dissenting-in-part.
I agree with the court’s thorough and well-reasoned analysis of the standing question, but respectfully dissent from Section II of the court’s opinion. The specification of U.S. Patent No. 7,756,129 (the “'129 patent”) repeatedly and unambiguously indicates that the term “Media *1351Access (MAC) address” refers to an address that is assigned to a peripheral device by a hub device. The patent neither contemplates nor enables any other method of network communication. The court offers no plausible justification for disregarding the unequivocal disclosure in the specification and dramatically expanding the definition of “Media Access address” to cover any address that uniquely identifies a device on a shared communication system. The decision to jettison the trial court’s claim construction is predicated almost exclusively on a purported industry definition of the term “Media Access Control address,” but that term is found nowhere in the '129 patent.
I.
Over and over again, the '129 patent makes clear that a “Media Access (MÁC) address” is an address assigned to a peripheral device by a hub device. The patent explains that “a single [h]ub device” communicates with multiple peripheral devices, id. col.3 1.28, and “orchestrates all communication in the [Personal Area Network],” id. col.3 1.33. Each peripheral device is assigned a “Media Access (MAC) address,” id. col.3 1.32, by the hub device when it becomes part of the communications network, id. col.ll 11.50-52. The Summary of the Invention describes a single method by which peripheral devices are attached to the network, a method that requires that the hub device assign a Media Access address to each peripheral device. Id. col.l 11.57-61 (explaining that “[t]he unattached peripheral device ... receives a new address from the hub device ... and communicates with the hub device using the new address”); id. col.2 11.8-14 (explaining that the unattached peripheral device “reeeiv[es] a new address from the hub device” and is “attach[ed] to the network using the new address”). The specification goes on to repeatedly describe a system in which the hub device assigns an address to each peripheral device. Id. col.ll 11.2-4 (“[T]he [h]ub ... assigns a MAC address to the [peripheral device].”); id. col.ll 11.50-54 (“When the [h]ub successfully receives the attach-request from the [peripheral device], it assigns a MAC address to the [peripheral device].”); id. col.ll 11.55-60 (“The [h]ub sends the new MAC address in an attach-assignment message to the now-identified [peripheral device].” (diagram reference numbers omitted)); id. col.12 11.31-34 (“The [peripheral device] waits for an attach confirmation from the [h]ub using the new MAC address ... and, upon receiving it, sends a final acknowledgement to the [h]ub.” (diagram reference numbers omitted)). Without exception, the '129 patent uses the term “MAC address” to refer to an address that is generated by the hub device. Nothing in the specification contemplates that a Media Access address will be assigned to a peripheral device at the time it is manufactured. Instead, as the district court correctly concluded, “what is defined and consistently disclosed” in the '129 patent “is a MAC address that originates with the hub device.” Azure Networks, LLC v. CSR PLC, No. 11-CV-0139, slip op. at 2 (E.D.Tex. May 17, 2013).
The specification typically provides “the primary guide to claim interpretation.” ArcelorMittal France v. AK Steel Corp., 700 F.3d 1314, 1320 (Fed.Cir.2012); see also Retractable Techs., Inc. v. Becton, Dickinson & Co., 653 F.3d 1296, 1305 (Fed.Cir.2011) (“It is axiomatic that the claim construction process entails more than viewing the claim language in isolation. Claim language must always be read in view of the written description.”). Here, because the specification repeatedly makes clear that the term “Media Access (MAC) address” refers to an address assigned by a hub device — and discloses no other method for network communication — the court has no warrant to vastly *1352expand the term to cover any address that uniquely identifies a device on a network. See Saffran v. Johnson & Johnson, 712 F.3d 549, 560 (Fed.Cir.2013) (explaining that a claim term should be construed in accordance with its “[e]xtensive, consistent usage in the specification”); Kinetic Concepts, Inc. v. Blue Sky Med. Grp., Inc., 554 F.3d 1010, 1019 (Fed.Cir.2009) (refusing to construe the term “wound” to cover fistu-lae because such a construction would “expand the scope of the claims far beyond anything described in the specification”). “Although the specification need not present every embodiment or permutation of the invention and the claims are not limited to the preferred embodiment of the invention, neither do the claims enlarge what is patented beyond what the inventor has described as the invention.” Netword, LLC v. Centraal Corp., 242 F.3d 1347, 1352 (Fed.Cir.2001) (citation omitted).
n.
Three fundamental errors infect the court’s decision to expand the '129 patent to cover an invention that the patentees neither claimed nor described. First, the court states that “[t]he parties do not meaningfully dispute the ordinary and customary meaning of a ‘Media Access Control’ or ‘Medium Access Control’ address (commonly referred to as a MAC address) to the relevant community at the time of the invention.” Ante at 1348. This is incorrect. The Defendants-Appellees vigorously — and persuasively — contend that Azure Networks, LLC (“Azure”) never established that, at the time of the claimed invention, there was a standard industry definition of the term “MAC address” which necessarily encompassed both addresses assigned locally and at the time of manufacture. To the contrary, Azure, in its opening claim construction brief, “cited and quoted two references that make clear that one of ordinary skill would understand MAC addresses as pre-programmed identifiers, not locally assigned addresses.” Br. of Defendants-Appellees at 39-40.
Azure’s problem is one of proof. If it wanted to enlarge its claims far beyond anything described in the specification, it had the burden of coming forward with evidence demonstrating not only that there was a common industry definition of the term “MAC address,” but that that definition necessarily encompassed both addresses assigned at the time of manufacture and by a local network. The simple fact, however, is that Azure failed to adduce any such evidence. Indeed, its references provide divergent definitions of the term. Azure relies heavily on an Institute of Electrical and Electronics Engineers (“IEEE”) specification which indicates that MAC addresses can be assigned either locally or at the time of manufacture. See J.A. 1604-07. Other references Azure provided to the trial court, however, indicate that a MAC address is typically a 48-bit number assigned to a network interface card at the time of manufacture, not an address assigned to a peripheral device by a local hub device as described in the '129 patent. See Hargrave’s Commo’ns Dictio-naey 313 (2001) (A “MAC address” is “[a] 48-bit number unique to each network interface card (NIC). Generally, the number is programmed into the NIC at the time of manufacture; hence, it is LAN and location independent.” (emphasis added)); Newton’s Telecom Dictionary 450 (2002) (A “MAC address” is a “Medium Access Control Address” or a “MAC name” that “traditionally is in the form of a 48-bit number, formerly known as an EUI-48 (Extended Unique Indentifier-48), which is unique to each LAN (Local Area Network) NIC (Network Interface Card). The MAC address is programmed into the card, usually at the time of manufacture.” (emphasis added)). The fact that Azure’s own references provide inconsistent definitions of the term “MAC address” belies its *1353contention that there was a “standard industry definition,” Br. of Plaintiffs-Appellants at 43, of the term sufficient to overcome its “[ejxtensive, consistent usage,” Saffran, 712 F.3d at 560, in the specification. See Renishaw PLC v. Marposs Societa’ per Azioni, 158 F.3d 1243, 1250 (Fed.Cir.1998) (emphasizing that “a common meaning, such as one expressed in a relevant dictionary, that flies in the face of the patent disclosure is undeserving of fealty”).
Apparently recognizing that Azure’s claim construction argument rests on a very feeble evidentiary foundation, the court creates out of whole cloth its own definition of “Media Access address.” In the court’s view, a Media Access address is anything that identifies a device on a network, and it has ferreted out two dictionary definitions that it believes support its definition of the term. See ante at 1348. These dictionary definitions are not part of the record, however, and the Defendants-Appellees have had no opportunity to challenge them. Even more fundamentally, the one cited definition that could even arguably support this court’s exceedingly broad claim construction — the one stating that a “Media Access Control address” is “a hardware address which uniquely identifies each physical connection,” ante at 1348 — pertains to a term found nowhere in the '129 patent.
Language matters. The definition upon which the court’s claim construction is predicated is a definition of a “Media Access Control address,” not a “Media Access address,” the term used in the '129 patent. See '129 patent col.3 11.31-32 (“Each device is identified by a Media Access (MAC) address.”). The fact that the patentees dropped the word “control” from their definition of the acronym “MAC” is highly significant. It indicates that they were not relying on any purported industry definition of the acronym, but were instead using the term “Media Access (MAC) address” consistently with the disclosure in the specification to refer to an address assigned to a peripheral device by a hub device.
Indeed, any doubt as to whether the meaning of the term “MAC address” as used in the '129 patent was intended to be different from any purported industry understanding of the term is resolved by reference to documents the inventors themselves provided. The '129 patent was developed in the course of the inventors’ work on a project known as “BodyLAN.” On six separate occasions, the BodyLAN specification uses the phrase “Media ACcess,” capitalizing both the “A” and the first “C” in “Access.” See J.A. 1664, 1686, 1691, 1701. The fact that the inventors capitalized the “M,” the “A,” and the “C,” in the term “Media ACcess” indicates that the letter “C” in the acronym “MAC” refers to the second letter in the word “access” and not to the word “control” in a media access control address. See Computer Docking Station Corp. v. Dell, Inc., 519 F.3d 1366, 1374 (Fed.Cir.2008) (emphasizing that “repeated and definitive remarks in the written description” can restrict a particular claim term); Renishaw, 158 F.3d at 1250 (“[Wjhere there are several common meanings for a claim term, the patent disclosure serves to point away from the improper meanings and toward the proper meaning.”). Significantly, the BodyLAN specification, just like the specification of the '129 patent, makes clear that a “Media ACcess” address originates with the hub device. See J.A. 1701.
Finally, even assuming arguendo that the term “MAC address” is ambiguous and could potentially be construed to cover addresses assigned to a device at the time of manufacture, such a construction would render the asserted claims invalid. See Free Motion Fitness, Inc. v. Cybex Int’l., *1354Inc., 423 F.3d 1343, 1349 n. 4 (Fed.Cir.2005) (explaining that a court can “construe claims to preserve their validity when after applying all the available tools of claim construction ... the claim is still ambiguous” (citations and internal quotation marks omitted)). The '129 patent describes a system in which a hub “orchestrates all communication” in the network, '129 patent col.3 1.33, by creating MAC addresses and assigning them to peripheral devices, id. col. 1111.2-4. The patent does not describe any other way for the hub device to perform the attachment and communication steps required by the asserted claims. It contains no disclosure of how the claimed network would function if the hub had to communicate with peripheral devices using unspecified and potentially proprietary addresses which had been assigned to those devices at the time of manufacture. To expand the definition of MAC address to include addresses assigned at the time of manufacture would permit the '129 patent to encompass a network that the inventors neither contemplated nor adequately described. See Nautilus, Inc. v. Biosig Instruments, Inc., - U.S. -, 134 S.Ct. 2120, 2124, 189 L.Ed.2d 37 (2014) (“[A] patent is invalid for indefiniteness if its claims, read in light of the specification delineating the patent, and the prosecution history, fail to inform, with reasonable certainty, those skilled in the art about the scope of the invention.”); MagSil Corp. v. Hitachi Global Storage Techs., Inc., 687 F.3d 1377, 1381 (Fed.Cir.2012) (“The specification must contain sufficient disclosure to enable an ordinarily skilled artisan to make and use the entire scope of the claimed invention at the time of filing.”).
Azure cannot have it both ways. If the term “MAC address” is properly construed to refer only to addresses assigned to peripheral devices by a hub device, then the accused products do not infringe. If, on the other hand, the term is broadly and unreasonably expanded to cover addresses assigned at the time of manufacture, then the asserted claims are invalid. See MagSil, 687 F.3d at 1381 (“[A] patentee chooses broad claim language at the peril of losing -any claim that cannot be enabled across its full scope of coverage.”); Liebel-Flarsheim Co. v. Medrad, Inc., 481 F.3d 1371, 1380 (Fed.Cir.2007) (“The irony of this situation is that [the patentee] successfully pressed to have its claims include a jacketless system, but, having won that battle, it then had to show that such a claim was fully enabled, a challenge it could not meet.”).