Defendant Premier Retail Networks, Inc. appeals from a partial default judgment by the United States District Court for the District of Minnesota, finding that Premier’s 37-day delay in filing an answer resulted from “contumacious” misconduct and thereby justified both the denial of Premier’s motion to set aside the entry of default and the grant of a default judgment. Advanced Communication Design, Inc. v. Premier Retail Networks, Inc., 186 F.Supp.2d 1009 (D.Minn.2002). Further, Premier appeals the district court’s entry of a preliminary injunction that, in addition to the partial default judgment, relied on the finding that Plaintiff Advanced Communication Design, Inc. would likely establish infringement of its patent, U.S. Patent No. 6,133,908, by three versions of Premier’s accused “Interactive Network” apparatus.
Because Defendant Premier’s delay involved unintentional wrongdoing only, we conclude that the district court abused its discretion by finding that a short delay and Premier’s awareness of the complaint evidenced the willfulness that could alone justify the entry of a default judgment. Accordingly, we .reverse and remand for additional proceedings, including a determination on liability.
For the preliminary injunction, we affirm the district court’s ruling as to one of the accused versions of the “Interactive Network” but reverse as to the other two versions. Acting within its discretion, the district court reasonably found that one of these versions likely does not contain a “local area network” or “LAN” — regardless of which definition it relied on for that term — and that this version of the accused device therefore likely falls within the scope of the claimed invention. Notably as well, Premier’s own engineer does not assert that this version operates without using a LAN.
Further, the district court correctly determined for purposes of the preliminary injunction that the “directly connected” limitation covers more than simply an embodiment described in the rest of the specification; it likely covers the ordinary meaning of that limitation, as understood by an artisan of ordinary skill. And the court did not clearly err in finding that the “data control means” limitation was likely satisfied. Given Advance’s likelihood of success of proving infringement as to this particular version of the “Interactive Network” and the absence of any serious challenge to the ’908 patent’s validity, we also find no abuse of discretion with the court’s *967analysis of the remaining preliminary-injunction factors, including its irreparable-harm analysis.
For the other two versions of the accused Interactive Network, however, the district court abused its discretion because not only did it fail to assign the claim term “local area network” its presumptive ordinary meaning, it also credited Advanced’s conclusory assertion that the “Interactive Network” does not contain a LAN. Indeed, the affidavit on which Advanced relies does not even specify whether all three versions of the accused device operate without a LAN or even whether three versions of the Interactive Network exist. Given that a movant for a preliminary injunction always maintains the burden of establishing an entitlement to such extraordinary relief, we conclude that Advanced’s conclusory assertion cannot establish an entitlement to that relief here. Accordingly, we reverse the district court’s order to the extent it enjoins Premier from making, using, selling or offering to sell these two versions of the Interactive Network.
The first part of this opinion will discuss the relevant background to the default judgment ruling and the district court’s analysis. It will then proceed directly to our analysis of that issue. We then discuss the ’908 patent, the district court’s preliminary injunction ruling and our analysis of it.
I
As recounted by the district court, the pretrial discussions between the parties here began well before the filing of Plaintiff Advanced’s complaint. To begin, on February 15, 2001, Advanced’s Chief Executive Officer, Marco Scibora, sent a letter to Premier’s president advising him that Premier’s “Interactive Network” product was infringing Advanced’s ’908 patent. A month later, on March 7, 2001, Premier’s CEO — Jeffrey Cohen — called Advanced’s Scibora, to see whether they could “amicably” resolve the dispute.
On April 9, 2001, lawyers from each company held a teleconference so that Advanced could learn more about Premier’s allegedly infringing Interactive Network. A few days later, on April 12, 2001, the CEO from each company (Scibora and Cohen) met to discuss a potential settlement. The parties offer different accounts about the conversation that followed: according to Scibora, he offered to settle the dispute “if and only if’ the parties could reach a “viable” business resolution; according to Cohen, the parties had reached a handshake agreement in which Advanced agreed not to sue Premier for its alleged infringement. (E.g., J.A. 407.)
Regardless, on April 20, 2001, Scibora and Advanced’s attorneys telephoned Premier and its outside patent counsel (Paul Davis, Esq., of the law firm Wilson Sonsini Goodrich & Rosati LLP) again to learn more about Premier’s accused product. During this conversation, asserts Advanced, Premier’s managerial and engineering personnel described the structure and function of the Interactive Network, a description that led Advanced to confirm that this product was in fact infringing the ’908 patent. (E.g., J.A. 287-88, 330.)
On May 3, 2001, Advanced’s patent attorney delivered another letter to Premier’s attorney, Davis. The letter described the accused Interactive Network and explained how it infringed the ’908 patent; further, it offered to resolve the dispute amicably. Neither Advanced nor its attorneys received a response to the letter, however, even though counsel subsequently called Premier’s attorney about the matter.
On June 1, 2001, Advanced filed — but did not serve — a complaint against Premi*968er alleging infringement of the ’908 patent. See generally Fed.R.Civ.P. 4(m) (giving plaintiff up to 120 days to serve the complaint and summons after fihng them with the court). On June 21, 2001, another attorney for Advanced wrote another letter to counsel for Premier, asking again for a response to Advanced’s earlier May 2001 letter. Premier did not respond.
Several weeks passed without any communication and then, on August 7, 2001, Advanced’s CEO Scibora wrote to Premier’s CEO Cohen, advising that Advanced had filed its complaint and that Premier’s attorney (Davis) had not yet responded to any of the previous letters from Advanced or its counsel. The letter also cautioned that if Advanced received no response by August 16, 2001, it would proceed with the litigation.
A
On August 21, 2001, having not heard from Premier, Advanced had its complaint served on Premier’s CFO Songey (J.A. 254, 321), thereby triggering the 20-day period for fifing a responsive pleading or motion. See Fed.R.Civ.P. 12(a). Songey forwarded the process to both Premier’s outside law firm and its CEO, Cohen. Cohen in turn told Songey and Davis that he himself would tend to the Advanced lawsuit, “CEO-to-CEO,” and that they should not concern themselves with it. (E.g., J.A. 413, 416.) Shortly thereafter, however, Cohen stepped down as CEO and Chairman of the Board; and while Cohen kept his seat on Premier’s Board of Directors, he “abruptly” vacated the Premier premises on August 24, 2001, or so. (J.A. 992.)
The 20-day period for fifing a response to the Advanced complaint passed without Premier fifing any response whatsoever. Approximately two weeks after this fifing deadline, on September 25, 2001, Advanced’s attorney contacted Wilson Sonsini to advise Davis of Premier’s overdue response and to ask about the status of Premier’s answer. (J.A. 305, 413.) According to a contemporaneous memorandum recorded by Advanced’s attorney, Davis said he knew about the complaint; but he expressed “surprise” that Premier had not yet responded. (See J.A. 946.) Davis also did not say whether his firm intended to represent Premier in this litigation, and nor did he affirmatively indicate that Premier intended to respond at all. (See J.A. 305, 413, 946.)
The next day, on September 26, 2001, Advanced applied to the district court clerk for the entry of a default against Premier. See Fed.R.Civ.P. 55(a) (authorizing the clerk of court to enter a default against a party who “has failed to plead or otherwise defend” against a litigation). The clerk granted the application and entered the default.
Various officers at Premier, including CFO Songey, aver that they only first became aware of its overdue answer on October 8, 2001, at a meeting with Premier’s attorneys. (J.A. 393, 416.) Premier then instructed its counsel to get a response on file as quickly as possible. And Premier’s attorneys, on October 15, 2001, then contacted Advanced’s attorneys and also learned about the entry of default. Two days later, on October 17, 2001, Premier filed its answer — 37 days after the fifing period had passed. Advanced, meanwhile, also moved on that same day for both a partial default judgment and a preliminary injunction. Premier opposed the motions and further moved to set aside the entry of default for good cause shown.
B
After a hearing, the district court denied Premier’s motion to set aside the entry of default and granted Advanced’s motion to *969enter a default judgment. As the district court saw it, Premier’s conduct evinced a pattern of 'willfully ignoring communication about the alleged infringement. The court drew this inference of culpability from the facts that Premier had been personally served with the summons and complaint on August 21, 2001, and that Premier, through its CFO Songey, had forwarded that complaint to its attorneys and the CEO at the time, Cohen. Moreover, reasoned the court, when Advanced’s counsel contacted Premier’s counsel Davis about the status of Premier’s overdue response, Davis indicated that he knew of the complaint but said nothing about whether Premier intended to answer.
Relying heavily on precedent from the Fifth Circuit, the Ninth Circuit and district court opinions from New York, New Jersey and other non-binding authorities, the court further rejected the excuse that CEO Cohen’s “unexpected and disruptive departure” from Premier caused or at least contributed to Premier’s late filing. Informed by these non-binding precedents, the court simply concluded that this excuse, even if true, did not reheve Premier of its obligation to defend the lawsuit. Citing an opinion from a district court residing in the Eleventh Circuit, the court determined that organizations must have “systems” or procedures for checking up on matters like these.
As to both the motion to set aside the entry of default on the one hand and to enter a default judgment on the other, the court concluded that it need only consider Premier’s “contumacious” conduct. This was so, the court continued, because an apposite Ninth Circuit decision instructed that a finding of culpable conduct can alone justify a default judgment and thereby excuse a district court from having to analyze other relevant considerations. (See J.A. 11-14) (comparing this case to Alan Neuman Prods., Inc. v. Albright, 862 F.2d 1388, 1391 (9th Cir.1988) and citing Inman v. Am. Home Furniture Placement, Inc., 120 F.3d 117 (8th Cir.1997), Ackra Direct Mktg. Corp. v. Fingerhut Corp., 86 F.3d 852 (8th Cir.1996)).
This appeal followed.
II
The entry of default must precede the entry of a default judgment if a party fails to respond to a pleading or to otherwise defend as required by the Federal Rules of Civil Procedure. Fed.R.Civ.P. 55(a), (b). A defaulting party may remove the entry of default, however, upon a showing of “good cause,” Fed.R.Civ.P. 55(c). Indeed, a party may even undo the entry of a default judgment, provided it can establish “excusable neglect” under the more stringent requirements of Fed.R.Civ.P. 60(b). Fed.R.Civ.P. 55(c); accord Johnson v. Dayton Elec. Mfg. Co., 140 F.3d 781, 783-84 (8th Cir.1998) (explaining that although “good cause” under Rule 55(c) and “excusable neglect” under Rule 60(b) examine the same factors, “most decisions ... hold that relief from a default judgment requires a stronger showing than relief from a mere default order”) (citations omitted). Eighth Circuit precedent governs our analysis of this procedural issue.
In addition, the “more lenient ‘good cause’ standard” under Rule 55(c) applies here because Premier filed an answer and moved to set aside the entry of default before the district court ever entered a default judgment. See Johnson, 140 F.3d at 783-84 (holding that the “more lenient” good cause standard — as opposed to the stricter Rule 60(b) “excusable neglect” standard — applied when defendant filed an answer and moved to set aside the entry of default before the district court had properly entered a default judgment). Further, we review a district court decision *970finding no good cause and entering a default judgment for an abuse of discretion, Hall v. T.J. Cinnamon’s, Inc., 121 F.3d 434, 435 (8th Cir.1997).
The Eighth Circuit has stated that the equitable “good cause” analysis requires consideration of “all [the] relevant circumstances” surrounding the default, with an emphasis on the following factors: (1) the length and reason for the delay in filing an answer or motion, along with consideration of whether the defendant acted in good faith; (2) the prejudice that this delay caused the plaintiff to incur; and (3) whether the defendant has a meritorious defense. E.g., Union Pac. R.R. Co. v. Progress Rail Servs. Corp., 256 F.3d 781, 782-83 (8th Cir.2001) (citing and quoting Pioneer Inv. Servs. Co. v. Brunswick Assoc. Ltd. P’ship, 507 U.S. 380, 394-95, 113 S.Ct. 1489, 123 L.Ed.2d 74 (1993)); Johnson, 140 F.3d at 784. As to the first factor, the precedents have consistently drawn a distinction between intentional, “contumacious” conduct on the one hand versus negligent conduct and a “marginal failure” to meet pleading deadlines on the other. E. g., Johnson, 140 F.3d at 784; Inman v. Am. Home Furniture Placement, Inc., 120 F. 3d 117, 119 (8th Cir.1997); Ackra Direct Mktg. Corp. v. Fingerhut Corp., 86 F.3d 852, 856 (8th Cir.1996). The former may warrant the denial of “good cause” and the entry of a default judgment; the latter may not, and a district court that essentially rules or finds to the contrary abuses its discretion. Id.; Jones Truck Lines, Inc. v. Foster’s Truck & Equip. Sales, Inc., 63 F.3d 685, 688 (8th Cir.1995).
The Eighth Circuit has also recently emphasized that while the reason for the delay, including a defendant’s putative negligence or “blameworthiness,” constitutes a “key consideration” in deciding whether to excuse a default or default judgment, a district court ought not focus narrowly on that behavior. Union Pac. R.R. Co., 256 F.3d at 783. Indeed, the “reason for a party’s delay” does not give a district court “carte blanche ... to disregard the other considerations” relevant to the default inquiry. Id. According to recent precedent, a “district court is required to engage in a careful balancing of multiple considerations,” including the other two factors listed above. Id. (emphasis added). At the same time, as Advanced hastens to note and as the district court here stated, other binding precedents clearly indicate that a district court may indeed disregard these other considerations when the defaulting party’s behavior involves “ ‘willful violations of court rules, contumacious conduct, or intentional delays.’ ” McMillan/McMillan, Inc. v. Monticello Ins. Co., 116 F.3d 319, 320 (8th Cir.1997) (citation omitted); Ackra Direct Mktg., 86 F.3d at 856-57.
In our view, we find it unnecessary to reconcile these seemingly inconsistent lines of Eighth Circuit precedent after we compare the conduct in this case to those excused in other binding precedents. In Union Pacific, for example, the Eighth Circuit determined that even under the more-strict “excusable neglect” standard of Rule 60(b), the district court abused its discretion by entering a default judgment against a defendant whose failure to timely answer resulted from a “recording error by its legal department.” 256 F.3d at 782. The defendant’s negligence involved a “minor mistake,” the court reasoned, and the trial court erred by focusing exclusively on that conduct instead of also considering (among other factors) the length of the delay, the prejudice to the plaintiff and the existence (if any) of a meritorious defense. Id. at 783. In addition, similar to this case, the defendant there contacted and began negotiating with plaintiff about having the default judgment set aside as soon as it learned of that judgment. Id. And *971when those negotiations failed, the defaulting defendant then moved to set aside the judgment “only three weeks after it had notice of the default and less than six months after [plaintiff had] filed its complaint.” Id. (emphasis added).
Similarly, in Johnson, the Eighth Circuit held an abuse of discretion occurred when the district court found no good cause to set aside a default that resulted from “poor communication” between the defendant, its attorney and its insurer. 140 F.3d at 784-85. As the reviewing court saw it, the defendant and its attorney did indeed act “carelessly,” and they even exhibited a “cavalier approach” to the filing deadline; but neither the defendant nor its counsel’s conduct demonstrated an “intentional flouting or disregard of the court and its procedures.” Id. at 785. In addition, the tardy defendant there attempted to cure its mistake within one day of learning of the default. Id. And thus, coupled with a showing of a meritorious defense and the absence of prejudice to the plaintiff, the Eighth Circuit reversed the trial court and found good cause to lift the entry of default. Id.
In Jones Truck Lines, the district court abused its discretion by declining to set aside a default judgment under the onerous “excusable neglect” standard (see Fed. R.Civ.P. 60(b)), given that the plaintiff there suffered no prejudice from the defendant’s otherwise untimely filing. 63 F.3d at 687. Further, defendant’s counsel in that case claimed the delay resulted from defendant’s hope that the parties could settle out of court, a claim that the parties’ correspondence substantiated. Id. at 687-88. This delay and the asserted reason for it simply did not show a “willful flaunting [sic] of the [filing] deadline,” the Eighth Circuit concluded. Id. That the defendant filed an answer as soon as the opposing party alerted it to the default judgment only further supported this conclusion, thereby rendering the trial court’s finding of willfulness “clearly erroneous.”
A
And so it goes in this case. Premier appears to have acted in good faith because while it did not respond to precomplaint offer letters, nor did its relatively brief failure to meet the filing deadline result from an intentional flouting of the rules themselves. To the contrary, as shown by the record, its 37-day delay instead resulted from otherwise excusable conduct, including a misunderstanding and perhaps even a “cavalier approach” by its attorneys, as well as a lapse in communication between the outgoing CEO Cohen and the rest of the Premier management team, e.g., CFO Songey. The district court erroneously transformed the relatively short delay and Premier’s awareness of the pending complaint into a reason in and of itself for finding intentional wrongdoing. Again, Premier’s error exhibited the same sort of lax approach found in Union Pacific, Johnson and Jones Truck, supra; but also like the tardy defendants in those precedents, it does not evince an intentional flouting or blatant disregard of court rules and procedures. Further, as stressed in the precedents discussed above, Premier took action as soon as it learned of the default, initiating communication via its attorneys with Advanced only a week thereafter and filing an answer only two days after that eounsel-to-counsel communication. Accordingly, the district court abused its discretion by finding that Premier’s behavior constituted inexcusable contumacious conduct.
Neither Hall nor Inman, both cited by Advanced, compels a different outcome. In Hall, no abuse of discretion stemmed from the district court’s entry of a default judgment when, unlike here, the defendant *972“watched the litigation from afar for over seven months,” despite its knowledge of both the lawsuit and plaintiffs request for a default judgment. 121 F.3d at 435. In addition, the defaulting party showed little or no regard for court rules and filing deadlines, as it even wrote a letter to the district court saying that it simply could not afford to hire local counsel to defend it in the subject litigation. Id. And yet, as the Eighth Circuit explained, the defendant belied its own written assertion to the trial court by thereafter retaining local counsel and subsequently moving to oppose the entry of a default judgment. Id. at 435-36.
Here, by contrast, Premier did not watch from “afar” as Advanced proceeded to obtain first an entry of default and then a default judgment. Instead, as soon as Premier realized that a default had been entered against it, Premier acted immediately and had its pleading filed — only 37 days late as opposed to the seven-plus months’ delay in Hall. And again, although Premier was aware that it had an imminent or even pending infringement action with Advanced, the record does not show that Premier was simply flouting the rules or otherwise playing games with either Advanced or (as in Hall) the district court. Indeed, the memorandum by Advanced’s own attorney reflects Premier counsel’s “surprise” that Premier had not yet responded to the service of Advanced’s complaint. As stated above, that failure perhaps constitutes a lax or “cavalier” approach to the filing deadlines; it does not show an intentional disregard or other contumacious conduct, the type of behavior needed to justify the entry of a default judgment without any consideration of the other “good cause” factors.
Likewise, in Inman, the Eighth Circuit found no abuse of discretion with the entry of a default judgment when the defaulting defendants blamed their attorney for failing to keep abreast of the litigation. 120 F.3d at 119. “Litigants choose counsel at their peril,” the court reasoned. Id. at 118-19. Further, the attorney there had failed to answer a three-year-old complaint and had neglected to respond to plaintiffs discovery requests and repeated inquiries from opposing counsel. Id. at 119. On top of that, the court reasoned, the defendants had also failed to appear for a scheduled deposition. Id. Last, neither the size of the damages judgment awarded against defendants nor the alleged strength of defendant’s defense could overcome defendants’ “dilatory conduct.”
Inman does repel some of the arguments that Premier makes on appeal, including arguments that a party simply cannot blame its attorney for a default and cannot use the size of the judgment as the sole basis for undoing a default judgment. (On the other hand, Advanced overstates the import of Inman, as well as another precedent (Ackra, supra), by maintaining that these precedents render consideration of the size of a damages award “irrelevant.” Those precedents said no such thing; they only indicated that the size of the judgments there could not overcome the other factors in those particular cases that augured in favor of entering a default.) Still, Inman does not purport to overturn the precedents, discussed above, that excuse negligent and even “cavalier behavior” — by either the defaulting party or that party’s attorneys — as to court rules and procedures. See Union Pac., 256 F.3d at 782-83 (excusing untimely fifing that resulted from a “recording error by [defendant’s] legal department”); Johnson, 140 F.3d at 784-85 (excusing untimely filing that resulted from defaulting party and counsel’s poor communication, carelessness and “cavalier approach” to the filing deadline). Equally important, the facts in In-man hardly resemble the background in *973this case. After all, unlike the situation there, Premier and its attorneys only failed to meet the filing deadline by 37 days (not three years).
In all, because of the short delay in this case, the poor communication that caused it, and the action that Premier immediately took to cure its filing oversight, the district court clearly erred by finding that this delay and Premier’s awareness of the complaint could alone amount to “contumacious” conduct and an intentional failure to comply with court-mandated deadlines. Accordingly, it abused its discretion in ruling on the motions. We therefore reverse the entry of the partial default judgment.
B
The analysis of the relevant prejudice and meritorious-defense factors shore up this conclusion. In the Rule 55(c) context, “prejudice” does not refer to a plaintiffs putative expectation that, after the filing period had passed without a response, it will automatically obtain a judgment or partial judgment in its favor See Johnson, 140 F.3d at 785 (“Relying on [a district court precedent], the [trial] court concluded that granting Dayton Electric relief would prejudice Johnson’s ‘expectations concerning the judgment’ and her ‘belief in the integrity of the system and the court’s authority.’ We reject this legal standard. Entry of default raises no protectable expectation that a default judgment will follow, and a party’s belief in the integrity of the system must include, to be reasonable, knowledge that a system of integrity makes exceptions for good cause shown.’ As numerous decisions make clear, prejudice may not be found from delay alone or from the fact that the defaulting party will be permitted to defend on the merits.”) (emphasis added). Rather, prejudice refers to more concrete and litigation-specific difficulties, “‘such as loss of evidence, increased difficulties in discovery, or greater opportunities for fraud and collusion.’ ” Id. (citation omitted).
In this case, the record does not reveal that Advanced would incur any of these difficulties if the default were set aside. Nor does Advanced even claim as much, focusing instead on Premier’s alleged contumacious conduct as the basis for upholding the district court’s ruling. Accordingly, the prejudice factor also cuts in Premier’s favor and further counsels that we overturn the partial default judgment.
C
So too does the meritorious-defense factor. Again, in the Rule 55(c) context, a meritorious defense does not mean a defense that would indisputably compel a finding in the defaulting party’s favor. See id. Rather, a meritorious defense means that the evidence or argument proffered by the defaulting party could reasonably lead to a finding in the defaulting party’s favor. See id.
At least one of the defenses that Premier puts forth could compel an ultimate finding in its favor. As more fully discussed below, Premier argues that at least two of its accused “Interactive Network” products do not infringe the ’908 patent because, unlike the claimed invention, these two versions use a “local area network” to connect the various stations of the preview device. The district court, for purposes of the preliminary injunction, essentially believed the contradictory (and conclusory) affidavit submitted by Advanced’s CEO, Scibora. Nevertheless, in our view, a reasonable fact finder could still believe the detailed testimony recounted in Premier’s affidavit, meaning a reasonable jury could indeed find in Premier’s favor. Thus, Premier has offered a meritorious defense as contemplated by Rule 55(c), thereby offering yet another reason *974to reverse the district court’s partial default judgment.
We next turn to the background underlying the district court’s preliminary-injunction analysis, including a discussion of claim 1 of the ’908 patent.
Ill
Broadly stated, Advanced’s ’908 patent claims an apparatus for allowing multiple customers at music or video retail stores to simultaneously sample an audio or visual product (e.g., a compact disc, a movie DVD or a video game) at the store itself. The invention purportedly reduces consumer frustrations with various audio-visual products and thereby leads to increased sales for a retailer that uses the apparatus.
To that end, as shown by the description of the ’908 patent’s preferred embodiment, a customer may take the video cassette or other audio/video product that interests them to the first of three main components of the claimed invention, the consumer “preview station 20.” At the preview station, the customer runs the bar code contained on the video or audio product across a bar card scanner. A “microcontroller 26” then takes the information fed into it by that bar code signal and sends it to the second main component of the claimed invention, the “preview station interface means,” using, e.g., a six-conductor modulator cable 50. This cable in turn has the requested bar-code information passed through to another microcontroller 42. A “device driver 64,” located within the third main component of the claimed apparatus (the “data control means 60”), then sends a request to that same microcontroller 42, which then passes on the request via a “data control interface means” 48.
The device driver 64 then “polls” each of the preview station interface means 40, to have information exchanged between that interface means and the data control means. The device driver 64 then sends the barcode data to the application software 68, also contained within the data control means. That application software, as explained by the specification, then conducts a database search for the requested information on the data control means’ hard drive 70. (The hard drive would need to have the requested information about the various video and audio products stored on it before a customer could actually use the claimed invention.) The data control means, using the various corresponding structures described in the specification, then has the requested information returned through the preview station interface means and finally back to the preview station itself, where the customer can view or listen to information about the scanned product, e.g., price of the product, artist of the CD or selected excerpts from the audio or video product itself. The whole process lasts one or two seconds.
*975[[Image here]]
The only claim at issue in this case is claim 1 of the ’908 patent:
1. A multi-station video/audio distribution apparatus that allows a plurality of users to simultaneously sample different video material and audio material, comprising:
(a) at least two preview stations, said at least two preview stations each having a user input, an audio output such that an analog audio signal corresponding to said user input may be heard, and a video display such that video material corresponding to said user input may be seen;
(b) data control means, said data control means for retrieving digitized audio and digitized video material corresponding to each of said user inputs;
(c) preview station interface means, said preview station interface means separate from and directly connected to said data control means and said at least two preview stations by non-switched connections not involving a local area network or wide area network, said preview station interface means for transferring each of said user inputs from its respective preview station to said data control means, for receiving each of said digitized audio material corresponding to each of said user inputs, for converting
each of said digitized audio materials to said analog audio signal and for transferring each of said analog audio signal to its respective preview station; and
(d) video display interface means, said video display interface means separate from and directly connected to said control means and said video display by connections not involving a local area network or wide area network, said video display interface means for receiving each of said digitized video materials corresponding to each of said user inputs, for converting each of said digitized video materials to a format acceptable to said video display, and for transferring said formatted video materials to said video display.
U.S. Patent No. 6,133,908, col. 8, lines 14-48 (issued Oct. 17, 2000). The three claim limitations at issue appear underlined, above: “local area network,” as in the claimed invention cannot use such a network or “LAN” to transfer information from the preview station interface means to the data control means; “directly connected,” as in the three main components of the claimed invention must exist separately but must also have a direct connection to one another; and “data control means,” one of the three major components (along with the preview station and *976the preview station interface means), as shown in the diagram above at 60.
As to each, the rest of the specification says nothing about a local area network or “LAN.” As to “directly connected,” the specification merely mentions during its description of an embodiment that the “video display 80 does not transfer data or information through the preview station microcontroller 26 but rather is directly connected to a video display interface means 82 through a high gauge wire 83 having RC connectors at either end.” (’908 Patent, col. 5, line 65 — col. 6, fine 2) (emphases added). Otherwise, the “directly connected” limitation is described only in the claims themselves; namely, dependent claims 17 (having “preview station interface means” and “video display interface means ... directly connected to said data control means by a computer bus”), 18 (having “low voltage preview station interface means ... directly connected to said at least two preview stations by low voltage cables not involving a local area network”) and 19 (having “low voltage preview station interface means ... directly connected to said at least two preview stations by low voltage cables not involving a local area network”) and independent claims 10 (requiring the same “directly connected” language as that used in claim 1, above) and 20 (same).
As to the third limitation in dispute, “data control means” (60, above), the specification makes clear that the structures used to perform the claimed function of “retrieving digitized audio and digitized video material corresponding to each of said user inputs” include:
a standard computer platform that incorporates a power supply, a back-plane/bus 61 that is connectable to the data control means interface 48 of the preview station interface means 40, a microprocessor 63, random access memory, and interfaces to various peripherals such as disk drives that operate in conjunction with disk controllers, modems, video displays, keyboards, and tape backup units. The data control means 60, or computer platform, utilizes operating system software 62 (e.g. UNIX) that has low level device drivers 64, file management utilities 66 and further utilizes application software 68 that operates within the computer platform to implement the full functionality of the apparatus 10. The computer platform also incorporates high-speed random access storage of audio and/or video material in digital form; the audio and/or video material is contained in data files stored on internal or external hard drives 70 that are connected to interfaces within the computer platform.
(’908 Patent, col. 3, line 64 — col. 4, line 14.)
As to the prosecution history, only Premier cites it; and then only to portions indicating that Advanced had added via amendment both the “directly connected” limitation and the “not involving a [LAN]” limitation so as to distinguish the ’908 patent from a prior art reference (“Allen”). CSee J.A. 506, 518, 526.) Unlike the Allen reference, said Advanced, “[c]laim 1 does not call for a network at all. Rather, claim 1 requires a direct, non-switched connection not involving a” LAN. (J.A. 551.)
A
The district court first justified the entry of a preliminary injunction on the strength of its partial-default judgment conclusion, saying that that judgment alone conclusively established the truth of the infringement allegations against Premier and thus entitled Advanced to an injunction pending a trial on damages. This conclusion is logically correct. But since the district court abused its discretion in entering a partial default judgment, *977we must also examine the separate analysis that the district court conducted under Fed.R.Civ.P. 65.
Citing the four well-known preliminary-injunction factors — the likelihood of success on the merits, irreparable harm, a balancing of the hardships and the public interest, Reebok Int’l Ltd. v. J. Baker, Inc., 32 F.3d 1552, 1555 (Fed.Cir.1994) — the court first found that Advanced would likely establish that the “Interactive Network” infringes claim 1 of the ’908 patent. This analysis revolved around the three limitations discussed above.
As to “non-switched connections not involving a local area network,” the court construed that limitation to cover a “network involving switching hardware or software” and a “network that switches hardware and software.” In so doing, the district court explained that the claim language itself requires that the invention not use a LAN or (as the district court defined that term, relying on the claim language itself) “switched connections. ” Further, as the district court understood it, the prosecution history clarified that any “local area network” is by definition “switched, not non-switched.” Moving to the fact-dependent infringement side of the analysis, the court also concluded that all three versions of Premier’s accused “Interactive Network” likely do not use “switched connections” or LANs, thereby bringing them within the scope of the claim. As support for this fact-based conclusion, the district court cited Advanced’s affidavits, as well as the affidavit from Premier’s engineer, Daniel Kyte.
As to the “directly connected” limitation, Premier argued to the district court that this term meant “joined by a single wire without deviation.” (J.A. 22, 427.) The court rejected this proposed construction, however, concluding instead that the limitation means “connected by non-switched connections not involving a local area network or wide area network.” According to the district court, the claim language in claim 1(c) itself directly supported this construction, as it explains that the required “direct connection” occur by way of “non-switched connections not involving a” LAN.
Contrary to Premier’s assertions, moreover, the specification did not purport to restrict the scope of the “directly connected” limitation to a “single wire”; such an understanding, it explained, would erroneously import a particular embodiment into the claim language. And the written description itself stated that, as to the described embodiment, the “present invention may be embodied in other specific forms without departing from the spirit of the essential attributes thereof.” Last, in the district court’s view, Advanced added the “directly connected” limitation during prosecution only to distinguish its invention from a prior art reference (Allen) that used a LAN; it did not add that language so as to narrow the limitation to a “single wire implementation.”
Having resolved the “directly connected” definition (at least for preliminary injunction purposes), the court indicated that no factual disputes remained as to the term and whether it covered the corresponding element in the accused Interactive Network devices. In short, Premier’s argument on this point rose and fell on the narrow interpretation it attempted to ascribe to the “directly connected” term. Accordingly, because the court rejected that narrow proposed construction, it found a likelihood of infringement as to this term, too.
As to the “data control means” limitation, the district court agreed with the parties that this was a means-plus-function limitation governed by 35 U.S.C. § 112 116. Again citing the language of the claim *978itself, the court construed the limitation’s function to cover “retrieving digitized audio and digitized video material corresponding” to the information requested by a customer via, e.g., the bar code contained on a particular CD package. And it determined that the corresponding structures that performed this function, see 35 U.S.C. § 112 If 6, included “file management utilities,” “data files stored on internal or external hard drives” as well as the other structures listed in the ’908 patent specification at “Col. 4, line 64 through Col. 5, line 17.” That citation to the specification identifies structures that constitute the invention’s “data control means.”
And, concluded the district court, the accused Interactive Network likely contains this limitation or its equivalent, too, given that it uses a “server” that performs the claimed function of retrieving digitized audio and video signals from a computer memory based on the information requested via the scanned-in bar code. In addition, in the district court’s view, the accused device also contained “identical structure”; as with the structures disclosed in the claimed invention’s specification, the Interactive Network also has: a standard computer platform with a microprocessor and random access memory; interfaces to various peripherals such as disk drives; operating system software; file management utilities; application software; and high-speed random access storage of audio or video material stored in an internal or external hard drive. The district court cited two affidavits to support this tentative finding.
B
Buttressed by a finding of a likelihood of infringement, the court’s remaining analysis of the irreparable-harm, balance-of-harms and public-interest factors fell into place. As to irreparable harm, -the court cited the rule that a “clear showing” of infringement and that the asserted patent would likely withstand a validity challenge perforce entitled the movant to a rebuttable presumption of irreparable harm. Advanced made that showing here, the court indicated, thus triggering the presumption. As the court saw it, moreover, Premier could not rebut the presumption based solely on the fact that Advanced had offered to license the claimed invention during the parties’ various communications.
Nor did the balance of harms tilt in Premier’s favor, the court found. In short, the “economic harm” posed to Advanced’s interests by the “unlicensed production and sale of the Interactive Network is obvious.” And any alleged contrary harm to Premier paled in comparison, the court noted. Last, the strong public interest in protecting patent rights also weighed in Advanced’s favor. Notwithstanding its ruling, the court agreed to stay the entry of the preliminary injunction pending this appeal.
On appeal, Premier challenges all but the public-interest finding. Given our deferential standard of review as well as the fact that the district court has not yet had the opportunity to conduct a Markman hearing, we reject these challenges as to one of the three versions of the Interactive Network. For the other two versions, we reverse.
IV
We review the grant of a preliminary injunction for an abuse of discretion. E.g., Amazon.com, Inc. v. Barnesandnoble.com, Inc., 239 F.3d 1343, 1350-51 (Fed.Cir. 2001). An abuse of discretion occurs when, for example, the trial court made a clear error of judgment in weighing the relevant factors or when it exercised that discretion based on an error of law or a clearly erroneous finding of fact. E.g., id. *979at 1350. We discern one such pivotal error with the court’s analysis.
A
Premier first argues that, as to the limitation “non-switched connections not involving a local area network,” the district court erred by disregarding an affidavit from Daniel Kyte, the director of Premier’s research and development. Kyte and Premier maintain that, unlike the claimed invention, accused versions 1.0 and 3.0 of the Interactive Network do in fact use a local area network or LAN. (J.A. 449-50.) (As the district court noted, not even Kyte avers that accused version 2.0 of the Interactive Network uses a LAN, meaning Premier seems to have conceded that this version likely contains this limitation.) Further, argues Premier, Advanced itself relied only on a conclusory affidavit from its CEO Scibora, as well as information that Advanced says it obtained via its telephone calls with Premier as to the elements of the Interactive Network. (This information is described in the May 3, 2001 letter that Advanced sent to Premier when it still thought the parties could resolve the dispute short of litigation.)
We agree and hereby undo the preliminary injunction as to versions 1.0 and 3.0 of the Interactive Network. Premier’s evidence of mfringement here consists of nothing more than a conclusory assertion by its CEO Scibora and an equally conclusory letter by its patent counsel. {See J.A. 301, 712.) Further, neither Scibora nor anyone else on Advanced’s behalf asserts that their conclusions turned on their having personally observed the accused 1.0 and 3.0 versions. (Indeed, nothing in Advanced’s affidavit or letter even recognizes that the accused Interactive Network has three versions.) Rather, Advanced’s knowledge appears to rest on the telephone conversation that it had with Premier personnel before suit began. While statements from Premier’s engineering and management personnel would not constitute inadmissible hearsay, see Fed. R.Evid. 801(d)(2)(D), nothing in the record reflects an admission that, as with the patented apparatus, all three accused versions do not use a LAN. Further, nor does Advanced claim it had to rely on these conclusory assertions because it never had the opportunity to observe versions 1.0 and 3.0 of the Interactive Network.
In short, we are convinced that the district court clearly erred by finding a likelihood of infringement by versions 1.0 and 3.0 based on the conclusory evidence submitted by Advanced, the party that bears the burden of showing an entitlement to the extraordinary relief of a preliminary injunction. See 11A Wright & Miller, Federal Practice & Procedure § 2949, at 215 (2d ed. 1995) (“All affidavits should state the facts supporting the litigant’s position clearly and specifically. Preliminary injunctions frequently are denied if the affidavits are too vague or conclusory to demonstrate a clear right to relief under Rule 65.”); see also Canon Comp. Sys., Inc. v. Nu-Kote Int'l Inc., 134 F.3d 1085, 1089 (Fed.Cir.1998) (rejecting challenges to district court’s findings for a preliminary injunction, as those challenges rested on a mere “conclusory statement”).
We further conclude that the district court’s finding of mfringement by these two versions tainted its analysis with regard to the remaining preliminary injunction factors as well. Thus, Advanced may not enjoin Premier from making, using, selling or offering to sell versions 1.0 and 3.0 of the accused Interactive Network. Because Premier has not challenged the district court’s conclusion about version 2.0 not containing a LAN — regardless of the definition ascribed to it — we uphold the *980preliminary injunction as to that version only.
Because Premier argues that other limitations are also missing from its accused devices, including version 2.0, we address those arguments, below.
B
Premier next essentially posits that the claim limitation “directly connected” can cover no more than a particular embodiment disclosed in the ’908 patent specification. In support of this position, Premier argues that the specification “only” once refers to the “directly connected” limitation, saying (as Premier recounts it) that the “video display unit does not transfer data or information through the preview station microcontroller” but instead has it “directly connected” to a video display interface means. From this, Premier extrapolates the following definition of “directly connected”: a “connection that does not transfer signals through a microcontroller.” (Appellant’s Br. at 49-50.) The district court therefore erred, goes the argument, because it defined “directly connected” using only the surrounding claim terms themselves as well as the prosecution history, thereby ignoring the “express teaching of the [’908 patent] specification.” (Id. at 50.)
For three reasons, we reject Premier’s argument. First, as Advanced suggests (Appellee’s Br. at 23-24), Premier is offering a new construction argument and definition on appeal, ones that it did not offer to the district court. As stated above, in the district court, Premier asserted that “directly connected” meant “joined by a single wire without any deviation.” (J.A. 427.) Accordingly, because the orderly adjudication of a claim construction dispute requires a consistent and clear presentation of the same arguments to both the trial and appellate courts, Premier has waived its new claim construction theory and accompanying definition on appeal. We simply should not and cannot tolerate moving targets.
Second, a claim is given its ordinary meaning unless otherwise specified or indicated. See, e.g., CCS Fitness, Inc. v. Brunswick Corp., 288 F.3d 1359, 1366-67 (Fed.Cir.2002) (discussing the claim construction presumption that a claim term covers the full scope of its ordinary meaning and the exceptions thereto). In this case, neither the specification’s terse mention of the “single wire” nor the oblique reference that Premier now points to on appeal imparts a clear indication to limit the ordinary meaning of the claim term “directly connected.” Thus, as the district court correctly explained, the intrinsic evidence does nothing here to detract from the scope of this limitation. Premier’s argument thus fails for this reason as well.
Third, Premier suggests that the district court erred by defining “directly connected” in terms of the related language that also appears in claim 1. Argues Premier, the district court should have instead assigned this claim language an “independent” definition altogether, lest it otherwise render that claim language meaningless and thereby read it out of the claim in its entirety. (See Appellant’s Br. at 49, 51) (citing Ethicon Endo-Surgery, Inc. v. U S. Surgical Corp., 93 F.3d 1572, 1578 (Fed.Cir.1996) (“If ... ‘connected to’ should be read broadly to include elements which are connected directly or indirectly, then this language would read on a lockout mechanism located anywhere in the surgical stapler, and the ‘connected to’ limitation would be meaninglessly empty.”)).
This contention has some force, as it appeals to the maxim that the construction of any legal document — like a statute, contract or patent — should try to give mean*981ing to every term in that document; otherwise, a lawyer or court will have erred by reading the chosen words of the document into oblivion. But in this instance, we see no abuse of discretion on these “independent-definition” grounds. In addition to the waiver and ordinary-scope impediments, Premier’s attack fails to confront the fact that if a claim term {e.g., “directly connected”) is sufficiently clear such that no other definition is needed, the district court simply has no duty to wave into existence a different definition, one that uses different words than the words actually used in the claim language itself. Notably, moreover, the only definition that Premier offers in place of the district court’s definition is the limiting definition that we have already rejected.
Accordingly, we reject the challenge to the district court’s construction of “directly connected.” Once the district court has an opportunity to consider more evidence and argument about the meaning of this and other disputed claim terms, it may change its ruling. Sofamor Danek Group, Inc. v. DePuy-Motech, Inc., 74 F.3d 1216, 1221 (Fed.Cir.1996). And, of course, we have that same freedom on appeal. Compare CVI/Beta Ventures, Inc. v. Tura LP, 112 F.3d 1146 (Fed.Cir.1997) (interpreting claims one way on appeal from a preliminary injunction based on the evidence presented) with CVI/Beta Ventures, Inc. v. Tura LP, 120 F.3d 1260 (Fed.Cir.1997) (interpreting the same claims a different way on appeal from a final judgment based on the same evidence presented at the preliminary injunction stage). Given our disposition of this issue, we need not address Advanced’s additional argument concerning the doctrine of claim differentiation.
C
The final infringement ruling that Premier challenges deals with the scope of the limitation “data control means,” which the parties agree and the district court determined is a means-plus-function limitation governed by the strictures of Section 112 116. As discussed above, the district court identified the function performed by this limitation as the very function claimed in claim 1; namely, the function of “retrieving digitized audio and digitized video material corresponding to each of said user inputs.” ’908 patent, col. 8, II. 24-25. Further, the court identified the corresponding structures that perform this function by listing some of those structures and then identifying the rest with a pinpoint cite to the portion of the specification that lists them.
On appeal, Premier asserts the court did not identify all the structures needed to perform the claimed function. Specifically, argues Premier, the district court failed to include among the corresponding structures the software that “poll[s] the preview station interface means”' — a “structure” that is absent from the accused Interactive Network. (Appellant’s Br. at 53) (emphasis added). Premier further argues that the corresponding structure must also include application software that “performs a database search,” and that the accused Interactive Network lacks such structure.
At this stage, however, without the benefit of a Markman hearing and without a more fully developed record, we are not prepared to identify the corresponding structure necessary to perform the claimed function. Accordingly, we do not decide whether the proper claim construction requires operating software that “polls” and application software that “performs a database search.”
Even if Premier were correct that operating software that polls and application software that performs a database search *982were required corresponding structure, the district court did not clearly err in finding Advanced’s likelihood of success of proving infringement to support this aspect of the preliminary injunction. Advanced CEO Marco Scibora testified that Premier’s operating system software that interrupted was equivalent to operating system software that could “poll”: “It is also my opinion that polling and interrupts ... are equivalent mechanisms for retrieving information from the user interface devices.” (Scibora Aff. H 10) (J.A. 1058.) Similarly, Scibora testified that Premier’s application software that read in parameters from an .ini text file was equivalent to the application software of the patent that “perform[ed] a database search.” (See Scibora Aff. 117) (J.A. 1058).
We see no basis for reversing the prehminary injunction on this ground.
V
Premier also argues that no presumption of irreparable harm should arise in this case, regardless of the likelihood-of-success analysis. Further, it asserts that even if the presumption does apply, Premier has rebutted it. Last, it challenges the district court’s balancing of the harms as erroneous. We address these three issues seriatim.
A
As the district court correctly stated, a presumption of irreparable harm arises when the party seeking a preliminary injunction makes a “clear showing” of likely infringement and that the asserted patent claims will withstand the accused infringer’s challenges to those claims’ validity and enforceability. Amazon.com, 239 F.3d at 1350; Reebok, 32 F.3d at 1556. Because we have already determined that Advanced’s conelusory assertions did not establish a likelihood of success as to versions 1.0 and 3.0, we must likewise determine that no presumption of irreparable harm applies to these two accused devices and that no preliminary injunction will enjoin them. Our remaining analysis on this point deals only with version 2.0.
Premier necessarily asserts that no presumption should arise as to version 2.0 either, even assuming the district court correctly analyzed the likelihood-of-success prong of the Rule 65 analysis. This is so, argues Premier, because the district court did not find the requisite “clear showing” of infringement. (Appellant’s Br. at 57.) And nor did Advanced make a clear showing of validity, claims Premier, for the ’908 patent has not yet undergone and survived the rigors of a trial on its validity, and it only issued a year before the start of this litigation.
We reject this pair of arguments. First, the district court plainly found the requisite “clear showing” of infringement, even stating in its analysis that only a “clear showing” could trigger the presumption. Premier seems to do nothing more than quibble over the court’s failure to repeat the words “clear showing” in the next sentence, when it applied the otherwise correctly stated rule to this case. Our court has admonished lawyers who pluck quotes from cases without giving effect to the case’s “actual holding.” Cf. Illinois Tool Works, Inc. v. Grip-Pak, Inc., 906 F.2d 679, 684, n. 5 (Fed. Cir.1990) (“Further, it is inappropriate to quote mere language from a court opinion, while disregarding the actual holding of the court and the factual pattern which gave rise to the quoted language.”).
Nor does Premier’s validity argument bar application of the presumption of irreparable harm. Simply put, Advanced needs to only defend against any challenges to the validity of the ’908 patent; it *983need not make an independent “clear showing” of patent validity when no such challenge arises in the first place, given that all patents are presumed valid and that the burden of proving invalidity always rests with the party asserting as such, the accused infringer. Canon Comp. Sys., Inc. v. Nu-Kote Int'l Inc., 134 F.3d 1085, 1088 (Fed.Cir.1998). Accordingly, as we have explained:
It is true that [plaintiff] Canon carries the burden of establishing a likelihood of success on the validity issue and thus must show that Nu-Kote will not likely prove that the patent is invalid. [Citations omitted] However, a patent is presumed valid, and this presumption exists at every stage of the litigation. [Citations omitted] Ultimately, at trial on the merits, Canon need only submit sufficient evidence to rebut any proof of invalidity offered by [accused infringer] Nu-Kote. Thus, where the challenger fails to identify any persuasive evidence of invalidity, the very existence of the patent satisfies the patentee’s burden on the validity issue.
Id. (emphasis added). Thus, because Premier — the party with the ultimate burden of proving invalidity — has not identified a single prior art reference that might diminish an otherwise unchallenged “clear showing” of patent validity, the presumption of irreparable harm likewise remains undiminished as to version 2.0 of the Interactive Network.
In addition, the two asserted grounds for Premier’s invalidity attack hardly amount to the requisite “persuasive evidence of invalidity.” Rather, Premier appears to have plucked these twin rationales from Amazon.com, Inc., where we said that a patentee (not an accused infringer) could perhaps rebuff a challenge to validity at the preliminary injunction stage by, “for example, ... showing that the patent in suit had successfully withstood previous validity challenges” or the relevant industry had long acquiesced in the patent’s validity. 239 F.3d at 1359. But the converse does not seem equally true. Otherwise, an accused infringer could undermine the “clear showing” needed to trigger the presumption simply because of the fortuity of litigation and the tuning of the Patent and Trademark Office, without any reliance at all on prior art or other more common attacks on validity. Without expressing any opinion about whether these asserted grounds could be relevant in a different case, we reject Premier’s strained contention here.
B
Next, Premier asserts that it has rebutted the presumption of irreparable harm even if that presumption does apply here. Specifically, Premier notes that Advanced waited 12 months from the time the ’908 patent issued to sue Advanced in federal court and seek a preliminary injunction, thereby undercutting the notion that Advanced needs emergency relief to prevent irreparable harm. (Appellant’s Br. at 59.) Citing High Tech. Med. Instr., Inc. v. New Image Indus., Inc., 49 F.3d 1551 (Fed.Cir. 1995), Premier asserts that, in addition, Advanced also “repeatedly” offered to license the ’908 patent to Premier during the parties’ pre-service negotiations. Thus, goes the reasoning, Advanced plainly showed itself willing to accept money damages in lieu of whatever emergency injunctive relief it now seeks.
This chorus of arguments has a much more plausible ring and further adds to the conclusion that Advanced merits no prehminary injunction as to versions 1.0 and 3.0 of the accused Interactive Network. But again, these arguments simply fall short of rendering the district court’s *984conclusion of an abuse of discretion as to version 2.0. After all, delay — the first cited ground for rebutting the presumption— constitutes but a single factor in the irreparable-harm balancing inquiry. See, e.g., High Tech. Med., 49 F.3d at 1557 (noting that “the period of delay may not have been enough, standing alone, to demonstrate the absence of irreparable harm”). Moreover, we excuse delayed requests for Rule 65 relief when, as in this case, the movant has offered a “good explanation” for that delay. Id. Here, Advanced has two: the ongoing delay cultivated by Premier’s on-again, off-again discussions with Advanced about whether the parties would settle the matter; and the fact that Advanced only learned for certain of the Interactive Network’s alleged infringement in May 2001, a month before Advanced filed suit and only three-and-half months before its complaint was served.
The licensing argument is also persuasive; but again, it simply does not do enough, either by itself or coupled with the “delay” argument, to overcome the presumption of irreparable harm. In High Tech. Med., by contrast, the court found the presumption rebutted when the trial court erroneously determined that the accused dental-camera device could be physically altered so as to fall within the scope of the asserted patent claims. 49 F.3d at 1555-56. Accordingly, upon vacating the district court’s likelihood-of-infringement analysis, we also vitiated its finding that a presumption of irreparable harm applied. Id. at 1556. Further, we reasoned, the patentee did not actually practice or commercially exploit his invention, meaning he would suffer no loss of goodwill or lost sales if no preliminary injunction issued. Id. at 1556-57. And the patentee expressed a willingness to license the patent, not to mention that he had waited “almost 17 months after the issuance of the re-examination certificate” before suing the accused infringer. Id. at 1557. The totality of these circumstances, along with the absence of any showing that money damages could not alone remedy the patentee’s harm, renders the district court’s conclusion of irreparable harm an abuse. Id.
In this case, on the other hand, we have found no reversible error with the district court’s infringement and likelihood-of-success analysis as to version 2.0, meaning the presumption still holds for that particular device. Contrast id. at 1556. Also, unlike the patentee in High Tech. Med., Advanced markets and commercially exploits its patented invention; and thus, it would indeed risk losing sales and market share, among other economic harms, should no preliminary injunction issue. Contrast 49 F.3d at 1556-57. And the more accurate gauge of the delay here is three or four months, not the “12 months” that Premier emphasizes. Thus, this argument is also unavailing.
C
Finally, Premier also contests the district court’s balancing of the harms. In so doing, Premier asserts that a preliminary injunction will effectively put it out of business and will “probably” force it to lay off all 150 of its employees. (Appellee’s Br. at 61.) In defense of the district court’s analysis, Advanced counters that this going-out-of-business assertion is inconsistent with the inference that no one at Premier thought enough of the pending complaint as to even timely respond to it. And one would expect, adds Advanced, that a party facing extinction via a lawsuit would aggressively defend against that suit, let alone timely answer it. (Appellee’s Br. at 61.)
This factor again solidifies the contrary results for the Interactive Network versions 1.0 and 3.0 on the one hand and *985version 2.0 on the other. As to versions 1.0 and 3.0, we cannot discount Premier’s sworn assertions that a preliminary injunction could well cause the company to dissolve (see J.A. 445), a relevant consideration. See Bell & Howell Doc. Mgmt. Prods. Co. v. Altek Sys., 132 F.3d 701, 708 (Fed.Cir.1997). As to version 2.0, the district court did not abuse its discretion because, in addition to the presumption of irreparable harm, Advanced also has allegedly lost other business opportunities with other large potential buyers, including Wal-Mart and Kmart. (See J.A. 445.) Further, as the presumption of irreparable harm implicitly recognizes, the patentee Advanced will continue to lose business opportunities unless the putative infringement immediately ceases. And of course, a court ought not reward a business built on an infringing product by allowing that infringement to continue in the face of a proper application for preliminary relief. Given the presumption, Advanced’s lost business, goodwill and sales opportunities and the possibility that Premier has already benefited at Advanced’s expense, the court had sufficient reasons to find in Advanced’s favor here as well. At the very least, its finding does not rise to the level of clear error.
VI
For the reasons stated above, we reverse the district court’s entry of a partial default judgment and the preliminary injunction insofar as it would enjoin versions 1.0 and 3.0 of the accused Interactive Network. We affirm the preliminary injunction as to version 2.0 of the accused Interactive Network.
COSTS
Each party will bear its own costs.