In 1977, General Business Systems (GBS) initiated a federal antitrust suit against N.V. Philips Gloeilampenfabrieken (NVP), a Netherlands corporation, and its American subsidiaries, North American Philips Corporation (NAP) and Philips Business Systems, Inc. (PBSI). The defendants cross-claimed against GBS and its sister corporation, Shasta General Systems (Shasta), alleging both federal antitrust and state law causes of action. GBS later sought sanctions against the Philips companies for alleged abuse of discovery. The district court refused to impose such sanctions and granted summary judgment against all parties. Both sides now appeal the denial of their respective complaints. We affirm the district court.

FACTS

This case arises out of the Philips companies’ attempt to penetrate the small business computer market in the United States. Between 1969 and 1978 NVP marketed its computers in this country through PBSI, its marketing agent, MLC, Inc., or independent distributors. During this period, Philips’ share of the small business computer market never exceeded five percent. GBS became the Philips distributor for Northern California in 1972. In return, GBS agreed that it would concentrate its “primary marketing efforts” on Philips computers and would not market “directly competitive hardware.” In 1974 PBSI added St. Louis, Missouri, to GBS’s market area. The agreement was renewed in 1977 to extend through August 1,1980. In 1978, however, Philips exercised a buy-out option, terminating the agency relationship.

Philips’ initial computer lines, the P300 and P350 series, stored memory on magnetic ledger cards (mlcs). Although a number of small business computers used mlcs at the time, mlcs generally were compatible with only one manufacturer’s brand of computers. The mlcs for Philips computers were produced primarily by two German companies, Jollenbeck-Kasten (JK) and Magnetdruck (MD). Philips provided design specifications and production assistance to JK and MD in order to assure maximum card-machine reliability. In return, JK and MD signed supply contracts restricting their sales of Philips mlcs to distributors “recognized as [members of Philips distribution network] pursuant to a written communication given by Philips to the Vendor.” Although it is disputed *970 whether a binding exclusive dealing relationship was created by this language, JK and MD generally sold mlcs exclusively to Philips affiliates or designees. PBSI was the designated Philips distributor for the United States at all times relevant to this suit. 1 PBSI, in turn, would sell the mlcs to GBS and other Philips computer distributors.

In 1974 GBS contacted JK and MD in an attempt to purchase mlcs directly. MD refused because of its exclusive dealing relationship with Philips. JK, however, had experienced negotiating difficulties with PBSI, and in 1975 inquired whether GBS would become an exclusive dealer of the JK-produced Philips mlcs. Although no exclusive dealing agreement was established, GBS did order and receive mlcs from JK during the latter half of 1976.

Philips did not ignore this interference with its distribution network. It repaired its relationship with JK by agreeing to split its orders for mlcs between JK and MD. JK reciprocated by ceasing to deal directly with GBS.

In the mid-1970’s the P300 and P350 faced increasing competition from more sophisticated computers using disk memories and cathode ray tube (CRT) displays. These computers threatened the Philips computers with obsolescence. Both Philips and GBS were concerned with the problem. Philips’ solution was to develop the P330, a disk/CRT unit. GBS, however, feared that the P330 represented too little, too late. GBS’s interest focused on a more viable product, the Diablo disk/CRT computer. In 1976 Diablo’s manufacturers inquired whether GBS would be interested in marketing the Diablo computer. GBS referred the opportunity to PBSI, which was uninterested. On September 3, 1976, GBS agreed to become Diablo’s exclusive U.S. distributor. Pursuant to this agreement, GBS’s two top management officials incorporated a sister corporation, Shasta, which shared office space and staff with GBS. A separate president, however, was designated to run Shasta. GBS assigned its exclusive Diablo dealership to Shasta, remaining liable should Shasta fail to perform.

Meanwhile, Philips continued to develop the P330. However, it was never able to market the new disk/CRT model successfully. GBS argues that Philips never made a production model available to U.S. distributors. Philips claims that GBS’s devotion of time and resources to the Diablo doomed the P330 from the start. In any event, Philips became dissatisfied with its import sales and, in November, 1978, withdrew from the computer import market. Philips sold its residual supply and service functions to Pertec Computer Corporation, which marketed its own small business computers.

II.

PROCEEDINGS BELOW

GBS alleged the following antitrust violations by the Philips companies:

(1) monopolization or attempted monopolization of a market or submarket for Philips-compatible mlcs, in violation of section 2 of the Sherman Act, 15 U.S.C. § 2 ;

(2) conspiracy to fix prices and allocate markets and concerted refusal to deal, in violation of section 1 of the Sherman Act, 15 U.S.C. § 1 ;

(3) product tying, in violation of section 3 of the Clayton Act, 15 U.S.C. § 14 .

GBS also sought to suppress the deposition testimony of certain witnesses because opposing counsel’s misconduct allegedly prevented it from adequately cross-examining those witnesses.

In their counter-claim against GBS and Shasta, the Philips companies alleged the following:

(1) conspiracy to eliminate Philips computers from the domestic computer market, in violation of section 1 of the Sherman Act;

*971 (2) attempt to monopolize, in violation of section 2 of the Sherman Act;

(3) breach of GBS’s distributorship contract with PBSI;

(4) tortious interference by Shasta with GBS’s obligations under the distributorship contract;

(5) tortious interference with the Philips companies’ business relationships with JK, MD, and others;

(6) abuse of process.

The district court, following the recommendation of a magistrate, denied GBS’s abuse of discovery motion. The court rejected GBS’s monopolization claim, primarily for failure to substantiate GBS’s market definition. From this it followed that evidence supporting the charges of attempted monopoly, conspiracy, and product tying was lacking, and the district court so held.

As to the Philips companies’ counterclaims, the district court concluded that GBS and Shasta were a single entity and were thus incapable of conspiring together in violation of the antitrust laws. The court dismissed the attempted monopolization claim because the Philips companies had failed to establish the requisite intent to monopolize or facts from which such intent could be inferred. The breach of contract claim fell in the absence of credible evidence of “direct competition” between the Philips and Diablo computers. Since GBS had not breached its contract with PBSI, the contractual interference claim against Shasta was rejected. The court also found no substantial evidence of improper interference with Philips’ business relationships. Finally, it rejected the abuse of process claim because no causal connection between the alleged antitrust conspiracy and GBS’s filing of this suit was shown, nor was any evidence introduced of improper use of trade secrets acquired through discovery. Summary judgment was accordingly granted against all claims.

III.

STANDARD OF REVIEW

Summary judgment should be granted only when, in light of the evidence and pleadings, a court is convinced that “there is no genuine issue as to any material fact and that the moving party is entitled to a judgment as a matter of law.” Fed.R. Civ.P. 56(c). Summary judgment is disfavored in complex antitrust cases involving elusive issues of intent or motive. Poller v. Columbia Broadcasting System, Inc., 368 U.S. 464, 473 , 82 S.Ct. 486, 491 , 7 L.Ed.2d 458 (1962). Nevertheless, this general reluctance does not preclude the use of summary judgment in antitrust litigation. Continental T. V., Inc. v. GTE Sylvania Inc., 694 F.2d 1132 , 1135 & n. 5 (9th Cir.1982). A party opposing summary judgment must present some “significant probative evidence tending to support the complaint.” First National Bank v. Cities Service Co., 391 U.S. 253, 290 , 88 S.Ct. 1575, 1593 , 20 L.Ed.2d 569 (1968); see, e.g., Betaseed, Inc. v. U & I Inc., 681 F.2d 1203, 1207 (9th Cir.1982); Thomsen v. Western Electric Co., 680 F.2d 1263, 1265 (9th Cir.), cert. denied, - U.S. -, 103 S.Ct. 348 , 74 L.Ed.2d 387 (1982); Ron Tonkin Gran Turismo, Inc. v. Fiat Distributors, Inc., 637 F.2d 1376, 1381 (9th Cir.), cert. denied, 454 U.S. 831 , 102 S.Ct. 128 , 70 L.Ed.2d 109 (1981). Such evidence must be relevant to disputed factual issues that are truly material to the litigation. Id. A lawfully displaced competitor cannot construct a justiciable controversy from artful pleading alone. The courts should not serve as a refuge from the give and take of a properly functioning marketplace. With these principles in mind, we turn first to the district court’s disposal of GBS’s claims.

IV.

GBS’S ANTITRUST COMPLAINT

A. Monopolization

The offense of monopolization rests upon the willful acquisition or maintenance of monopoly power in the relevant market, United States v. Grinnell Corp., 384 U.S. 563, 570-71 , 86 S.Ct. 1698, 1703-04 , 16 L.Ed.2d 778 (1966); Greyhound Computer *972 Corp. v. IBM, 559 F.2d 488, 492 (9th Cir. 1977), cert. denied, 434 U.S. 1040 , 98 S.Ct. 782 , 54 L.Ed.2d 790 (1978), and the existence of actual injury to competition in that market. See Betaseed, 681 F.2d at 1231-32 ; California Computer Products v. IBM, 613 F.2d 727, 735 (9th Cir.1979). The threshold consideration in establishing market power is the relevant market. See Grinnell, 384 U.S. at 571 , 86 S.Ct. at 1704 . A firm cannot impose monopoly prices if buyers are free to purchase a competitor’s goods. Thus, all products that are “reasonably interchangeable,” and so can be said to compete with each other for the same buyers’ dollars, are included in the market definition. United States v. E.I. du Pont & Co., 351 U.S. 377 , 76 S.Ct. 994 , 100 L.Ed. 1264 (1956). For purposes of antitrust analysis, the relevant market may also be a sub-market delineated by:

Brown Shoe Co. v. United States, 370 U.S. 294, 325 , 82 S.Ct. 1502, 1523 , 8 L.Ed.2d 510 (1962); see, e.g., M.A.P. Oil Co. v. Texaco Inc., 691 F.2d 1303, 1307-08 (9th Cir.1982).

Nonetheless, the designation of the relevant market requires considerable judgment. Its dimensions include the product involved, the geographical limits within which it functions, and the appropriate time frame. The latter two pose no difficulty in this case. We can assume that the United States fixes the geographical limits and the years 1969 to 1978 the appropriate time frame.

The choice of the appropriate product is another matter. According to GBS, the appropriate product in this case is the Philips-compatible mlc when sold at wholesale. Its entire case is dependent on this choice of appropriate product. Only in this narrowly defined market did Philips possess a significant share. GBS relies on several factors to establish its market definition. It asserts that owners of Philips computers were “locked-in” to Philips-compatible mlcs. These owners could find few, if any, adequate substitutes for the JK and MD cards. As a result, GBS argues, Philips was able to raise its prices without facing competition from other mlc manufacturers. GBS also claims that the industry recognized a distinct Philips-compatible mlc market. The cards produced by JK and MD met Philips’ stringent specifications and were not sold for use in other manufacturers’ computers. The majority of these cards were handled by dealers who dealt only in Philips products. The foregoing allegedly gave Philips the ability to maintain “premium” prices in this distinct market.

The district court rejected as a matter of law GBS’s market definition. It is this interpretation of the law that we must review. If the district court erred, we must reverse its judgment with respect to GBS’s claims and remand for further proceedings. The primary source of the failure of GBS to establish its position is the district court’s conclusion that the market for Philips-compatible mlcs was not insulated from the competitive struggle between computer systems. That is, in its view Philips had little or no power to raise the price of its mlcs without reducing its profits because any such increase would diminish sales of its computer system and very likely adversely affect aggregate profits. Were mlc prices significantly increased, computer system buyers quickly could shift to other sellers who, in turn, could profitably expand their output to meet the new demand. We agree with the district court that the undisputed facts permit only one conclusion, viz., that the market for mlcs cannot be separated from the general market for small business computer systems.

1. The Undisputed Facts

The performance of mlcs was a major concern for businesses contemplating purchase of a P300 or P350. At first this worked to Philips’ advantage. The reliability of its mlcs, which reputedly worked better than those of its competitors, played an important role in the campaign to sell Phil *973 ips computers. Businesses unwilling to risk processing delays or loss of information would find the Philips computers and their mlcs an attractive package. As other companies developed more modern disk computers, however, Philips’ early advantage became a serious disadvantage. Its failure in the American market is attributable to its attempt to cling to mlc computers even after the newer disk computers had proven their greater productivity. Philips’ sales of computers dropped as purchasers turned away from systems based on the outmoded mlc technology.

In spite of contrary allegations in GBS’s brief, the facts before the district court demonstrate that prospective purchasers also viewed the cost of mlcs as a major barrier to purchase of a Philips computer. While Philips sold fewer than 3000 computers in the American market between 1974 and 1978, it sold more than 19 million mlcs in the same market in the shorter period from 1975 to October 1978. At between twenty and twenty-eight cents a card, the mlcs’ cost was a substantial part of total cost for many users. It is hardly surprising, then, that Shasta included the high price of mlcs among the factors it used to convince customers to buy the more modern, disk-based Diablo. Even Philip Parise, GBS’s vice-president, who maintained that he didn’t know of any particular computer sale lost because of the cost of Philips mlcs, conceded that it could have been easier to sell Philips computers had mlc prices been lower. As we view the record, there is no genuine dispute concerning the fact that the high price of these mlcs, which cost as much as two or three times those of other manufacturers, reduced the competitiveness of the Philips computers.

GBS asserts that the viability of a separate market for mlcs is placed at issue by the price insensitivity of the market for Philips mlcs after Philips withdrew from the computer market. This is not the case. Philips adhered to high prices after its withdrawal because, as GBS’s president, Lawrence Finch, stated in his Declaration in Opposition to Defendant’s Summary Judgment Motion, it had higher overhead costs than its competitors. The period during which GBS alleges that these prices were successfully foisted on the market is the same period during which competitors forced Philips, laden with its expensive mlcs, out of the American small business computer market. In due course Philips abandoned its mlc operation as well. It would not have done so had that market been the separate, profitable market alleged by GBS. Philips was the victim, not the beneficiary, of its high prices in the mlc market.

The other factors cited in support of GBS’s market definition merit little discussion. Mlcs were sold separately from computers because of the dynamics of use, not market differentiation. Users could hardly be expected to buy a supply of mlcs good for the product life of the computer when they made their initial purchase of a P300 or P350. The physical division of consumer behavior does not alone establish that it is directed at separate markets. Industry recognition is, of course, a valid indicator of relevant market. See, e.g., Greyhound Computer, 559 F.2d at 495 . GBS, however, has failed to direct us to any significant probative evidence of such recognition. The only evidence to which it alludes is the unsupported statement of one witness that Philips-compatible mlcs constituted a market. This does not amount to substantial evidence of industry recognition, particularly when contrasted with the persuasive evidence already discussed of the interdependence of Philips computers and mlcs.

2. The Cases

GBS relies on two cases, Heatransfer Corp. v. Volkswagenwerk, A.G., 553 F.2d 964 (5th Cir.1977), cert. denied, 434 U.S. 1087 , 98 S.Ct. 1282 , 55 L.Ed.2d 792 (1978), and Greyhound Computer Corp. v. IBM, 559 F.2d 488 (9th Cir.1977), cert. denied, 434 U.S. 1040 , 98 S.Ct. 782 , 54 L.Ed.2d 790 (1978), to refute the district court’s determination of the relevant market. The confidence of GBS is misplaced. In Heatransfer the jury recognized a separate market in air-conditioning units for Volkswagen auto *974 mobiles. These units, designed to accommodate Volkswagens’ air-cooled, rear-mounted engines, were not suitable for other types of engines. Several manufacturers competed in this special market, which was not quickly responsive to price changes in air conditioners suitable for other engine types. Thus, a jury could easily find a market limited to Volkswagen air conditioners. See 553 F.2d at 980-81 .

The market for mlcs is quite different. Unlike an air conditioner, which is not an integral part of an automobile “system,” mlcs are an essential component of the Philips computer system. Not only is an mlc computer useless without cards, but also the quality of the mlcs determines the overall performance of the system. As already pointed out, the rational purchaser of a computer system must consider the cost, availability, and reliability of mlcs when deciding which system to purchase. Conversely, changes in the price, whether direct or indirect, of competing computer systems will quickly affect demand for the Philips computer systems and their mlcs.

GBS’s reliance on Greyhound Computer is similarly misplaced. In that case the evidence sustained, “though by no great margin,” the jury’s finding of a market consisting of leased computers. 559 F.2d at 494-95 . Contrary to GBS’s contention, we expressly declined to rule on the lock-in theory asserted by Greyhound. Id. at 495-96 . Rather, we placed heavy reliance on industry recognition of the distinction between leases and sales. See id. at 495 .

Standing against GBS’s “lock-in” theory is a line of cases rejecting attempts to establish a relevant market with respect to a “locked-in” component. Telex Corp. v. IBM, 510 F.2d 894 (10th Cir.), cert. dismissed, 423 U.S. 802 , 96 S.Ct. 8 , 46 L.Ed.2d 244 (1975), and ILC Peripherals Leasing Corp. v. IBM, 458 F.Supp. 423 (N.D.Cal. 1978), aff’d sub nom. Memorex Corp. v. IBM, 636 F.2d 1188 (9th Cir.1980), cert. denied, 452 U.S. 972 , 101 S.Ct. 3126 , 69 L.Ed.2d 983 (1981), both rejected claims that the relevant market was IBM-compatible peripheral equipment. In Telex, the Tenth Circuit reversed a district court’s definition of the relevant market for peripherals as being that for devices compatible with IBM central processing units. The circuit court expanded the market definition to include non-IBM plug-compatible peripherals. The ease with which the incompatible could be made compatible established the reasonable substitutability of these peripheral products. 510 F.2d at 919.

We adopted a similar approach in ILC Peripherals, in which the district court’s directed verdict rested in part on the plaintiff’s failure to establish a submarket for IBM peripherals. The district court emphasized the “potential competition” from non-IBM plug-compatible manufacturers who could have converted to production of IBM-compatible equipment. 2 IBM had demonstrated that the costs and engineering barriers to conversion were not excessive. 458 F.Supp. at 429 . It also relied heavily on *975 competition between computer systems. Because peripherals formed a large part of total system cost, an increase in peripheral prices would increase the demand for competitive systems. Systems competition was therefore an important restraint on peripheral pricing. Id.

It is true that in Telex and ILC Peripherals the peripheral equipment represented a more substantial portion of the total system cost than do the mlcs. The same, however, cannot be said about Advance Business Systems & Supply Co. v. SCM, 287 F.Supp. 143 (D.Md.1968), modified on other grounds, 415 F.2d 55 (4th Cir.1969), cert. denied, 397 U.S. 920 , 90 S.Ct. 928 , 25 L.Ed.2d 101 (1970). In Advance Business Systems the plaintiff, a distributor of copying machine supplies, contended that the relevant market was paper that could be used in SCM copiers. SCM controlled this market, although a number of manufacturers produced paper that could be used with SCM’s direct electrostatic process. Copying paper constituted a small portion of overall system cost. The court found the plaintiff’s market definition to be unrealistically narrow after citing the competition among both paper suppliers and copier manufacturers. Id. at 153-54 . It opted instead for a definition that included paper compatible with any copier operating on the direct electrostatic process used by SCM.

Advance Business Systems is consistent with the relevant market cases decided in this circuit. In Kaplan v. Burroughs Corp., 611 F.2d 286 (9th Cir.1979), cert. denied, 447 U.S. 924 , 100 S.Ct. 3016 , 65 L.Ed.2d 1116 (1980), the plaintiff alleged that the initial cost of programming equipment locked-in customers to bulk processing centers using Burroughs equipment and, thus, that such centers composed the relevant market. Similarly, in In re Data General Corp. Antitrust Litigation, 529 F.Supp. 801 (N.D.Cal. 1981), commitment to certain software programs had allegedly locked-in customers to the central processing units that executed the programs. In both cases, definitions of the relevant market that were limited to a single company’s systems were rejected. Such definitions did not describe “economically significant” markets. In Kaplan we stressed the high cross-elasticity of demand as between data processing systems. 611 F.2d at 295 . Data General also turned on the widespread competition in the market for computer services. See 529 F.Supp. at 813-14 .

GBS’s lock-in theory must also be rejected. Philips has amply established that intense systems competition drove it from the market for small business computers and for mlcs. Had there been a demand for substitute mlcs, other manufacturers would have experienced no technological or financial barriers in responding to it. Taken to its logical conclusion, GBS’s approach would dictate that a manufacturer, facing competition against which it cannot prevail in the sale of its end product, could be found to monopolize the market for each unique component that goes into the product. This is surely to lose sight of the forest because of fascination with the trees.

We conclude, therefore, that the district court did not err in rejecting a market definition limited to Philips-compatible mlcs.

B. Attempted Monopolization

Three elements are necessary to prove an attempted monopolization claim in this circuit: “(1) specific intent to control prices or destroy competition in some part of commerce; (2) predatory or anticompetitive conduct directed to accomplishing the unlawful purpose; and (3) a dangerous probability of success.” William Inglis & Sons Baking Co. v. ITT Continental Baking Co., 668 F.2d 1014, 1027 (9th Cir.1981), cert. denied, - U.S. -, 103 S.Ct. 57 , 74 L.Ed.2d 61 (1982). In establishing the presence of these elements we permit a dangerous probability of success to be inferred from proof of specific intent which, in turn, can be inferred from conduct that forms the basis of a substantial claim of restraint of trade, or conduct that is exclusionary or threatening to competition. Id. at 1029 & n. 11.

*976 GBS’s attempt at direct proof of probability of success is unavailing because of its failure to establish a relevant market limited to Philips-compatible mlcs. See generally M.A.P. Oil Co. v. Texaco Inc., 691 F.2d 1303, 1308-10 (9th Cir.1982) (explaining role of relevant market in attempt to monopolize claims). Philips’ diminishing share of the market for small business computers cannot support a finding of probable success in that market. Nor can we infer specific intent because, as the district court found, GBS has failed to show anticompetitive conduct sufficient to survive the grant of summary judgment. It is true, as GBS argues, that direct evidence of specific intent may lessen the burden of the conduct requirement if it clarifies the purpose of otherwise ambiguous conduct. See Inglis, 668 F.2d at 1030 . However, here GBS has made no showing of predatory conduct whatsoever. Philips competed ineffectually in the market for small business computers and their accessory products. To hold that such conduct supports a charge of attempted monopolization is to suggest that the antitrust law punishes losers in the competitive struggle rather than protecting them from the avarice of winners. The district court properly rejected this claim.

C. Price Fixing

GBS’s price fixing claim rests on the complex arrangement Philips created to market its products in the United States. JK and MD manufactured mlcs for the American market under contract with NVP. American suppliers could not deal directly with either manufacturer; NVP designated distributors for the mlcs and renegotiated prices annually with JK and MD. NVP set up a two-tier distribution system for its American market. Mlcs were first funneled through PBSI, its wholly owned subsidiary. In some areas PBSI then sold these mlcs directly to retail customers; in other markets designated agents like GBS distributed the mlcs.

GBS, in support of its price fixing claim, points out that between 1975 and 1978 the prices charged for JK and MD mlcs were nearly identical. Although GBS is ambiguous on this point, it apparently is complaining about the prices JK and MD negotiated with NVP, not the prices charged by PBSI. GBS relies on the rather vague testimony of a GBS officer who fails to differentiate between sales by PBSI and sales by JK and MD. In addition, it relies on a series of letters from NVP to PBSI which detail NVP’s price agreements with JK and MD. GBS argues that negotiations with NVP provided the opportunity for JK and MD to set prices.

We hold that GBS, as an indirect purchaser, lacks standing to complain about prices JK and MD charged PBSI. See Illinois Brick Co. v. Illinois, 431 U.S. 720 , 97 S.Ct. 2061 , 52 L.Ed.2d 707 (1977). The cases cited by GBS, United States v. Socony-Vacuum Oil Co., 310 U.S. 150 , 60 S.Ct. 811 , 84 L.Ed. 1129 (1940), and Ambook Enterprises v. Time Inc., 612 F.2d 604 (2d Cir.1979), cert. dismissed, 448 U.S. 914 , 101 S.Ct. 35 , 65 L.Ed.2d 1179 (1980), do not involve similar indirect purchasing arrangements. Even were GBS to have standing our result very likely would not be different because we find no evidence of any illicit agreement between JK and MD. Price parallelism alone generally will not establish price fixing. Other circumstantial evidence is required. Id. at 615; see, e.g., Zoslaw v. MCA Distributing Corp., 693 F.2d 870, 885 (9th Cir.1982). No such evidence has been shown to exist here. Thus, summary judgment on the price fixing claim was proper.

D. Market Allocation

GBS’s per se attack on Philips’ agreement to divide purchases evenly between JK and MD is also without merit. GBS asserts that this constitutes an illegal horizontal restraint. Numerous cases have condemned such restraints. In United States v. Topco Association, 405 U.S. 596 , 92 S.Ct. 1126 , 31 L.Ed.2d 515 (1972), for instance, potential competitors in the grocery industry combined to allocate markets and compete with larger retailers. Timken Roller Bearing Co. v. United States, 341 U.S. 593 , 71 S.Ct. 971 , 95 L.Ed. 1199 (1951), *977 involved a horizontal restraint among manufacturers of antifriction bearings. Here JK and MD did not combine or agree to allocate markets. The Third Circuit in American Motor Inns, Inc. v. Holiday Inns, Inc., 521 F.2d 1230 (3d Cir.1975), it is true, did treat a vertical allocative scheme as a horizontal agreement where Holiday Inns, in response to pressure from its franchisees, distributed new franchises in a manner that achieved the same ends as would have a horizontal agreement between the franchisees. Pitchford v. PEPI, Inc., 531 F.2d 92 (3d Cir.1975), cert. denied, 426 U.S. 935 , 96 S.Ct. 2649 , 49 L.Ed.2d 387 (1976), also cited by GBS, presented an analogous arrangement between an electronics manufacturer and its dealers. No such joint pressure was exerted here and Philips was responding to none such in dividing purchases between JK and MD. In fact, it was responding to the unilateral pressure of JK which GBS was assisting to some extent.

A direct challenge to PBSI’s contract as an illegal vertical restraint would be no more successful. In Continental T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36 , 97 S.Ct. 2549 , 53 L.Ed.2d 568 (1977), the Supreme Court perhaps left open the possibility that per se standards would apply to vertical restraints where a plaintiff could show “pernicious economic effect or lack of any redeeming virtue.” First Beverages, Inc. v. Royal Crown Cola Co., 612 F.2d 1164 , 1170 (9th Cir.), cert. denied, 447 U.S. 924 , 100 S.Ct. 3016 , 65 L.Ed.2d 1116 (1980). GBS, however, has not attempted to provide the factual evidence necessary to meet its burden even under the rule of reason. See Continental T.V., Inc. v. G.T.E. Sylvania Inc., 694 F.2d 1132, 1136 (9th Cir.1982); see also Chicago Board of Trade v. United States, 246 U.S. 231, 238 , 38 S.Ct. 242, 243 , 62 L.Ed. 683 (1918); cf. First Beverages, 612 F.2d at 1168 n. 3 (reproducing district court’s jury instruction, drawn from Chicago Board of Trade , on reasonableness of territorial restraint). Accordingly, its market allocation claim must fail.

E. Tying

GBS rests its tying claim on an alleged attempt by Philips to expand sales of JK and MD mlcs, the tied product, by denying service and warranty protection, the tying service, to computer owners who did not use these mlcs. The elements necessary to establish such a claim are: “(1) Two distinct products or services are in fact tied such that the products are offered as á single package; (2) The defendant has sufficient economic power in the tying market to impose restrictions in the tied product market; (3) The amount of commerce in the tied product is not insubstantial.” Portland Retail Druggists Association v. Kaiser Foundation Health Plan, 662 F.2d 641, 648 (9th Cir.1981); Moore v. Jas. H. Matthews & Co., 550 F.2d 1207, 1212 (9th Cir.1977). GBS has not presented facts from which the second element could be inferred. To possess substantial power in the tying market, a firm must be able to “raise prices or to require purchasers to accept burdensome terms that could not be enacted in a completely competitive market.” U.S. Steel Corp. v. Fortner Enterprises, 429 U.S. 610, 620 , 97 S.Ct. 861, 867 , 51 L.Ed.2d 80 (1977) (Fortner II). Under the facts before the district court, Philips possessed no such power with respect to the tying service. To have attempted to impose significant pressure to buy JK and MD mlcs by use of the tying service only would have hastened the date on which Philips surrendered to its competitors in the small business computer market.

It is true that Philips mlcs had a higher price than other similar cards. This could be under proper circumstances the result of being tied. See Fortner Enterprises v. U.S. Steel Corp., 394 U.S. 495, 504-05 , 89 S.Ct. 1252, 1259 , 22 L.Ed.2d 495 (1969) (Fortner I); Moore, 550 F.2d at 1216 & n. 8. That, however, is not the case here. The higher prices of Philips mlcs were incident to their greater reliability. Their prices did not depend on Philips’ involvement in the tying service. The cost of Philips-compatible mlcs did not drop after Philips withdrew from the market for small business computers in 1978 and sold its interest in the tied product to a third party. GBS itself sought *978 to sell more expensive mlcs from another manufacturer in this period. The Philips mlc was hardly a product whose prices were inflated by the vigor of its producer’s strength in a tying market. The tying claim also fails.

F. Refusal to Deal

GBS, in pressing its section 1 claims, argues that the Philips companies’ exclusive dealing arrangements with JK and MD constituted violations under either a per se or rule of reason analysis.

To establish a per se violation of section 1, a plaintiff generally must show that the defendant engaged in concerted anticompetitive activity with others at the same level of market organization. Exclusive dealerships usually escape the proscriptions of section 1 because, as already indicated, they establish a vertical relationship among firms that do not compete with each other. See A.H. Cox & Co. v. Star Machinery Co., 653 F.2d 1302, 1306 (9th Cir.1981); Gough v. Rossmoor Corp., 585 F.2d 381, 387 (9th Cir.1978), cert. denied, 440 U.S. 936 , 99 S.Ct. 1280 , 59 L.Ed.2d 494 (1979).

Restraints solicited by a distributor but implemented by a manufacturer are not automatically per se violations. A.H. Cox, 653 F.2d at 1306 . Restraints of this type come within the per se rule only if they “clearly had, or [were] likely to have, a pernicious effect on competition and lacked any redeeming virtue.” Ron Tonkin Gran Turismo, Inc. v. Fiat Distributors, 637 F.2d 1376, 1386-87 (9th Cir.), cert. denied, 454 U.S. 831 , 102 S.Ct. 128 , 70 L.Ed.2d 109 (1981); see, e.g., Betaseed, Inc. v. U & I Inc., 681 F.2d 1203, 1235 (9th Cir.1982).

GBS invokes Ron Tonkin’s per se standard as applied in Cernuto, Inc. v. United Cabinet Corp., 595 F.2d 164 (3d Cir.1979). In Cernuto, the Third Circuit implied an anticompetitive effect to reverse a grant of summary judgment where a distributor allegedly had convinced its supplier, a cabinet manufacturer, to cease dealing with a competing distributor. Id. at 170 . GBS appears to read Cernuto to hold that all a plaintiff need show to establish a per se violation is that a purchaser requested a common supplier to cut off a price-cutting competitor and that the request was honored. The Third Circuit, however, placed equal reliance on evidence of the defendants’ concerted activity to restrain price movement. See id. at 168-70 ; Edward J. Sweeney & Sons, Inc. v. Texaco, Inc., 637 F.2d 105, 115 (3d Cir.1980), cert. denied, 451 U.S. 911 , 101 S.Ct. 1981 , 68 L.Ed.2d 300 (1981). This GBS has not shown.

GBS would use Philips’ allegedly premium prices to prove its price maintenance motivation. We have already rejected GBS's claim that it demonstrated an artificial price premium. According to GBS’s own testimony, the Philips companies had high overhead costs which explain, at least partially, their high prices. Moreover, Philips mlcs were more reliable than the competition’s; this explains the market’s acceptance of higher prices. GBS’s only direct evidence of price motivation is the testimony of a Philips employee that Philips “was aware” of GBS’s lower prices during the period it bought mlcs directly from JK. The employee denied any concern about those prices apart from his general concern about GBS’s interference with the Philips companies’ exclusive relations with JK and MD. Equally damaging to GBS’s Cernuto theory is the absence of evidence that Philips communicated its price concerns to JK or MD. Ron Tonkin and Cernuto each revealed much stronger evidence of complicity between manufacturer and distributor than exists in this case.

GBS claims that even if the Philips companies’ exclusive dealerships were not invalid under Cernuto, the use of pressure against JK and MD to maintain the exclusive distributorship constituted an unreasonable restraint of trade. For support GBS cites a PBSI employee’s testimony that JK was threatening its business relations with Philips by dealing with GBS. The “threat” here, however, is little different than the one in Determined Productions, Inc. v. R. Dakin & Co., 514 F.Supp. 645 (N.D.Cal.1979), aff’d mem., 649 F.2d 866 *979 (9th Cir.1981). In Determined Productions, the defendant contracted with a third party to produce stuffed toy animals, which the defendant, in turn, supplied to the plaintiff. The third party was the largest but by no means the only such manufacturer in Korea. As in this case, the plaintiff tried to sidestep the middle man and deal directly with the Korean toy producer. The defendant responded with a threat to break relations with the Korean toy producer unless it returned to an exclusive dealing relationship. The Korean acquiesced. The district court, having first rejected the plaintiff’s per se claim, discarded the rule of reason challenge because of the plaintiff’s failure to give evidence of a relevant market and because of the ready availability of alternative sources of supply. Id. at 648.

GBS attempts to distinguish Determined Productions on the ground that GBS did not have alternative sources of mlcs. This is not so. Alternatives did exist. It is true that substitute mlcs might not have been available on the same terms from other suppliers; this, however, does not indicate economic injustice. As the court noted in Determined Productions, a defendant, having succeeded in legitimately controlling “the best, most efficient and cheapest source of supply in Korea,” does not have to share “the fruits of its superior acumen and industry.” Id. In trying to secure direct access to mlcs jointly developed by Philips and JK, GBS was attempting to accomplish just that.

Finally, GBS condemns the refusal to deal as a section 2 violation. Our rejection of GBS’s notions of the relevant market foreshadows our response. The relevant market was small business computers. The Philips mlcs were not such unique resources as the railroad terminals in United States v. Terminal Railroad Association, 224 U.S. 383 , 32 S.Ct. 507 , 56 L.Ed. 810 (1912), or the power grid in Otter Tail Power Co. v. United States, 410 U.S. 366 , 93 S.Ct. 1022 , 35 L.Ed.2d 359 (1973). Nor does the evidence support allegations that Philips used its worldwide market position to coerce JK to refuse to deal with GBS. The only passage GBS cites when referring to Philips’ pressure on JK deals explicitly with the United States market. In sum, GBS’s section 2 claim is no more supported than its various claims under section 1. We affirm the district court’s decision.

THE ABUSE OF DISCOVERY CLAIM

GBS moved to have the deposition testimony of two witnesses, Mr. Weniger and Mr. Ockenfels, suppressed unless they were produced for cross-examination along with a third witness, Mr. Sievers, who was never deposed. The district court refused. Our inquiry on appeal is limited to examining the district court’s ruling for an abuse of discretion. See generally Wells Fargo & Co. v. Wells Fargo Express Co., 556 F.2d 406 , 430 n. 24 (9th Cir.1977) (dictum) (decision to grant discovery to determine jurisdiction lies in trial court’s discretion); e.g., Data Disc, Inc. v. Systems Technology Association, 557 F.2d 1280 , 1285 n. 1 (9th Cir. 1977); cf. Liew v. Breen, 640 F.2d 1046, 1049 (9th Cir.1981) (same standard to determine relevancy in discovery).

As to Mr. Weniger, the court found that GBS terminated discovery and therefore could not complain. This finding is clearly supported by the record and did not amount to an abuse of discretion.

GBS’s claim with respect to the Ockenfels’ deposition is stronger. Again, however, the district court’s refusal to suppress did not amount to an abuse of discretion. The magistrate recommended denial of the plaintiff’s suppression request based upon the following facts: (1) GBS had an equal opportunity to cross-examine the witness and filled more transcript space with its deposition than did Philips’ counsel; (2) GBS went beyond the scope of direct examination on cross-examination; and (3) GBS improperly delved into areas that may have involved trade secrets and were certain to cause objections. The district court’s acceptance of his recommendation was not an, abuse of discretion.

*980 VI.

THE COUNTER-CLAIMS

A. Exclusionary Conspiracy

Philips asserts that GBS and Shasta joined co-conspirators Paradata (a trade organization of Philips distributors) and its lawyer, and Xerox Corporation, its subsidiary Diablo Systems, and their officers, to eliminate Philips from the domestic market. Our review of the record reveals no significant probative evidence to support Philips’ charge of conspiracy.

Xerox and Diablo Systems are guilty only of anticipating a shift in demand toward more efficient computers. These companies did not solicit GBS to drop the Philips line or otherwise restrict its existing relationship with Philips. Nor do we find them guilty of any other conspiratorial acts that would violate section 1. The crux of the claim against Paradata and its lawyer is that they sought to prevent Philips from marketing the P330 in Northern California without using GBS, its exclusive distributor in that area. This attempt by Philips dealers to maintain their distributing network indicates a desire to keep Philips in the American market, not to drive it out. They sought to exclude other distributors, not Philips.

The claim against GBS and Shasta is not as easily resolved. The district court’s finding that they operated as a single business entity, and thus could not conspire in violation of the antitrust laws, concerns a frequently litigated issue. A section 1 violation requires, at the outset, concerted activity. Harvey v. Fearless Farris Wholesale, Inc., 589 F.2d 451 (9th Cir. 1979). This requirement precludes liability for coordinated activity among multiple corporations operated as a single entity. Thomsen v. Western Electric Co., 680 F.2d 1263, 1266 (9th Cir.), cert. denied, - U.S. —, 103 S.Ct. 348 , 74 L.Ed.2d 387 (1982); William Inglis & Sons Baking Co. v. ITT Continental Baking Co., 668 F.2d 1014, 1055 (9th Cir.1981), cert. denied, - U.S. -, 103 S.Ct. 57 , 74 L.Ed.2d 61 (1982); Murray v. Toyota Motor Distributors, Inc., 664 F.2d 1377, 1378-79 (9th Cir.), cert. denied, - U.S. -, 102 S.Ct. 2905 , 73 L.Ed.2d 1314 (1982); Las Vegas Sun, Inc. v. Summa Corp., 610 F.2d 614, 617 (9th Cir.1979), cert. denied, 447 U.S. 906 , 100 S.Ct. 2988 , 64 L.Ed.2d 855 (1980); Harvey, 589 F.2d at 455 . The single entity test is easily satisfied when corporate policies are set by one individual or by a parent corporation. See, e.g., Las Vegas Sun, 610 F.2d at 618 ; Harvey, 589 F.2d at 454, 457 . At the other extreme, jointly owned corporations that compete in the marketplace, hold themselves out to the public as competing organizations, and set policy independently are as capable of conspiring to restrain trade as unrelated corporations. See Murphy Tugboat Co. v. Shipowners & Merchants Towboat Co., 467 F.Supp. 841, 859 (N.D.Cal. 1979), aff’d, 658 F.2d 1256 (9th Cir.1981), cert. denied, 455 U.S. 1018 , 102 S.Ct. 1713 , 72 L.Ed.2d 135 (1982).

The relationship between GBS and Shasta does not fall clearly at either of these extremes. Finch and Parise, GBS's principal officers and its sole directors, held 90% of Shasta’s stock and two out of three positions on its board of directors. Shasta and GBS shared offices and personnel. Finch and Parise participated, as directors, in general policy formulation for Shasta. On the other hand, they left much of the day-to-day operation to Shasta’s president, Joseph Walsh. Philips claims that Finch and Parise “swore they had little awareness of Walsh’s management decisions.” The cited portions of the record indicate that neither was much involved with policy decisions during the period in which the alleged conspiracy was to have been formed. Walsh was apparently given general control over Shasta’s sales policies. In sum, the degree of control exercised by GBS is unclear.

On this record, the district court should not have decided the legal status of GBS’s relationship with Shasta. We have found. the single entity issue unsuited for resolution as a matter of law when faced with similar factual uncertainty. See Murray, *981 664 F.2d at 1379 ; Inglis, 668 F.2d at 1055 ; Hunt-Wesson Foods, Inc. v. Ragu Foods, Inc., 627 F.2d 919 , 927 n. 5 (9th Cir.1980), cert. denied, 450 U.S. 921 , 101 S.Ct. 1369 , 67 L.Ed.2d 348 (1981); compare Thomsen, 680 F.2d at 1266-67 . GBS has raised a genuine issue of material fact that cannot be resolved conclusively on the basis of the evidence before us.

We need not reverse the district court’s disposal of Philips’ section 1 claim, however. On so clear a record, we are free to affirm the district court on any ground supported by the evidence before us. Turf Paradise, Inc. v. Arizona Downs, 670 F.2d 813, 821 (9th Cir.), cert. denied, 456 U.S. 1011 , 102 S.Ct. 2308 , 73 L.Ed.2d 1308 (1982); United States v. County of Humboldt, 628 F.2d 549, 551 (9th Cir.1980). Even if GBS and Shasta operated independently and could be guilty of conspiring, Philips has not pointed us to any probative evidence of either specific intent or conduct to eliminate it from the small business computer market.

Philips claims that GBS and Shasta “formulated and undertook a plan to take Philips’ marketing position in the United States and eliminate Philips as a competitor in the small business competitor [sic] market.” The evidence, even when interpreted most favorably to Philips, does not support this claim. Philips lost the American market; it was not taken from it. GBS, far from seeking to eliminate Philips, urged Philips early on to add the Diablo to its product line. Only after Philips declined did GBS set up its marketing arrangement with Shasta. Philips points to GBS’s plan to form “NUCO systems” to market Diablo’s Ranger computer. This plan, however, which GBS claims was never pursued, did not itself drive Philips from the American market. Although the Diablo was clearly more attractive than the old Philips computers, GBS continued to maintain the infrastructure of its Philips distributorship. There is no evidence that either GBS or Shasta ever sabotaged Philips’ sales. GBS sold 67 Philips computers in 1976, the year it began marketing the Diablo, up from 49 the year before. GBS’s later declining sales of Philips computers was part of a nationwide decline from 832 computers in 1976 to 250 in 1978. In this same period the total annual sales of small business computers jumped from 27,000 to 38,000. GBS’s poor sales were the result of an impersonal shift in demand that Philips did not anticipate, or at least did not respond to effectively. Philips cannot retrieve its lost market share by blaming GBS, which was merely the messenger with the bad news.

In view of these conclusions, the district court could properly have granted summary judgment against Philips had it reached beyond the intracorporate defense. We therefore affirm the district court’s disposition of Philips’ section 1 claim.

B. Attempted Monopoly

The Philips companies allege that GBS and Shasta attempted to monopolize Philips’ American trade. 3 We find that this claim easily fails, for we agree with the district court that Philips provides no credible evidence of either specific intent to monopolize or predatory conduct.

C. Breach of Contract and Tortious Interference

The district court rejected the breach of contract claim for want of evidence that GBS, contrary to the terms of the contract, marketed “directly competí *982 tive” equipment. 4 Philips claims that disk/CRT and mlc computers were “directly competitive” from statements by GBS officers that these computers, although vastly different in performance and cost, fulfilled the “same basic tasks” and competed for the same “profile” of customers. Philips attempts to buttress this claim by citing testimony that some of its customers did turn to disk/CRT computers such as the Diablo.

This evidence, while showing that disk/CRT computers replaced some mlc computers, does not show “direct competition” within the meaning of the contract. As Louis Fuchs, Philips’ manager for new products when the Diablo came out, testified, mlc computers were no longer competitive with newer CRT models. The computer industry has experienced the rapid obsolescence characteristic of high technology markets. In such technologically sensitive markets, where every advance brings a major increase in productivity or decrease in price, old products do not remain in “fierce, direct competition” with the new, as Philips claims. The group of customers may remain the same, but their demand shifts from one product group to another as technology redefines their needs and expectations. “Direct competition,” within the meaning of the contract, appropriately should embrace only computers of approximately the same technological generation.

Philips mentions competition between old and new products in the markets for flexible wrappings, United States v. E.I. du Pont & Co., 351 U.S. 377 , 76 S.Ct. 994 , 100 L.Ed. 1264 (1956); energy, United States v. General Dynamics Corp., 415 U.S. 486 , 94 S.Ct. 1186 , 39 L.Ed.2d 530 (1974); fire protection, United States v. Grinnell Corp., 384 U.S. 563 , 86 S.Ct. 1698 , 16 L.Ed.2d 778 (1966); and automobiles, Sherman v. British Leyland Motors, Ltd., 601 F.2d 429 (9th Cir.1979); but in none of these markets have new products vanquished the old as completely as in the functionally sensitive market for small business computers. In this market yesterday’s technology is simply not competitive. Philips’ departure from the American market demonstrates this even with respect to only marginally obsolescent computer products.

Philips relies on certain portions of the record to prove that GBS breached its contractual duty to focus its “primary marketing effort” on the P330 when it became available. Philips claims that its computer was “operational and ready for market” in mid-1977. The cited portions of the record do not support the claim, however. These portions refer only to dates in 1978 and give no detailed information about what “operational” means. Philips does not answer GBS’s claim that workable software and peripheral equipment, which are crucial to successful marketing efforts, were unavailable in 1978. Indeed its evidence boils down to an assertion that a P330 computer demonstrated at a computer show in March, 1978, the same year it abandoned the American market, was a production rather than prototype model.

Be that as it may, GBS did not shirk its duty to Philips. To the contrary, GBS rejected PBSI’s offer to release it from its agency agreement and expressed continued interest in the P330. Philips, on the other hand, withdrew from the American market rather than pressing forward to develop a fully operational P330. Thus, we need not decide whether such a P330 would have been in direct competition with the Diablo.

*983 Philips’ claims of tortious interference with its contracts with GBS and other dealers have no substance. Shasta did not tortiously interfere with Philips’ contracts with GBS; nor did GBS and Shasta interfere with other dealers' contracts. Philips refers us to no breached contracts, no direct evidence of intent on the part of GBS or Shasta to induce a breach, and no damages connected to any contracts they had with Philips’ dealers.

Similarly, inasmuch as customers were offered a new and better computer, no claim of tortious interference with Philips’ customer relations is made out. Shasta sold the Diablo to a wide range of customers to whom Philips no longer offered competitive equipment. Neither the deposition of Shasta salesperson Shirley nor the “NUCO” document shows that the Diablo computer was particularly targeted at Philips’ customers, as Philips asserts. Shasta sought to sell the Diablo to whomever it could.

Finally, with respect to Philips’ relationship with JK and MD, we find no evidence that. GBS, whose direct dealing with JK in 1976 was immediately terminated by PBSI, sought improperly to tie up the mlc market or to use its distribution of mlcs to identify Philips’ customers. In short, Philips has adduced no evidence to support its many claims of tortious interference.

D. Abuse of Process

Philips claims that GBS brought its suit hoping to procure trade secrets and then to manufacture mlcs itself. There is no evidence that GBS was interested in manufacturing mlcs after the period in 1976 when it was able to obtain them from JK. The cited portions of the record, contrary to Philips’ assertions, reveal that GBS later lost interest in manufacturing mlcs. No evidence supports the charge that GBS improperly misused trade secrets, or had an improper motive at the time this suit was filed. Philips’ allegations rest on pure speculation.

VII.

CONCLUSION

There is no support for any of the claims asserted in this case. The judgment of the district court is affirmed.

AFFIRMED.

. The Philips designee until mid-1975 was MLC, Inc., which operated as a subsidiary corporation. PBSI took over as designated distributor in 1975 after complaints about MLC’s effectiveness in handling mlc distribution. MLC is not involved in this litigation.

. Fontana Aviation, Inc. v. Cessna Aircraft Co., 460 F.Supp. 1151 (N.D.Ill.1978), rev’d on other grounds, 617 F.2d 478 (7th Cir.1980), although cited heavily by GBS, is not good law in this circuit to the extent that it disregards the role of supply substitutability in the product market definition, see id. at 1156. Whether real or only potential, reasonable interchangeability of supply is significant because the existence of possible competitors constrains prices even where there is no direct competition. See ILC Peripherals, 458 F.Supp. at 427-29 . Fontana’s identification of a market for Cessna avionics seems internally inconsistent as well. The court noted that customers bought airplanes without paying any attention to avionics. Had the court adhered to its own admonition that real-world consumer conduct should dictate the product market definition, 460 F.Supp. at 1157 n. 20, it would have been forced to conclude that the relevant market was for aircraft. Fontana’s strategy in approaching the avionic market by buying unequipped planes and then selling packaged planes with avionics in competition with Cessna acknowledged the inseparability of the two markets.

GBS also criticizes the district court’s reliance on Bushie v. Stenocord Corp., 460 F.2d 116 (9th Cir. 1972), and Spectrofuge Corp. v. Beckman Instruments, Inc., 575 F.2d 256 (5th Cir. 1978), cert. denied, 440 U.S. 939 , 99 S.Ct. 1289 , 59 L.Ed.2d 499 (1979). We find its criticism marred by its failure to distinguish existing and potential competition, and its unwillingness to follow the court in ILC Peripherals and impute proper significance to the latter.

. Although the counterclaim also appears to state a claim of monopolization, Philips concedes in its reply brief that it raised only an attempt claim. Were Philips to raise a straight monopolization claim, it could be disposed of quickly. Our rejection of GBS’s market definition would be equally fatal to any counterclaim predicated on Philips mies as a relevant market. GBS had no significant power in other markets. It played no significant role in the general mlc market. Charges of monopoly in the small business computer market would be equally insupportable, given GBS’s de minimis portion of the growing and increasingly competitive market already shared, as Philips admits, by nearly seventy firms. There is no evidence of its power in this market, as required by United States v. Grinnell Corp., 384 U.S. 563, 570 , 86 S.Ct. 1698, 1703 , 16 L.Ed.2d 778 (1966).

. The court also noted that the withdrawal of the Philips computer from the domestic market motivated GBS’s defection. Philips correctly urges that the district court erred if it believed Philips withdrew before GBS agreed to market the Diablo. GBS incorporated Shasta in 1976, two years before Philips pulled out of its American computer operations. The district court may have been referring to Philips’ effective withdrawal from the market by its failure to provide an operational CRT computer in time to compete with the Diablo, not to its final departure in 1978. In any event, Philips’ failure to provide substantial evidence that its computers were “directly competitive” with the Diablo makes the district court’s additional reliance on the timing of Philips’ retreat harmless error at worst. Thus, although we agree that the court’s second finding may well be either misleading or incorrect, we need only address the first finding, as it is dispositive.