrade parts from IBM without IBM labor included at prices reflecting a labor content value equal to 1.2 percent of the net MES price. Allen, Tr. 373-74. Mr. Allen’s testimony contradicted his earlier statement that no service organization, including AMI, could survive on a 1.87 percent return on its investment in parts needed to perform upgrades. See DX 2066. I do not credit Mr. Allen’s testimony that AMI could survive on a 1.2 percent rate of return on its investment in upgrade parts. See also infra note 68. There was no other evidence that AMI could survive on such a rate of return. If AMI were to buy the upgrade parts from IBM at IBM’s “MES price without labor” (see DX 1805A, Column 6), it would not, on the whole, be able to perform the upgrades and charge customers what IBM charges for the same upgrades on a net price basis and still recover what AMI would regard as an acceptable profit. Were IBM required to price and sell 308X upgrade parts and labor separately, it would not be required to sell the parts at a price lower than the difference between the MES net prices and Mr. Ross’s figures for the value of the labor content of 308X upgrades, provided that Mr. Ross's calculations allow IBM to recover its direct and indirect labor costs, plus a reasonable profit. IBM 308X upgrade parts sold at such prices would not present AMI with a viable business opportunity. Unless Mr. Ross’s estimates are unduly low, AMI has not been forceclosed from a viable business opportunity as a result of net pricing. AMI argues that “[t]he labor content of net priced MIPs upgrades, as estimated by IBM, is significantly understated,” A.T.M. at 46, and that IBM’s estimated value of the labor content is “unrealistically low.” Id. at 47. Of course, where two products are sold only as a package, it is difficult to ascertain the true cost of either. See Hyde, 466 U.S. at 16, 104 S.Ct. at 1559. a. Mr. Ross’s figures are based on the pricing methodology that is used by IBM to determine its SWRPQ prices for memory and model upgrades. Ross, Tr. 1369, 1381-82; see also Granito, Tr. 984; Beckwith, Tr. 1284-96, 1298, 1358. IBM sells the J to K MIPS upgrade (which does not require the removal of TCMs) optionally on an SWRPQ basis at the price listed on Mr. Ross’s chart. It is IBM’s actual SWRPQ price. IBM used the same method to determine the value of the labor component of the other MESs which IBM sells only on a net priced basis. See Ross, Tr. 1881-82. Thus, although the calculations for upgrades other than the J to K upgrade were made for trial purposes, the methodology employed was not. As Mr. Ross’s chart explains, the labor content of every MIPS upgrade is determined by: 1) taking the average installation hours for each upgrade from IBM’s field service cost estimates; 2) multiplying those hours by the 1985 field installation rate for IBM labor; 3) adding 7 percent for travel time and 10 percent for contingency; and 4) subtracting $100 for handling. See Ross, Tr. 1378-79, 1381-83, 1396-97; DX 1805A; DX 2215. As AMI contends, IBM’s figures underestimate the value of the alleged tied product because Mr. Ross’s estimates account only for field installation labor. AMI’s definition of the tied product includes all engineering labor required to perform the upgrade, including the engineering services performed prior to the installation of the upgrade. The fabrication work not taken into account by Mr. Ross’s estimates includes planning (e.g., determining the computer’s E.C. level) and diagnostic work, and the revision of the (upgraded) computer’s microcode. Levin, Tr. 64-66. Fabrication work may require a different degree of skilled labor than is required by field installation work. See Gibson, Dep. Tr. 333-34; Beckwith, Tr. 1316-17. Although the amount and quality o