ULSI System Technology, Inc. appeals from the order of the United States District Court for the District of Oregon granting Intel Corporation's motion for a preliminary injunction enjoining ULSI from infringing U.S. Patent Re. 33,629. Intel Corp. v. ULSI Sys. Technology, Inc., 782 F.Supp. 1467, 21 USPQ2d 1922 (D.Or.1991). Because the district court clearly erred in concluding that Intel had established a reasonable likelihood of success on the issue of infringement, we reverse.

Intel is the assignee of U.S. Patent Re. 33,629 to John F. Palmer, et al., entitled "Numerical Data Processor."1 The claims of the '629 patent are directed to the design and operation of a floating-point arithmetic processor capable of mixed precision calculations, mixed mode arithmetic calculations, and rounding operations. Intel has developed a line of math coprocessors2 covered by the patent, including the Intel 8087, 80287, and 80387 coprocessors.

On January 10, 1983, Intel and the Hewlett-Packard Company (HP) entered into a cross-licensing agreement to "increase their freedom of design by obtaining a license under present and future patents and patent applications owned or controlled by the other." Under that agreement, Intel and HP each granted to the other an "irrevocable, retroactive, nonexclusive, world-wide, royalty-free license" under all patents and patent applications "having an effective filing date prior to January 1, 2000, said license to be effective until the expiration of said patents."

ULSI sells a math coprocessor known as the US83C87 ('C87 coprocessor) which is compatible with the Intel 80386 microprocessor and competes commercially with the Intel 80387 coprocessor. Since September 22, 1989, ULSI has purchased the 'C87 coprocessors from HP under an agreement entered into on August 2, 1988, in which HP agreed to manufacture the coprocessors for ULSI. As is apparently common in such "foundry" arrangements in the semiconductor industry, ULSI supplied HP with proprietary design specifications and HP then manufactured and shipped completed coprocessor chips to ULSI, which resold them as ULSI products.3

Intel first became aware of ULSI's 'C87 coprocessor sales on February 4, 1991. On July 29, 1991, Intel brought an action in the U.S. District Court for the District of Oregon alleging infringement of the '629 patent by ULSI's "making and selling, and inducing others to make, sell and use, the 'US83C87' [coprocessor]."4 Intel filed a motion for a preliminary injunction. In considering Intel's motion, the district court weighed several factors relating to injunctive relief, including the likelihood of Intel's success on the merits, irreparable harm, the balance of hardships, and the public interest.

The district court determined that the public interest favored neither party because the public's interest in the protection of patents was balanced by its interest in allowing an accused company to continue to operate until the issue of liability was fully adjudicated. Additionally, the court determined that the balance of hardships tipped in favor of ULSI because the 'C87 coprocessor was ULSI's only product and ULSI "would in all likelihood be forced out of business" if it were enjoined. However, the district court concluded that Intel had established a likelihood of success on the merits with respect to the issues of infringement, validity, and enforceability of the '629 patent. Furthermore, it determined that ULSI failed to rebut the presumption of irreparable harm that arose from Intel's "clear showing" of validity and infringement. Because the "likelihood of success on the merits and the irreparable harm (both of which favor Intel)" outweighed "the balance of hardships (which favors ULSI)," the district court granted Intel's motion.5

The issuance of a preliminary injunction under 35 U.S.C. § 283 (1988) is a matter of discretion for a district court. That discretion, however, is not absolute and must be reviewed in light of the equitable standards governing the issuance of injunctions. Smith Int'l, Inc. v. Hughes Tool Co., 718 F.2d 1573, 1579, 219 USPQ 686, 691 (Fed.Cir.), cert. denied, 464 U.S. 996, 104 S.Ct. 493, 78 L.Ed.2d 687 (1983). In determining whether a movant has established a right to preliminary injunctive relief, the district court must consider a number of factors, viz., (1) whether the movant has sufficiently established a reasonable likelihood of success on the merits; (2) whether the movant would suffer irreparable harm if the injunction were not granted; (3) whether the balance of hardships tips in the movant's favor; and (4) the impact, if any, of the injunction on the public interest. Hybritech Inc. v. Abbott Lab., 849 F.2d 1446, 1451, 7 USPQ2d 1191, 1195 (Fed.Cir.1988). Each factor must be weighed and assessed against the others and against the form and magnitude of the relief requested. Id. We have cautioned, however, that a preliminary injunction is a drastic and extraordinary remedy that is not to be routinely granted. Nutrition 21 v. United States, 930 F.2d 867, 869, 18 USPQ2d 1347, 1349 (Fed.Cir.1991); Illinois Tool Works, Inc. v. Grip-Pak, Inc., 906 F.2d 679, 683, 15 USPQ2d 1307, 1310 (Fed.Cir.1990).

In opposition to the motion, ULSI maintained that HP was permitted under the licensing agreement to act as a foundry for ULSI and that the sale of the coprocessors by HP to ULSI was a "first sale" that extinguished Intel's patent rights with respect to those products. The district court, however, rejected ULSI's argument because it determined that the licensing agreement did not grant HP the "power to sublicense" the '629 patent. On appeal, ULSI claims that the district court erred in concluding that the "patent exhaustion" or "first sale" doctrine did not shield ULSI from Intel's claim of infringement.6

The law is well settled that an authorized sale of a patented product places that product beyond the reach of the patent. See Bloomer v. Millinger, 68 U.S. (1 Wall.) 340, 350-51, 17 L.Ed. 581 (1864). The patent owner's rights with respect to the product end with its sale, United States v. Univis Lens Co., 316 U.S. 241, 252, 62 S.Ct. 1088, 1094, 86 L.Ed. 1408, 53 USPQ 404, 408 (1942), and a purchaser of such a product may use or resell the product free of the patent, id. at 250, 62 S.Ct. at 1093, 53 USPQ at 408. This longstanding principle applies similarly to a sale of a patented product manufactured by a licensee acting within the scope of its license. See Unidisco, Inc. v. Schattner, 824 F.2d 965, 968, 3 USPQ2d 1439, 1441 (Fed.Cir.1987), cert. denied, 484 U.S. 1042, 108 S.Ct. 774, 98 L.Ed.2d 860 (1988).

In the instant case, the issue as to whether ULSI is free from infringement liability turns on whether there was a sale of 'C87 coprocessors by HP to ULSI. Intel argues that the "patent exhaustion" doctrine does not apply because HP never sold a product to ULSI. Although Intel claims, as it must, that the 'C87 coprocessor infringes the '629 patent, it maintains that what was actually sold by HP under the foundry agreement was its fabrication services with an ancillary sale of wafers and chemicals. Intel asserts that HP could not have sold a product covered by the '629 patent because HP never had or retained any ownership rights in the 'C87 coprocessors. Thus, according to Intel, no sale ever took place that could support ULSI's "first sale" defense. That argument is incorrect.

Interpretation of a contract is a question of law which we review de novo. See Interstate Gen. Gov't Contractors, Inc. v. Stone, 980 F.2d 1433, 1434 (Fed.Cir.1992). After reviewing the HP-ULSI contract, we cannot accept Intel's characterization of that agreement as one in which HP merely provided fabrication services to ULSI. That agreement, entitled "Terms and Conditions of Sale," is replete with references to the sale of semiconductor wafers (i.e., chips) that incorporate the 'C87 coprocessor design. For example, the section of the agreement headed "Section 2: Production Fabrication" provided that HP "will sell CMOS34 wafers to" ULSI. That section recites prices for the chips and includes a delivery schedule for shipments of the chips to ULSI. Although the agreement also includes a section delineating the "engineering services" to be provided by HP, the agreement clearly involved the sale of chips, not merely the sale of fabrication services.

Nor, as Intel contends, must the licensed seller of a patented product own intellectual property rights to the product in order for there to be a sale. Intel makes much of the fact that the 'C87 chip was based on a design provided by ULSI. Intel confuses the issue of design origin with the issue of sale. Who designed the chip and whether it embodies inventions other than Intel's have no bearing on the controlling issue whether the 'C87 coprocessors were sold by HP to ULSI and thus extinguished Intel's patent rights relating to those products.

That ULSI, rather than HP, might have owned any existing intellectual property rights to the chips was a matter between ULSI and HP, and did not concern Intel. Intel does not dispute that HP was authorized under the broad terms of the licensing agreement to sell the chips at issue. To the extent that Intel had a patent covering the chips, HP's conceded right to sell the chips deprives Intel of any claim of infringement, as long as HP sold the chips. If it had not granted that license or if the license had been limited in some relevant way, that would be a different case from the one before us. Intel might thereby have retained its right to proceed against those who entered into foundry agreements such as the present one. While Intel may not in retrospect be pleased with the deal that it made permitting HP to make unrestricted sales, it nevertheless granted HP that right in 1983, presumably for consideration it believed to be of value at that time. It cannot now renege on that grant to avoid its consequences.

We also reject Intel's contention that the sale of chips by HP to ULSI constituted a "de facto sublicense" prohibited by the licensing agreement. We found a similar argument to be "without merit and specious" in Lisle Corp. v. Edwards, 777 F.2d 693, 227 USPQ 894 (Fed.Cir.1985). In Lisle, a licensed manufacturer sold products covered by the licensor's patent to a third party which resold them under its trademark. The licensor brought an infringement action against both the licensee and the third party on the basis that the manufacture of the patented product for the third party constituted a sublicense. Because such sublicensing was prohibited under the licensing agreement, the patent owner claimed that the products were infringing. The court in Lisle, however, concluded that the licensee's sales were authorized and that the resale by the third party did not create a sublicense. Id. at 695, 227 USPQ at 895. Similarly, the sale by HP to ULSI here did not create a sublicense. HP did not empower ULSI to make Intel-patented chips or to use or sell any such chips except those lawfully sold to it by HP; these would have been the incidents of a sublicense.

Relatedly, we do not agree with the district court's conclusion that the sale of chips by HP to ULSI was not a "first sale" because HP was not authorized to sublicense ULSI to design products covered by the '629 patent. That HP did not have the authority to sublicense the '629 patent to ULSI is irrelevant. The agreement between HP and ULSI was not a sublicense, but a contract for the manufacture and sale of chips. Thus, HP did not grant a sublicense; it sold a product, albeit one designed by its purchaser. ULSI is immune from infringement, not because it was a sublicensee, which it was not, but because HP was a licensed and therefore legitimate source of the chips. Moreover, ULSI was not required to be sublicensed in order to provide its chip design to HP.

Both parties cite our earlier decision in Intel Corp. v. United States Int'l Trade Comm'n, 946 F.2d 821, 20 USPQ2d 1161 (Fed.Cir.1991) ("Atmel "), as supporting authority. Atmel is similar to, but distinguishable from, the instant case. Under the agreement in Atmel, Intel granted Sanyo a "non-exclusive, world-wide royalty-free license without the right to sublicense except to Subsidiaries, under Intel Patents which read on any Sanyo [devices] for the lives of such patents, to make, use and sell such products." (Emphases added). Intel alleged that Atmel, among others, violated 19 U.S.C. § 1337 (1988) by importing EPROMs (Erasable Programmable Read Only Memories) that infringed a number of Intel patents. Atmel claimed that its imported EPROMs were not infringing because they were manufactured and sold by Sanyo under its agreement with Intel. Intel, on the other hand, argued that Sanyo was not permitted to sell EPROMs to Atmel for resale as Atmel products because the licensing agreement only authorized Sanyo to sell Sanyo products.

As an initial matter, the court in Atmel expressly recognized the freedom from patent infringement of one purchasing products from a licensed party under a foundry agreement. The court stated that

[i]f the Intel/Sanyo agreement permits Sanyo to act as a foundry for another company for products covered by the Intel patents, the purchaser of those licensed products from Sanyo would be free to use and/or resell the products. Such further use and sale is beyond the reach of the patent statutes. See United States v. Univis Lens Co., 316 U.S. 241, 250-52, 62 S.Ct. 1088, 1093-94, 86 L.Ed. 1408, [53 USPQ 404, 408] (1942) (the first vending of any article manufactured under a patent puts the article beyond the reach of the patent).

Id., 946 F.2d at 826, 20 USPQ2d at 1166. Thus, the court agreed that Atmel would be shielded from Intel's claims of infringement if Atmel could establish that Sanyo was authorized to sell the EPROMs to Atmel.

In determining whether the licensing agreement provided for foundry rights, the court focused on what was meant by the "Sanyo limitation" in the agreement. The court concluded that the limitation precluded Sanyo from serving as a foundry for non-Sanyo EPROMs because Sanyo was only permitted to sell Sanyo products. Sanyo was prohibited from producing and selling EPROMs to Atmel for resale as Atmel products, and the court thus held that Atmel could not rely on the license defense. Id. at 828, 20 USPQ2d at 1167-68. In contrast, the licensing agreement between Intel and HP here contains no restriction on HP's right to sell or serve as a foundry.

In light of our discussion above, we hold that the 'C87 coprocessors were insulated from Intel's claim of infringement because they were sold to ULSI by HP, which was authorized to do so under its licensing agreement with Intel. Accordingly, we conclude that Intel cannot establish a likelihood of success on the issue of infringement.

To obtain reversal of a grant of a preliminary injunction, an alleged infringer must establish that a determination regarding one or more of the factors relied on by the district court was clearly erroneous. New England Braiding Co. v. A.W. Chesterton Co., 970 F.2d 878, 882, 23 USPQ2d 1622, 1625 (Fed.Cir.1992). Although none of the factors alone is dispositive, the absence of a sufficient showing with regard to any one factor may, in light of the weight assigned to the other factors, preclude preliminary injunctive relief. See Chrysler Motors Corp. v. Auto Body Panels of Ohio, 908 F.2d 951, 953, 15 USPQ2d 1469, 1471 (Fed.Cir.1990).

The district court's finding on the likelihood of success is clearly erroneous because it was based on a legal error concerning the application of the first sale doctrine. As to the other preliminary injunction factors, the district court presumed irreparable harm because it found that Intel had made a clear showing that the '629 patent was valid and infringed. Because that presumption was based on a clearly erroneous finding on the likelihood of success, it too was clearly erroneous. We discern no clear error in the district court's determination that the balance of hardships tipped in ULSI's favor and that the public interest favored neither party. In view of the totality of these factors, as weighed by the district court, we conclude that the district court abused its discretion in granting Intel's motion for a preliminary injunction.

The district court erred in determining that the Intel-HP licensing agreement did not provide ULSI with a defense against Intel's claim of infringement, a predicate to its finding a likelihood of success in proving infringement, and derivatively, irreparable harm. The district court's grant of Intel's motion for a preliminary injunction is therefore reversed.

REVERSED.

PLAGER, Circuit Judge, dissenting.

I respectfully dissent. ULSI has managed to take a shield the law provides to purchasers of products containing patented inventions and turn it into a sword to cut off the legitimate rights of the patent owner. While a wrong result in a particular case is always unfortunate, a wrong precedent in this particular context is a cause for concern, since it can only lead to even further confusion regarding an important issue of law. In hopes that this case will not establish itself as such, I write to explain what went wrong.

In its simplest terms, the case is this. Company A and Company B are major competitors in a highly volatile industry. The industry experiences constant innovation in its products and processes, and is under great pressure from other domestic and foreign competitors. A and B both maintain large R & D operations, and obtain patents on their various innovations. Since both companies are engaged in continuing programs of research and development, they agree that it is in their mutual interest to avoid spending resources litigating with each other over patent rights rather than inventing.

In order to increase their freedom of design they cross-license each other in such a way that each is free to innovate and market their own similar products without fear of infringing upon the patent rights of the other. There is no intent to authorize third parties to make, have made, use, or sell the inventions covered by these patents beyond the rights the law accords to purchasers in the ordinary course in the marketplace.

Company C, a small company seeking to break in to the same market, approaches Company B with a proposition. C will provide B with details of its (C's) invention (a design similar to that patented by A). C will provide complete design and manufacturing specifications, and warrants to B in writing that C rightfully obtained the design involved and that it does not infringe the patent rights of others. Using its manufacturing facilities, B is to manufacture the item to C's specifications. B will provide the raw materials, and will be paid on a per completed unit basis. B, having excess manufacturing capacity, agrees, and produces the contracted-for item. B then delivers the item to C, and C provides the finishing touches.

Later, C markets its product, describing it in terms that suggest it is the same as a product manufactured by A. A examines C's product, concludes that it is so much like A's product that it infringes one of A's patents, and sues C. C then defends on the grounds that, since B manufactured the item that infringes A's patent, and since B is immune from liability for infringement of A's patents under the A-B cross-license, C also is immune under the doctrine of 'first sale,' sometimes known as 'patent exhaustion.'

Of course C is wrong.

The principle of 'first sale,' simply stated, is that when a patent owner (or the owner's authorized licensee) sells to another a product which incorporates the patented invention, the other may convey the product to third parties free of any claim of patent infringement.1 That is, third parties may use or sell the product without a license as such from the patent holder. This is but another variant on the common law rule that an owner of property may be estopped from claiming an interest in property which, through voluntary act of the owner, has found its way from the owner's transferee into the hands of a third party.2

This rule has no application to the case of C for two reasons. First, the 'first sale' of the allegedly infringing product was by C to third parties, not by B to C. Second, the claimed activity of B must be within the scope of authority granted by the license from A in order for B's activity to immunize C; here it is not.

As for the first point, B did not make a sale to C of a product incorporating A's patented invention. It sold C the raw materials that went into the product. It manufactured C's product to C's specifications. It sold its manufacturing expertise, but the product was always C's, never B's (and assuredly not A's).

If B were to have manufactured a thousand extra embodiments of C's design, performed the finishing touches itself, and then sold them to third parties, would B's cross license with A protect it from a claim of conversion by C? The answer of course is "no." The issue of infringement under the cross-license--i.e., the rights of A and B vis-a-vis each other--is distinct from the question of C's legal rights against B or A, and vice versa.

If the issue is whether the chips made by B infringe A's patent, the fact that C rather than B designed them, or that C owns the manufactured item rather than B, would be irrelevant. If the accused product delivered to C infringes A's patent, then A has a cause of action for infringement of its patent right against both B and C.

B's defense is, I am not liable for infringement because I have a license from the patent owner.

C's defense is not the license; his defense is, I am not liable for infringement because I am a purchaser of the patented device in ordinary course in the marketplace. In which case the question is, did B sell the accused device to C, or did B sell C something else, like manufacturing services. The question of whose design and whose property was involved is not only relevant, but determinative.

As for the second point, the scope of B's authorization, neither A nor B intended by their cross-license to immunize third-party infringers of the patents of either. It is fundamental contract law that the A-B cross-license cannot afford rights and immunities to third parties that are not within the contemplation of the contracting parties. The duty that A and B owe to each other with regard to their own inventions and products is determined by the terms of the A-B cross-license; the duty that B and C owe to each other is determined by the B-C manufacturing agreement; the duty that C owes to A is determined by neither of these agreements, but by the law of patent infringement.3

In the case before us, A is Intel, B is Hewlett-Packard (HP), and C is ULSI. By law Intel has been granted the "right to exclude others from making, using, or selling" the invention claimed in the Palmer patent. 35 U.S.C. § 154 (1988). Intel alleges that ULSI "without authority ... sells" a chip which is within the scope of the Palmer patent, and therefore is an infringer. 35 U.S.C. § 271(a) (1988). ULSI is correct that, if the C87 chip had previously entered the stream of commerce with Intel's permission via an authorized sale of the C87 chip by HP to ULSI, then ULSI's sales to others need not be authorized by Intel. Bloomer v. Millinger, 68 U.S. (1 Wall.) 340, 350-51, 17 L.Ed. 581 (1864). But if ULSI's sale to others is the first sale of that chip (i.e. it is ULSI and not HP or Intel who launches the C87 chip into the stream of commerce), or if Intel never authorized HP to manufacture infringing embodiments of the Palmer patent at the behest of others, so that there could be no authorized sale by HP under the license, then ULSI's defense must fail. In my view, the defense fails on both counts.

As an initial matter, if we assume for the sake of argument that HP sold an embodiment of the Intel patented invention to ULSI, we must determine whether that sale was authorized by the patent owner; absent authorization the transferee obtains no protection. General Talking Pictures Corp. v. Western Elec. Co., 304 U.S. 175, 182, 58 S.Ct. 849, 853, 82 L.Ed. 1273 (1938); Mallinckrodt, Inc. v. Medipart, Inc., 976 F.2d 700, 703 (Fed.Cir.1992). The answer to this question is found in the HP-Intel agreement.

The terms of the cross-license between Intel and HP are broadly stated. The cross-license notes that both parties are "engaged in continuing programs of research and development," and that both parties "want to increase their freedom of design by obtaining a license." In light of this expressed intent, HP and Intel then entered into a reciprocal grant--an "irrevocable, retroactive, nonexclusive, world-wide, royalty-free license under all patents and patent applications" filed before January 1, 2000, effective until the expiration of the designated patents.

It is correct that an appellate court approaches contract interpretation as a question of law. We thus owe no special deference to the trial court's view of that law; that is the meaning of review de novo. But that is not the same as saying we are free to ignore the trial court's factual determinations regarding the intent of the parties, based on testimony at trial. Especially is this the case when the matter before us is not a final judgment but a preliminary injunction, a matter expressly within the broad discretion of a trial court.

Both parties to the agreement--Intel and HP--agree that the cross-license was intended only to provide each party with the freedom to conduct research and development work free of wasteful litigation, and that there was no intent to immunize third party infringers. For example, the Associate General Counsel and Director of Intellectual Property for HP testified that "[n]either Intel nor HP intended to grant the other the right to sublicense any patents licensed to the other under the patent cross-license agreement." Further, he testified that "HP did not intend to grant a sublicense under any Intel patents licensed to HP...." Jt.App. at 408.

The trial court, after considering the evidence presented by the parties and following a hearing on the motion, agreed with Intel's interpretation of the contract. The trial court stated:

In examining the breadth of the Intel/Hewlett-Packard agreement, it is clear that neither Intel nor Hewlett-Packard intended their agreement to be so broad as to grant the other party the power to sublicense any patent granted under the Intel/Hewlett-Packard agreement.

782 F.Supp. at 1474, 21 USPQ2d at 1928 (emphasis added).4

The use of the term "sublicense" by counsel and court must be understood to mean that the cross-license did not give HP power to authorize third parties to separately design and manufacture (or have manufactured) products incorporating the patented invention. Thus HP could itself manufacture and sell products with the patented invention incorporated in them (and the purchasers of those products from HP would be protected in their use under the 'first sale' principle), but HP was not licensed to authorize others to do so.

In another context this court described as "without merit and specious" a patentee's argument to the effect that the activity involved was a prohibited sublicense. Lisle Corp. v. Edwards, 777 F.2d 693, 227 USPQ 894 (Fed.Cir.1985). And on those facts, the argument was specious. In Lisle, the patentee had granted the manufacturer (Lisle) a nonexclusive license to make, have made, use, and sell the patented tool. Lisle produced the tool in accordance with the patentee's design and paid the patentee a royalty on all sales. One of Lisle's customers, Snap-On Tools, purchased the patented tools but also arranged for Lisle to affix the tool with the Snap-On trademark. The patentee argued that the arrangement between Lisle and Snap-On was a "de facto sublicense," which made Snap-On a "de facto manufacturer." This court noted that "[t]he sales by Lisle were authorized by the nonexclusive license agreement. Resale did not create a sublicense." Id. at 695, 227 USPQ at 895.

Lisle does not stand for a general rule that an argument that sublicensing is prohibited under a particular license is necessarily "specious." The substance of the transaction at issue should control whether it is 'sublicensing,' and the terms of the license as intended by the parties should determine whether such 'sublicensing' is permitted. In Lisle, the substance of the transaction was the licensee (Lisle) manufacturing a product containing the patented invention, with a subsequent authorized sale of the product to a third party (Snap-On) interested in buying it. Here, in contrast to Lisle, the third party, ULSI, purported to have no interest in buying the patentee's invention--the substance of the transaction was the licensee making the third party's invention for the third party's account. And record testimony indicates unequivocally that neither party to the cross-license understood or intended their cross-license to bring this activity within the scope of the cross-license.

This is not the first time that the Federal Circuit has been called upon to determine the scope of a broad Intel license. In Intel Corp. v. United States Int'l Trade Comm'n, 946 F.2d 821, 20 USPQ2d 1161 (Fed.Cir.1991) (Atmel ), Intel had similarly entered into broad cross-licenses with two foreign manufacturers (both with names including the word Sanyo). The two Sanyo companies entered into an agreement with a third company, Atmel, under which the Sanyo companies manufactured Atmel-designed chips which allegedly infringed Intel patents when offered for sale by Atmel in this country. This court held that the importation of such goods was in violation of the patent rights of Intel, since the Intel license to Sanyo did not permit the licensees to manufacture the designs of third parties and thus immunize the fruits of those designs from charges of infringement.

The court quoted with approval the ITC's description of the incongruity of the result argued for by Atmel:

The interpretation of the licensing agreement as proposed by Atmel would mean that any company that was unable to obtain a license from Intel but still wanted to make its own parts practicing Intel patents could employ Sanyo as a foundry and circumvent Intel's patents. Without something to explain why the parties would have intended such a result, the agreement will not be given this strained construction.

Id. at 827, 20 USPQ2d at 1166-67 (emphasis in original).

The second element to a successful 'first sale' defense is the existence of a sale of the accused product to the defendant--here ULSI--by the owner or someone authorized to sell the product--here HP--prior to ULSI's own alleged infringing use or sale. The question is whether HP sold the accused product to ULSI. The preamble to the HP-ULSI agreement states: "[t]his Agreement details prototype and production fabrication of an integrated circuit designed by ULSI ..." (emphasis added). Jt.App. at 38. The contract specifies that "the ownership rights to the chip designs, mask works, and physical layout computer data and processes shall be in the party who develops and designs the same." Jt.App. at 43 (emphasis added); those items were provided by ULSI. "[ULSI] will provide [HP] with the GDS-II data files containing the artwork database for the Part." ULSI "warrants that it owns all the rights to the information and processes including specifications, designs, instructions and Confidential Information provided to HP ... [and] warrants that it has the full power and authority to supply and to disclose such information to HP." Jt.App. at 41 (emphasis added). ULSI further "warrants that it has not improperly or unlawfully acquired the information and processes submitted to HP for the purpose of this agreement." Jt.App. at 42. The contract then proceeds to indemnify HP against any liability for, inter alia, patent infringement.

The affidavit of Richard R. Duncombe, Sales Manager for the contracting HP fabrication facility, provides the background for these contractual provisions:

I asked [the ULSI representative] whether he had any concerns about intellectual property rights. By this question, I wanted to make sure that ULSI's product could be fabricated without exposing HP to liability for patent, copyright, or trade secret infringement. [The ULSI representative] replied that ULSI had no concern about intellectual property implications. Nonetheless, in keeping with HP's custom and practice, ULSI agreed to indemnify HP against any claims of patent infringement, or other legal problems, which might arise from ULSI's design of the product.

Jt.App. at 411. Had HP intended to sell to ULSI a chip containing the Palmer invention pursuant to the authority granted by the HP-Intel cross-license, and had ULSI intended to use HP in such a capacity, none of the warranties and indemnifications would have been required.

"HP has no ownership rights to ULSI's products, and no rights to sell or use ULSI's products. HP is required to return or destroy the magnetic design tape and reticles after HP completes the foundry services...."

Jt.App. at 413. Because ULSI had provided its own integrated circuit design, HP never had ownership rights in the invention to sell to ULSI; if there was a 'sale,' it must have been of something else. Under the HP-ULSI contract, HP provided the blank silicon wafers upon which the ULSI design was etched. Jt.App. at 60. HP further provided the fabrication services for embedding the ULSI design on the blank HP wafer. Jt.App. at 57. ULSI paid HP for certain non-recurring expenses, Jt.App. at 58, and paid a scheduled amount for each fabricated chip containing the ULSI design. Jt.App. at 60. Despite language which might imply that HP sold "chips," Jt.App. at 60, the overall context of the contract demonstrates that the sale was of services, measured per chip, rather than sale of any technology (be it Intel's or ULSI's), as embodied in each chip.

Thus HP basically sold its fabrication services and not the C87 chip with the allegedly infringing invention. The first sale of the C87 chip--the accused device in this infringement action, allegedly with the Palmer invention in it--was by ULSI and not by HP. That sale by ULSI was therefore within the scope of § 271, and is not a protected action of a purchaser in the marketplace.

Again, ULSI's argument confuses the question of whether HP manufactured a product that infringes Intel's patent rights--allegedly it did, although that issue has yet to be definitively determined--with the question of whether that product was owned and sold by HP to ULSI, thus creating the 'first sale' defense.

The district court understood that the parties to this suit and their arrangements fit the pattern described earlier. This is not a situation in which there is a dispute between the parties to a contract regarding their understanding of their respective contract rights. Rather, the district court recognized that the so-called 'patent exhaustion' or 'first sale' defense was simply lawyer argument by ULSI trying to capitalize on the existence of an agreement between other parties (Intel and HP) in which ULSI had no part. The district court rejected the defense, granted a temporary injunction against ULSI (a matter within her discretion), and prepared for trial on the merits.

The majority overrules this action by the district court, and effectively ends the case by granting judgment in ULSI's favor on this so-called defense. Thus a sensible and socially desirable agreement between Intel and HP is turned into an unintended gift to all manner of infringers. The result creates a disincentive among competitors to invent rather than litigate, potentially disadvantaging companies in a volatile industry such as this in competing world-wide. ULSI's position is both bad law and bad policy, and does not merit approval by this court.

Because the District Court did not abuse its discretion, commit an error of law, or seriously misjudge the evidence in holding that Intel was entitled to a preliminary injunction, I would vacate the stay and affirm the trial court's judgment.5 Even if I had reservations about the propriety of a preliminary injunction at this stage, which I do not, I believe the majority's conclusion that "ULSI is immune from infringement," Op. at 1570, is premature, since, if taken literally, it not only overturns the injunction but terminates the litigation.ORDER

Aug. 26, 1993.

A combined petition for rehearing and suggestion for rehearing in banc having been filed by the APPELLEE, and a response thereto having been invited by the court and filed by the APPELLANT, and the petition for rehearing having been referred to the panel that heard the appeal, and thereafter the suggestion for rehearing in banc and response having been referred to the circuit judges who are in regular active service,

ORDERED that the petition for rehearing be, and the same hereby is, DENIED, and it is further

ORDERED that the suggestion for rehearing in banc be, and the same hereby is, DECLINED.

The mandate of the court will issue on September 2, 1993.

Circuit Judges PLAGER, NEWMAN and RADER would rehear the appeal in banc.

PLAGER, Circuit Judge, with whom NEWMAN and RADER, Circuit Judges, join, dissenting from the denial of rehearing in banc.

I respectfully dissent from the denial by the full court of Intel's Suggestion for Rehearing In Banc. The panel in this case held that a broad cross-license between companies--permitting them to compete rather than litigate--had the effect of immunizing third parties whose products otherwise infringe the patents which were the subject of the cross-license. To reach this result, the panel applied an old rule in a new way, imposing on the cross-license certain contractual effects not provided for in the agreement, not agreed to by the parties to the cross-license (by their own testimony), and directly contrary to the trial judge's findings on the point.

The result frustrates the contracting parties' reasonable commercial expectations, and destroys important property rights they had under the patent laws. No effort was made to justify this result on the grounds of public policy. For the reasons stated in my dissent to the panel opinion, in my view neither precedent nor policy is shown to be furthered by this new rule.

It can be expected that American industry, facing a worldwide competitive environment, will seek new and innovative arrangements for sharing emerging technology. The question raised by the case is whether the law permits American industry to freely share creative assets--patented inventions--by cross-licensing without running the risk of being held to have inadvertently lost the benefits of the patent system. The type of cross-license involved here apparently is not that uncommon, and likely presages a variety of new contractual arrangements among and between industry partners and competitors. Absent a showing of illegality or compelling public policy, it is not the place of courts to limit or expand these arrangements beyond the bounds of the agreements, and in a manner that is avowedly inconsistent with what the parties intended. Particularly is this so when the consequences of doing so, no matter how well-intentioned, are economically far-reaching and likely to be harmful.

It cannot be denied that in banc cases add more to the court's burden. Nevertheless we have demonstrated a willingness to take cases in banc to ensure that important principles of law are honored. The rule imposed on this type of cross-license by this case is of sufficient moment to principles of contract law and competitiveness to deserve the attention of, and decision by, the full court. Whether the result is changed or not, I would take the case in banc in order to give our decision the benefit of the views of the full court.