EireOg Innovations Ltd. v. Cisco Systems Inc., No. 2:24-cv-00224 (E.D. Tex. Apr. 9, 2026)

Before the Court is Defendants Cisco Systems, Inc., Fortinet, Inc., International Business Machines Corporation, and Palo Alto Networks, Inc.’s Motion to Exclude Improper Opinions of Dr. Michael C. Brogioli. Dkt. No. 204. The motion is fully briefed. See Dkt. Nos. 259, 294, 323. In the Motion, Defendants seek to exclude Dr. Brogioli’s opinions concerning benefits of the ’777 and Interrupt Patents, both under Daubert and as untimely. They also seek to exclude his opinions on infringement as untimely. For the reasons set forth below, the Motion is DENIED.

I. LEGAL STANDARD

An expert witness may provide opinion testimony if “(a) the expert’s scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue; (b) the testimony is based on sufficient facts or data; (c) the testimony is the product of reliable principles and methods; and (d) the expert has reliably applied the principles and methods to the facts of the case.” Fed. R. Evid. 702.

Rule 702 requires trial courts to make a preliminary determination, when requested, as to whether the requirements of the rule are satisfied regarding a particular expert’s proposed testimony. See Kumho Tire Co. v. Carmichael, 526 U.S. 137, 149 (1999); Daubert, 509 U.S. at 592–93 (1993). Such courts are given broad discretion in making Rule 702 admissibility2 determinations. Kumho Tire, 526 U.S. at 152 (“a judge must have considerable leeway in deciding in a particular case how to go about determining whether particular expert’s testimony is reliable”). Although the Fifth Circuit and other courts have identified various factors that the court may consider in determining whether an expert’s testimony should be admitted, the nature of the factors that are appropriate for the court to consider is dictated by the ultimate inquiry—whether the expert’s testimony is sufficiently reliable and relevant to be helpful to the finder of fact and thus to warrant admission at trial. See United States v. Valencia, 600 F.3d 389, 424 (5th Cir. 2010).

Importantly, in a jury trial setting, the Court’s role under Daubert is not to weigh the expert testimony to the point of supplanting the jury’s fact-finding role; instead, the Court’s role is limited to that of a gatekeeper, ensuring that the evidence in dispute is at least sufficiently reliable and relevant to the issue before the jury that it is appropriate for the jury to consider. See Micro Chem., Inc. v. Lextron, Inc., 317 F.3d 1387, 1391–92 (Fed. Cir. 2003) (applying Fifth Circuit law) (“When, as here, the parties’ experts rely on conflicting sets of facts, it is not the role of the trial court to evaluate the correctness of facts underlying one expert’s testimony.”); Pipitone v. Biomatrix, Inc., 288 F.3d 239, 249–50 (5th Cir. 2002) (“‘The trial court’s role as gatekeeper is not intended to serve as a replacement for the adversary system.’ . . . Thus, while exercising its role as a gatekeeper, a trial court must take care not to transform a Daubert hearing into a trial on the merits” (quoting an Advisory Committee Note to Fed. R. Evid. 702)). As the Supreme Court explained, “vigorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but admissible evidence.” See Mathis v. Exxon Corp., 302 F.3d 448, 461 (5th Cir. 2002).

3II. DISCUSSION

A. The ’777 Patent

i. Reliability of Benefits Opinions

Defendants argue that Dr. Brogioli’s benefits opinions related to the ’777 Patent should be excluded as unreliable. Dkt. No. 204 at 4–7. Defendants argue that the “Streaming Load/Store” document, on which Dr. Brogioli primarily relies, is hearsay that is “undated,” “written by an unknown author,” and EireOg has produced no evidence that it was peer reviewed, published, or otherwise relied upon by any expert. Id. at 4. Defendants argue no expert would reasonably rely on such a document. Id. at 5. Defendants argue the document contains no information about who performed testing, what computer systems were used, the purpose of the testing, or how the testing was performed. Id. at 5. Defendants argue the document does not state whether the non-temporal instructions were used on any accused product or Intel processor, or whether such instructions were used in an accused mode or in a “non-modifying mode.” Id. at 6. Defendants argue that Dr. Brogioli’s endorsement of the methodology used in the study is insufficient because experts cannot render their own opinions reliable by approving of their own methods. Dkt. No. 294 at 1.

EireOg argues that Defendants’ complaints about the Streaming Load/Store document go to the weight rather than admissibility of Dr. Brogioli’s opinions. Dkt. No. 259 at 3. EireOg argues that reliance on hearsay is allowed and that the proper standard is whether experts in Dr. Brogioli’s field would rely on it, and it has proferred sufficient evidence that such experts would. Dkt. No. 259 at 3–4. EireOg cites that the document reflects a study conducted by CERN, a reputable organization in the field, and was found on their website. Id. at 4. EireOg also cites Dr. Brogioli’s explanation that he would have used the same methodology as the document explained that CERN used. Id. (citing Dkt. No. 259-3 at 370:1–372:22). EireOg argues that the document contains sufficient identifying information, including that it was produced by CERN, when it was4 published, and that a CERN Research Fellow who previously worked at Intel conducted the study. Id. EireOg argues that Dr. Brogioli explains why the study is relevant, and that the “non-temporal instructions”—and not the processor, computer systems, or pointers—provide the benefits of the patented invention. Id. at 5–6. EireOg also argues that the document provides information on the instructions used in the study, snippets of the code used, explanations of what the code is showing and what it is used for, and the references to “icc” in the code are references to Intel’s C compiler. Id. at 6–7 (first citing Dkt. No. 204-9 at 67733–38; and then citing Dkt. No. 259-3 at 281:1–20).

Having considered the briefing, the cited portions of Dr. Brogioli’s opening and supplemental reports, and the Streaming Load/Store document, the Court finds Dr. Brogioli’s benefits opinions are sufficiently reliable to survive this Daubert challenge. Experts are allowed to and routinely do rely on inadmissible hearsay, see Fed. R. Evid. 703; United States v. Portillo, 969 F.3d 144, 170 (5th Cir. 2020), and EireOg has proffered evidence to show that experts in Dr. Brogioli’s field would rely on it. Notably, Defendants drop the argument that the study methods were not disclosed in their reply. Dkt. No. 294 at 1–3. Additionally, Dr. Brogioli does not endorse methods he used, as Defendants seemingly argue, id. at 1, but methods used by the author of the study, as a reason he reasonably relied upon the study. Dkt. No. 259-3 at 372:15–22. Dr. Brogioli also sufficiently explains why he believes the 42% figure he used, based primarily on the Streaming Load/Store document ratios, reflects the benefits of only the patented features. Dkt. No. 204-8 ¶¶ 97–107. Defendants’ other arguments are unpersuasive. Accordingly, the Court finds that the Motion is denied to the extent it seeks exclusion of Dr. Brogioli’s technical benefits analysis concerning the ’777 Patent.

ii. Timeliness of Infringement Opinions

Defendants argue that Dr. Brogioli’s infringement opinions on the “decoration operation” limitation of Claim 16 of the ’777 Patent are untimely. They allege that EireOg contended that this5 limitation was met by “the memory controller perform[ing] the decoration operation . . . on a location in memory designated by the target address. For example, when an instruction in one processor core requests to write to memory that is connected to the memory controller on a different processor core.” Dkt. No. 204 at 8 (quoting Dkt. No. 204-11 at 14). Defendants allege that Dr. Brogioli’s opinions now focus on a partial write to memory with a “byte enable” “decoration” that “results in a different operand size,” which is inconsistent with the theory disclosed in EireOg’s infringement contentions. Id. (quoting Dkt. No. 204-7 Ex. C ¶¶ 104, 144). While they concede that the contentions mention the VEX prefix, they contest that the contentions reference operand size and that the cited Intel manual discusses any prefix. Dkt. No. 294 at 3.

EireOg argues that its contentions were sufficient to put Defendants on notice of its infringement theory with respect to the “decoration operation” limitation. Dkt. No. 259 at 10. EireOg alleges that its contentions disclosed theories as to both REX and VEX prefixes, and that the Intel manual cited in its contentions describes both prefixes in the same way that Dr. Brogioli uses them. Dkt. No. 259 at 10–11 (citing Dkt. No. 259-4 at 4–5, 10–11, 13–14). EireOg alleges the Intel manual also explains that these prefixes can change the operand size. Id. (citing Dkt. No. 259-4 at 11). EireOg argues that Defendants improperly focus on an example disclosed in its contentions, to which EireOg’s theory is not limited. Dkt. No. 323 at 2.

The Court agrees that Defendants improperly focus on a non-limiting example provided in EireOg’s contentions to argue that Dr. Brogioli’s infringement read on the “decoration operation” limitation is untimely. Having reviewed the briefing, the cited portions of the contentions, and Dr. Brogioli’s report, the Court finds that Dr. Brogioli’s theory is consistent with the theories disclosed in EireOg’s contentions. The contentions do allege that the operation instruction “can be decorated with [either] the REX or VEX prefix to access additional registers,” and that “use of the REX.W6 prefix promotes operation to 64 bits” rather than “default operation size [of] 32 bits.” Dkt. No. 259-4 at 10–11 (emphasis added). Accordingly, the Court finds that Defendants’ Motion is denied to the extent it seeks to exclude Dr. Brogioli’s infringement opinions related to the “decoration operation” limitation of the ’777 Patent as untimely.

B. The Interrupt Patents¹

iii. Reliability of Benefits Opinions

Defendants next argue that Dr. Brogioli’s opinions on the benefits of the Interrupt Patents should be excluded as unreliable. Dkt. No. 204 at 11–13. Defendants’ argument is based on their understanding that Tu, a 2015 journal article on direct interrupt delivery (“DID”), “does not reflect the operation of the accused products or the claimed methods of the” Interrupt Patents. Id. at 11. Defendants argue that, despite Dr. Brogioli’s theory that DID is equivalent to the “posted interrupt” feature, Tu specifically distinguishes DID from Intel’s APICv technology, which includes that feature. Id. at 11–12. Defendants argue that Dr. Brogioli fails to explain how DID reflects the claimed “interrupt packages,” “partition identifiers” and use of a “processor,” or how DID is practiced by the accused products. Id. at 12–13. Defendants also argue that Dr. Brogioli’s apportionment analysis is without support because he arbitrarily chooses Tu’s 9.2% benefit for “application-level CPU saving.” Id. at 13.

EireOg concedes that Tu distinguishes DID from the APICv interrupts but argues that Tu still bears relevance because APICv/VT-d are more advanced, vendor-specific architectures that achieve better results than DID. Dkt. No. 259 at 12–13 (citing Dkt. No. 204-16 at 67699, 67701). In other words, EireOg argues, Tu provides “a conservative estimate” of the benefits of “bypassing7 the hypervisor,” which is what Dr. Brogioli explains is the claimed improvement of the Interrupt Patents. Id. at 13 (first citing Dkt. No. 204-16 at 67699; and then citing Dkt. No. 259-3 at 176:21– 177:23). As to the 9.2% benefits figure, EireOg argues that Tu provides improvement figures based on different packets per second rates ranging from 100–600K, and Dr. Brogioli explained that he chose the 600K figure as a “conservative estimate of the benefits conferred by the Interrupt Patents” because “the Accused Products use substantially higher rates of packets per second than 600K.” Dkt. No. 259 at 13–14 (citing Dkt. No. 204-8 ¶¶ 115–22).

Having considered the briefing, the Tu article, and the cited portions of Dr. Brogioli’s opening report, the Court finds Dr. Brogioli’s reliance on Tu and the technical benefits analysis related to the Interrupt Patents to be sufficiently reliable. While DID is distinguished from the APICv architecture, Tu does explain that DID allows software-based “hypervisor bypassing,” and, as a result, that “DID can achieve the same . . . goals of minimizing the VM exits overheads . . . without requiring advanced vendor-specific hardware support,” like Intel’s APICv. Dkt. No. 204-16 at 67699, 67701. Dr. Brogioli further ties DID to the accused products and apportions to what he contends reflects the benefits of only the patented invention. Dkt. No. 204-8 ¶¶ 108–22. Nothing further is required, and Defendants’ other arguments are unpersuasive. Accordingly, the motion is denied to the extent it seeks exclusion of Dr. Brogioli’s benefits opinions related to the Interrupt Patents.

iv. Timeliness of Doctrine of Equivalents Opinions

Defendants argue that Dr. Brogioli’s opinions on infringement of the Interrupt Patents under the doctrine of equivalents (“DOE”) are untimely. Dkt. No. 204 at 13–15. Specifically, Defendants contend that EireOg’s contentions only included a DOE theory that “checking the PIR bits of the currently running VM’s Posted Interrupt Descriptor is equivalent to the current-VMCS register,” to satisfy the “processing the interrupt package . . . by comparing the . . . partition8 identifier against at least a . . . stored partition identifier” limitation. Id. at 14. However, Defendants allege that Dr. Brogioli’s opinions include three DOE theories: (1) that the “interrupt package” includes a “partition identifier identifying a partition at the processor,” (2) processing an interrupt “by comparing the . . . partition identifier against at least . . . a stored partition,” and (3) processing an interrupt using information “retrieved from special purpose registers at the processor.” Id. Defendants concede that the second theory was sufficiently disclosed in EireOg’s contentions. Id.

EireOg disagrees with Defendants’ characterization of its contentions. Dkt. No. 259 at 15. EireOg alleges it disclosed that “checking the PIR bits of the currently running VM’s Posted Interrupt Descriptor is equivalent and interchangeable with comparing the partition identifier associated with the incoming interrupt to the current VMCS register.” Id. (emphasis in original). EireOg argues this discloses that PIR bits are equivalent to partition identifiers. Id. EireOg argues that Dr. Brogioli’s opinions that “the use of PIR bits in a PID to convey partition identity is an insubstantial difference from the patent’s explicit identifier,” and “when VT-d set[s] a bit in the [PIR] field of a PID, it is signaling that an interrupt is pending for the partition tied to that PID,” are thus entirely consistent with this disclosure and not untimely. Id.

Defendants improperly focus on the fact that the DOE theory was within the corresponding cell of the “processing the interrupt package . . .” limitation of EireOg’s infringement contentions, and not in corresponding cells related to the two other limitations. Dkt. No. 294 at 4 (citing Dkt. No. 204-21 at 36; Dkt. No. 204-22 at 26). However, the statement in the contentions that “checking the PIR bits . . . is equivalent [to] . . . comparing the partition identifier” is sufficient to give notice of Dr. Brogioli’s theory that the use of PIR bits to convey partition identity is equivalent to comparison involving the claimed partition identifier. Dkt. No. 204-21 at 36; Dkt. No. 204-22 at 26; Dkt. No. 204-7 Ex. D ¶¶ 143–46; Dkt. No. 204-7 Ex. E ¶¶ 97–100. The complete contentions’9 statement, that “checking the PIR bits of the currently running VM’s [PID] is equivalent [to] . . . comparing the partition identifier associated with the incoming interrupt to the current VMCS register,” is also consistent with Dr. Brogioli’s theory that “the role of the ‘stored partition identifier’ is satisfied by the processor’s use of the VMCS-to-PIC association” because “[w]hen VT-d set a bit in the [PIR] field of a PID, it is signaling an interrupt is pending for the partition tied to that PID.” Dkt. No. 204-21 at 36; Dkt. No. 204-22 at 26; Dkt. No. 204-7 Ex. C ¶¶ 228–29; Dkt. No. Ex. E ¶¶ 140–41. Accordingly, the motion is denied to the extent it seeks to exclude Dr. Brogioli’s DOE opinions related to the Interrupt Patents as untimely.

C. Timeliness of Benefits Theories

Defendants argue that Dr. Brogioli’s benefits opinions related to both sets of patents are untimely because EireOg did not disclose such theories during discovery. Dkt. No. 204 at 7–8, 13; Dkt. No. 294 at 5. Defendants allege they requested in an interrogatory that EireOg “identify and describe . . . any performance benefits.” Dkt. No. 204 at 7. Defendants allege that EireOg responded that it may rely on unidentified “improvements in the accused products’ functionality,” but never provided any information. Dkt. No. 294 at 5. Defendants allege that EireOg cited the disputed Streaming Load/Store and Tu documents, along with 150 others, on the last day of discovery, in response to a different interrogatory. Id. Defendants argue this disclosure was insufficient to provide notice of Dr. Brogioli’s benefits theories, and that they were prejudiced by their inability to conduct discovery on such theories. Id.; Dkt. No. 204 at 7.

EireOg argues that it was not required to disclose Dr. Brogioli’s technical benefits analysis during fact discovery, and its disclosure of the factual basis for Dr. Brogioli’s opinions during fact discovery was sufficient. Dkt. No. 259 at 7–8. EireOg alleges that it properly answered Defendants’ interrogatory by responding that the benefits “will likely encompass improvements in the accused products’ functionality and/or cost savings related to increasing processing10 efficiencies,” and incorporated by reference its response to Interrogatory No. 1, which included the Stream Load/Store and Tu documents. Id. at 8. EireOg also argues that even if its disclosure was late, Defendants have suffered no prejudice because they deposed Dr. Brogioli at length regarding his benefits opinions, and Defendants’ expert, Dr. Witchel, responded to Dr. Brogioli’s opinions in detail in his rebuttal report. Id. at 9–10.

Having reviewed the briefing, EireOg’s interrogatory responses, and the cited portions of Dr. Brogioli’s reports, the Court finds that EireOg’s theory was sufficiently disclosed. Dkt. No. 204-1 at 8–10. EireOg responded to Interrogatory No. 1, that, while an expert has not opined yet, it plans to evaluate “technical benefits . . . via comparison with [NIAs] . . . which will likely encompass improvements in the . . . functionality . . . related to increased processing efficiencies.” Id. at 9. EireOg referenced “its response to Defendant’s Interrogatory No. 1.” Id. at 9–10. In that . response—after a similar statement identifying that the benefits theory would likely relate to increased processing efficiencies—Eirog cited the Streaming Load/Store document and Tu article. Dkt. No. 259 at 8; Dkt. No. 204-3 at 20. Dr. Brogioli’s benefits opinions are consistent with that disclosure. Accordingly, the motion is denied insofar as it seeks exclusion of Dr. Brogioli’s benefits opinions as untimely.

III. CONCLUSION

For the reasons provided above, the Court finds that the Motion (Dkt. No. 204) should be and hereby is DENIED.

SIGNED this 3rd day of January, 2012.

SIGNED this 9th day of April, 2026.

ROY S. PAYNE

UNITED STATES MAGISTRATE JUDGE