dissenting.
This is not a close case. One of the Apple patents at issue covers a spelling correction feature not used by Apple. Two other patents relate to minor features (two out of many thousands) in Apple’s iPhone—linking a phone number in a document to a dialer, and unlocking the screen. Apple alleged that it would suffer irreparable harm from lost sales because of Samsung’s patent infringement. For support, Apple relied on a consumer survey as direct evidence, and its allegations of “copying” as circumstantial evidence. The district court rejected both evidentiary bases. On the record of this case, showing clear error in the district court’s factual findings is daunting, if not impossible. Not surprisingly, Apple principally presses a novel legal theory in this appeal: that the narrowness of its injunction request eliminated its burden to show nexus between its alleged irreparable harm and Samsung’s patent infringement. The majority correctly rejects this theory and the case should have ended there.
So why doesn’t it? Because the majority finds legal error by the district court where none exists. Then, under the guise of the purported “legal error,” the majority reverses without deference the district *657court’s rejection of Apple’s survey evidence, never mentioning that the survey was rejected by the district court because Samsung’s serious challenges to its techniques and conclusions were unrebutted by Apple. The majority further relies on “evidence,” found nowhere in the record, that carriers or users preferred having the patented features on Samsung’s phones. It also concludes — contrary to our case law— that Apple’s alleged evidence of “copying” is sufficient to show nexus to Apple’s alleged lost-sales. Because the majority here reaches a result that comports with neither existing law nor the record in this case, I must respectfully dissent.
A
Injunctions in patent cases, as in other areas of law, require evaluating the traditional four factors, including irreparable harm. Following eBay Inc. v. MercExchange, L.L.C., 547 U.S. 388, 126 S.Ct. 1837, 164 L.Ed.2d 641 (2006), when the infringing feature is but one of several components of the accused product, our precedent has clearly and consistently required patentees requesting injunctions to establish a nexus between the alleged irreparable harm and the patent infringement. This nexus showing is, of course, an indispensable pre-requisite in a case such as this, where, we are told, the infringed features are merely three of potentially hundreds of thousands of patented features in a single product. Requiring a showing of nexus is necessary to prevent undue leverage wielded by patents on minor features.
B
Turning to the case, the majority’s first error is its determination that the district court’s analysis was legally erroneous. Specifically, the majority states that, regarding Apple’s alleged irreparable harm from lost sales, “it was legal error for the district court to effectively require Apple to prove that the infringement was the sole cause of the lost downstream sales.” Majority Op. at 641; see also id. at 641-42, 644. But the majority quotes nothing from the district court’s opinion to show there is such an error. And for good reason: there is nothing. Hence, there is no error.
The words “sole” and “predominant” are not even present in the district court’s opinion.1 There is simply nothing in the district court’s opinion that explicitly or implicitly required Apple to show that the patented features were the “sole,” “predominant,” or “exclusive” reasons for purchasing Samsung’s products. Nevertheless, the majority concludes that the district court’s rejection of Apple’s di*658rect evidence — the consumer survey and testimony by its expert, Dr. Hauser— “seems to be predicated on an incorrect understanding of the nature of the causal nexus requirement, as discussed' above,” i.e., the so-called “legal error” by the district court. Id. at 644.
In reality, however, the district court simply weighed the evidence and found it lacking: “[t]he weight of the evidence shows that Apple’s conjoint study fails to demonstrate that the features claimed in the '647, '721, and '172 patents drive consumer demand for Samsung’s infringing products.” Injunction Order at *131 The district court reasoned that Apple made “only cursory arguments” about Dr. Hauser’s survey, while in contrast, Samsung challenged its myriad deficiencies including that the survey “omitted the major factors and major drivers of sales,” “overstated the scope of the claimed features and improperly included noninfring-ing alternatives,” and “produced nonsensical results, such as the conclusion that the patented word correction feature (corresponding to the '172 patent) was worth about $102 on a phone that cost $149.” Id. at *12-13. The district court therefore found that Apple “d[id] not rebut Samsung’s critiques of Dr. Hauser’s techniques or show that Apple’s conjoint study in this case established] a causal nexus.” Id. at *13.
In making these factual findings, the district court followed our case law faithfully. Nothing in the district court’s opinion suggests that it deviated from our precedent. Rather, it is the majority that deviates from our precedent by repeating as a mantra the phrase “some connection” in Apple Inc. v. Samsung Electronics Co., 735 F.3d 1352 (Fed.Cir.2013) (“Apple III”) detached from the causal nexus standard explained in our prior cases. See Majority Op. at 640, 640-41, 641 n. 1, 641-42, 644. The phrase “some connection” was not made in a vacuum, but instead in the context of our prior decisions, which goes ignored by the majority here.
Specifically, in Apple II, we explained that the connection between harm and infringement must be more than “insubstantial:”
It is not enough for the patentee to establish some insubstantial connection between the alleged harm and the infringement and check the causal nexus requirement off the list. The patentee must rather show that the infringing feature drives consumer demand for the accused product.
Apple Inc. v. Samsung Elecs. Co., 695 F.3d 1370, 1375 (Fed.Cir.2012) (“Apple II”) (emphasis added). And in Apple III, we explained that the requisite connection might be met by evidence showing that the patented feature is a “significant” driver of demand:
There might be a variety of ways to make this required showing, for example, with evidence that a patented feature is one of several features that cause consumers to make their purchasing decisions. It might also be shown with evidence that the inclusion of a patented feature makes a product significantly more desirable. Conversely, it might be shown with evidence that the absence of a patented feature would make a product significantly less desirable.
Apple III, 735 F.3d at 1364 (emphasis added).
The majority simply dismisses these examples from Apple.III on grounds that they “do not delineate or set a floor on the strength of the connection that must be shown to establish a causal nexus” but instead are “examples of connections that surpass the minimal connection necessary.” Majority Op. at 641, n. 1. But divorcing Apple III’s “some connection” *659language from the examples directly following those words morphs our decision into something it was not. And the majority’s attempt to sidestep that language by relying on Apple III’s cup holder example (illustrating an insufficient connection) fares no better. Id. In this case, Apple’s evidence fell far short of even the meager cup holder example, as Apple failed to offer any defensible evidence on consumers’ willingness to pay even a nominal premium for the patented features over non-infringing alternatives. See Injunction Order at *13 (finding that Apple “d[id] not rebut Samsung’s critiques of Dr. Hauser’s techniques or show that Apple’s conjoint study in this case establishes a causal nexus.”).
Perhaps recognizing its error, the majority reissued its opinion in this case to remove the implication that even an insignificant connection might be enough to satisfy the causal nexus requirement.2 While this change is a more accurate reflection of our law, it does not obviate the central problem with the majority’s conclusion in this case. As we stated in Apple III, “[t]he question becomes one of degree, to be evaluated by the district court.” Apple III, 735 F.3d at 1368. Here, the district court weighed the evidence and found it lacking. • Injunction Order at *13 (“[T]he weight of the evidence shows that Apple’s conjoint study fails to demonstrate that the features claimed in the '647, '721, and '172 patents drive consumer demand for Samsung’s infringing products.”). The majority identifies no basis for overturning this finding with its conclusion — unsupported by the record — that “Apple did, however, show that ‘a patented feature is one of several features that cause consumers to make their purchasing decisions.’ ” Majority Op. at 644 (quoting Apple III, 735 F.3d at 1364).
C
Hamstrung by the deficiencies in Apple’s direct survey evidence, the majority trumpets instead Apple’s “copying” evidence and even creates new evidence:
Given the strength of the evidence of copying and Samsung’s professed belief in the importance of the patented features as a driver of sales, and the evidence that carriers or users also valued and preferred phones with these features, the district court erred by disregarding this evidence, which further establishes a causal nexus and Apple’s irreparable harm.
Majority Op. at 643. All three parts of this statement are wrong: there was no evidence at all of such “carriers’ or users’ preference;” there was no “strong” evidence of “copying;” and “copying” alone is not dispositive to establish a causal nexus to Apple’s alleged irreparable harm from lost sales.
First, the majority’s “carriers’ or users’ preference” theory was not mentioned at all by the district court. The majority asserts that “[t]he district court acknowledged that Apple presented evidence that carriers ('721 patent) and users ('172 patent), not just Samsung, preferred and valued the infringing features and wanted them in Samsung phones.” Id. at 643. The majority again quotes nothing from the district court’s opinion to show there is *660such an acknowledgement. Again for good reason: there is nothing. As the majority notes just two sentences later, the district court “failed to appreciate” that the evidence cited by Apple “did not just demonstrate that Samsung valued the patented features, but also that its carriers or users valued the features.” Id. The district court could not have “acknowledged” what it “failed to appreciate.” The majority reaches its creative interpretation of the evidence to find “carriers’ or users’ preference” all on its own.
The majority also cites nothing from the record to support its “carriers’ or users’ preference” theory. I can only guess that the majority’s “users (T72 patent) preference” theory is relying on its earlier statement that “users criticized Samsung’s non-infringing keyboards and word-correction designs,” for that is the only reference by the majority to anything in the record in connection with users and the '172 patent. See id. at 642 (citing J.A. 20985). The document in the Joint Appendix on page 20985, however, is merely an internal Samsung e-mail message that mentioned “carrier issues” with Samsung’s keyboard user interface and referred to a table of information immediately following. Apple’s expert, Mr. Cockburn, concluded from the table that Samsung was proposing to use the feature defined in the '172 patent. J.A. 10700-02 (“And the next column across says ‘shows word in suggestion bar but does not change in the text field until user accepts or hits space.’ So this is the infringing method.”). Immediately below the text interpreted by Apple’s Mr. Cock-burn as proposing the infringing feature, Samsung’s employees noted “[Carrier requests additional information] It is not clear exactly what the issue is.” J.A. 20988 (brackets in original). The carrier was concerned about and had some “issue” with Samsung’s proposal to change to the purported infringing feature; the “carrier issues” were not about Samsung’s previous non-infringing method.
This e-mail message mentioned no users’ or carriers’ criticisms of Samsung’s non-infringing alterative to the '172 patent’s method. See J.A. 20983-88, J.A. 10700-02. Moreover, even if this e-mail were to show such criticisms, a negative view towards a non-infringing feature does not prove a positive preference towards the patented feature. Consumers could have preferred many other non-infringing word correction alternatives to the '172 patent, including Apple’s implementation in its “undisputedly successful” products that do not practice the '172 patent. See Majority Op. at 638.
Likewise, I can only guess that the majority’s “carriers ('721 patent) preference” theory is relying on its earlier statement that “Samsung e-mails not[ed] that certain carriers disapproved of the noninfringing ‘circle lock’ alternative,” for that is the only statement by the majority tying the '721 patent to carriers. See id. at 642 (citing J.A. 21019). The document in the Joint Appendix on page 21019, however, is merely an internal Samsung e-mail message referring to a single carrier’s “negative response towards our company’s circle lock playing the role of the unlock visual cue.” The majority’s characterization of the negative response as a “disapproval” is much too strong because the response was only preliminary; the carrier had not reviewed an actual working sample and was “requesting] to review actual working sample.... ” See id. Moreover, as discussed above, a negative view towards a non-infringing feature does not prove a positive preference for the patented feature. The evidence cited by the majority of a “negative response” does not show that any carrier preferred the feature defined by the '721 patent. The majority’s “carriers’ or users’ preference” arguments *661and the factual record it builds for support dissolve upon review of the evidence.
What we are therefore left with is the majority’s reliance on the so-called “copying” by Samsung to justify its reversal of the district court’s finding of no irreparable harm from lost sales. And the factual support is weak. The majority concedes as much in concluding that “the evidence may not make a strong case'of irreparable harm....” Id. at 647.
Nevertheless, the majority states that “[t]he district court wrote that there was evidence ... ‘indicative of copying.’ ” Id. at 642. The quotations upon which the majority relies, however, are not the district court’s findings. Rather, they are the district court’s recitation of Apple’s contentions, with which the district court disagreed. As the district court noted, “[wjhile indicative of copying by Samsung, this evidence alone does not establish that the infringing features drove customer demand for Samsung’s smartphones and tablets.” See Injunction Order at *14 (emphasis added). The district court, of course, did not mean that Apple proved copying for all three patents-in-suit. As the district court noted, Apple did not practice or allege copying of the '172 patent. Id. The district court also rejected Apple’s only support for its contention that it practiced the '647 patent. Id. at *15 (finding Apple’s only evidence of its own use “did not directly equate asserted claim 9 of the '647 patent with ‘data-detectors’ ”). Without Apple practicing these patents, Samsung obviously could not-have copied the patented features from Apple’s products.
The district court also discounted Apple’s evidence of “copying,” because “[s]ome of the cited Samsung documents show that Samsung valued numerous other nonin-fringing features.” See id. In fact, Apple’s evidence of “copying” lacked any connection to the critical details that define the patented features. The handful of internal Samsung documents cited by Apple merely addressed generic or un-patented aspects of Apple’s linking and screen-unlocking features. For example, one internal Samsung analysis recommended that Samsung provide “Links for memo contents such as Web, Call and E-mail, that can be linked.” J.A. 20584. But the asserted '647 patent claim does not monopolize the general concept of linking from documents; it is limited instead by specific elements such as. “displaying] a pop-up menu of the linked actions” and more. See Injunction Order at *1 (detailing asserted '647 patent claim 9). None of these critical elements were addressed in Apple’s “copying” evidence.
Similarly, another internal Samsung analysis compared Apple’s “unlocking standard by sliding” with Sam-sung’s “unlocking] with only a slight flick motion.” J.A. 20347. But the '721 patent does not deal with an innovation based on the strength and speed of the touch input, i.e., “sliding” versus “slight flick motion;” it requires instead details such as “displaying] visual cues to communicate a direction of movement of the unlock image required to unlock the device” and more. See Injunction Order at *2 (detailing asserted '721 patent claim 8). Again, none of these critical elements were addressed in Apple’s “copying” evidence. Merely mentioning generic or un-patented aspects of Apple’s linking and screen-unlocking features is clearly insufficient to show copying of the relevant patented features.
Finally, the majority concludes that the evidence in this case, which boils down to Apple’s allegations of “copying,” is enough to show nexus to Apple’s alleged irreparable harm from lost sales. This conclusion is contrary to our precedent. As the district court stated, “the parties’ subjective *662beliefs about what drives consumer demand are relevant to causal nexus, but do not independently satisfy the inquiry.” Id. at *14. Once again, the district court was doing nothing more than faithfully following our case law. We have repeatedly affirmed the district court’s previous rejections of the same allegations of “copying” as insufficient to show irreparable sales-based harm. Apple I at 1327; Apple III at 1367. As we have explained, to prove nexus to the alleged lost-sales, “the relevant inquiry focuses on the objective reasons as to why the patentee lost sales, not on the infringer’s subjective beliefs as to why it gained them (or would be likely to gain them).” Apple I at 1327-28.
The district court was well within its discretion to reject Apple’s contentions of “copying.” There is simply no basis for this court, on an abuse of discretion review, to reverse the district court’s denial of Apple’s injunction request.
D
In sum, the majority states that “the evidence [of consumer preferences and buying choices] is to be determined on a case-by-case basis based on what the evidence indicates.” Majority Op. at 643. The district court did exactly that in this case. Given the unassailable factual findings by the district court, the majority faces a tough mountain to climb to reach a reversal.
Thus, in order to reach its result, as described above, the majority rests on findings of non-existent legal error, of “carriers’ preference” created without record support, and of “copying” as disposi-tive to. show causal nexus to lost sales that is contrary to our case law. I must disagree with the majority’s approach and its conclusion that Apple would suffer irreparable harm from Samsung’s patent infringement.3
E
Finally, I also note the majority’s discussion on the public interest factor. I agree with the majority that the public’s interest in competition, without more, does not necessarily decide this- factor against granting an injunction. But it does not follow that the public interest “nearly always” favors granting an injunction as the majority states. According to the majority, “injunctions are vital to this system. As a result, the public interest nearly always weighs in favor of protecting property rights, especially when the patentee practices his inventions.” Id. at 647.
The majority repeatedly relies on the statutory right to exclude others from practicing a patent and the public policy embodied in the statute. See id. at 638-39, 641-42, 646. But I am confident that we all remain mindful that pre-eBay, “[a]c-cording to the Court of Appeals, this statutory right to exclude alone justifie[d] its general rule in favor of permanent injunc-tive relief.” eBay, 547 U.S. at 392, 126 S.Ct. 1837. The Supreme Court, however, unanimously rejected that approach, rea*663soning that “the creation of a right is distinct from the provision of remedies for violations of that right.” Id. For the same reason, the statutory right to exclude should not categorically bias the public interest factor “strongly ” in the determination of the injunctive remedies as the majority asserts. See Weinberger v. Romero-Barcelo, 456 U.S. 305, 312, 102 S.Ct. 1798, 72 L.Ed.2d 91 (1982) (“The Court has repeatedly held that the basis for in-junctive relief in the federal courts has always been irreparable injury and the inadequacy of legal remedies.”). The particular facts of a given case matter. As Justice Kennedy explained, “[w]hen the patented invention is but a small component of the product the companies seek to produce and the threat of an injunction is employed simply for undue leverage in negotiations, legal damages may well be sufficient to compensate for the infringement and an injunction may not serve the public interest.” eBay, 547 U.S. at 396-97, 126 S.Ct. 1837 (Kennedy, J., concurring).
Based on this record, I cannot agree with the majority’s broad warning that “[i]f an injunction were not to issue in this case, such a decision would virtually foreclose the possibility of injunctive relief in any multifaceted, multifunction technology.” See Majority Op. at 647. Rather, injunctive relief will be appropriate when and if, consistent with our case law, the causal nexus requirement is met. This is not such a case.
. The district court used the words "exclusivity” and "exclusively,” but only in the context of rejecting Apple's contention of irreparable reputational harm. See, e.g., Apple Inc. v. Samsung Elecs. Co., No. 12-CV-00630-LHK, 2014 WL 7496140, at *11 (N.D.Cal. Aug. 27, 2014) ("Injunction Order ”). And while the majority declines to reach the reputational harm issue, the concurrence does not. Here, the district court found that Apple’s licenses to other competitors were fatal to its claim that it had a "reputation for exclusivity” over the patented features. Id. The concurrence simply disregards this finding. Instead, the concurrence faults Samsung for failing to show evidence "that any of the licensees practiced any of the patented features” and states that Apple only "bears the burden of production,” after which the burden shifts to Samsung. Concurrence at 655. But the party requesting injunction bears more than the "burden of production;” it bears the “burden of proving irreparable harm,” a burden which Apple failed to meet. See Robert Bosch LLC v. Pylon Mfg. Corp., 659 F.3d 1142, 1154 (Fed. Cir.2011), Apple’s failure of proof cannot be excused by disregarding the district court’s factual findings and arbitrarily shifting the burden of proof to Samsung.
. The majority’s original opinion stated that: "Apple did not establish that that. [sic ] these features were the exclusive or significant driver of customer demand, which certainly would have weighed more heavily in its favor.” Apple Inc. v. Samsung Elecs. Co., 801 F.3d 1352, 1363 (Fed.Cir.2015), vacated, Order (Fed.Cir. Dec. 16, 2015) (emphasis added). The majority’s reissued opinion removes the words "or significant” from this sentence. Majority Op. at 644.
. I also disagree with the majority's reversal of the district court’s findings that remedy at law would be adequate. This reversal is premised on the majority’s disagreement with the district court's findings of no irreparable harm and the majority’s acceptance of Apple’s contention that any lost downstream sales would be "difficult to quantify.” Majority Op. at 645. We noted previously that if "Apple cannot demonstrate that demand for Samsung’s products is driven by the infringing features, then Apple’s reliance on lost market share and downstream sales to demonstrate the inadequacy of damages will be substantially undermined.” Apple III at 1371. Because I agree with the district court that Apple failed to show irreparable harm, I would also affirm the district court’s finding that Apple failed to show inadequacy of legal remedy.