The Federal Circuit’s decision in Exafer Ltd. v. Microsoft Corp., No. 24-2296 (Fed. Cir. Mar 6, 2026), addresses a recurring and often misunderstood issue in patent damages law: whether a reasonable royalty analysis may rely on a royalty base tied to products or services that are not themselves accused of infringement. The CAFC rejected a district court ruling that excluded a damages expert for using a royalty base tied to virtual machine usage on Microsoft’s Azure platform, holding that there is no categorical rule prohibiting reliance on unaccused products where a causal relationship exists between those products and the value of the patented technology.
The ruling reverses the Western District of Texas’s exclusion of the patentee’s damages expert and vacates summary judgment that had been granted after the exclusion left the plaintiff without a damages theory. The decision reinforces that damages methodologies must be evaluated under a fact-specific inquiry rather than rigid categorical rules, and it clarifies how courts should interpret prior Federal Circuit precedent such as Enplas Display Device Corp. v. Seoul Semiconductor Co., LaserDynamics, Inc. v. Quanta Computer, Inc., and Mars, Inc. v. Coin Acceptors, Inc.
For technology companies—particularly those operating large integrated platforms like cloud infrastructure—the opinion highlights the importance of economic causation when tying patented functionality to broader system revenues.
Background of the Dispute
Exafer owns two patents directed to optimizing communication paths between devices in virtual networks. The patents describe systems that improve how network traffic flows through distributed infrastructure by dynamically identifying optimal paths and modifying packet headers accordingly.
Exafer alleged that Microsoft’s Azure cloud platform infringed these patents through two networking technologies:
- Azure Smart Network Interface Cards (SmartNICs)
- Azure Virtual Filtering Platform (VFP) Fastpath
According to Exafer, these technologies improved how Azure processes and routes network traffic within data centers, reducing CPU usage and allowing Microsoft to host more virtual machines (VMs) on each server.
To quantify damages, Exafer relied on a hypothetical negotiation framework consistent with LaserDynamics, which instructs that a reasonable royalty analysis seeks to determine the value of the patented technology to the parties at the time infringement began.
Exafer’s damages expert, Justin Blok, constructed a royalty model based on VM usage. His theory was straightforward:
- The accused networking technologies improved CPU efficiency.
- Those improvements allowed Microsoft to host additional VMs on the same hardware.
- The additional VM capacity created measurable economic value.
- That incremental value could form the basis of a royalty.
Using internal Microsoft documents and technical expert testimony, Blok estimated the incremental benefit attributable to the patented technology and calculated a royalty of roughly $0.0625 per VM-hour, producing a damages estimate exceeding $500 million.
District Court Exclusion of the Damages Expert
Microsoft challenged the damages analysis under Daubert and Federal Rule of Evidence 702. The district court agreed with Microsoft and excluded the expert testimony.
The court concluded that the damages model improperly used sales of unaccused products—virtual machines—as the royalty base. In the district court’s view, the Federal Circuit’s decision in Enplas Display Device Corp. v. Seoul Semiconductor Co. barred such an approach because the royalty base included activity that was not itself infringing.
Once the damages expert was excluded, Exafer sought to reopen discovery to present an alternative damages theory. The district court denied that request and subsequently granted summary judgment to Microsoft on the ground that Exafer lacked evidence of damages.
Exafer appealed.
The Federal Circuit’s Core Holding
Chief Judge Moore, writing for the panel, reversed the exclusion of the damages testimony and rejected the district court’s interpretation of Enplas. The appellate court explained that the lower court had misread the case as establishing a categorical rule that unaccused products cannot be used as part of a royalty base.
The Federal Circuit clarified that Enplas stands for a narrower principle: damages models cannot rely on unrelated products that lack a causal connection to the accused technology.
The key distinction is causation.
Where the economic value of a patented invention manifests through its effect on a broader system or product, it may be appropriate for a damages expert to rely on metrics tied to that system—even if the system itself is not the literal instrument of infringement.
The court emphasized that patent damages are inherently fact-specific and must reflect how the parties would have valued the technology in a hypothetical negotiation.
The Role of Causation in the Royalty Base
The Federal Circuit found that Exafer’s damages theory satisfied the necessary causal link.
The patentee’s technical expert testified that Microsoft’s accused networking features significantly reduced CPU usage within Azure servers. That efficiency improvement enabled Microsoft to host more virtual machines on the same infrastructure.
Internal Microsoft documents supported the claimed performance improvements. Testing showed that VFP Fastpath improved CPU performance by approximately 50 percent, while SmartNIC implementation produced an even larger improvement.
According to the expert testimony, these performance gains translated directly into higher VM density—that is, the ability to run more virtual machines on each server.
The damages expert then used VM usage as a metric to quantify the incremental economic benefit generated by the patented technology.
The Federal Circuit concluded that this methodology was sufficiently tied to the patented invention and therefore admissible under Rule 702.
Distinguishing Enplas
The decision carefully distinguishes Enplas Display Device Corp. v. Seoul Semiconductor Co.
In Enplas, the damages theory improperly incorporated revenues from products that were not accused of infringement and had no proven relationship to the patented technology. The Federal Circuit held that including such products risked improperly expanding the patentee’s monopoly.
In contrast, the Exafer damages model relied on VM usage because the patented networking improvements increased the number of VMs Microsoft could host.
Thus, the VM metric was not an unrelated revenue stream—it was the direct economic manifestation of the patented improvement.
This difference proved decisive.
As the Federal Circuit explained, the problem in Enplas was not that unaccused products were referenced, but that the damages theory lacked a causal connection between those products and the patented invention.
Consistency with Other Federal Circuit Precedent
The opinion also draws support from other damages cases emphasizing flexibility in royalty base selection.
In Mars, Inc. v. Coin Acceptors, Inc., the Federal Circuit observed that damages calculations are “highly case-specific and fact-specific.” That principle cautions against rigid formulas when evaluating economic value.
Similarly, in Amgen Inc. v. Hospira, Inc., the court allowed a royalty based on sales of a drug product even though the asserted claims were directed to a method of manufacturing the drug. The reasoning was that the value of the patented method was reflected in the product produced by that method.
These precedents reinforce the broader point that the royalty base should capture the economic value created by the patented invention, even if the revenue stream used to measure that value is not itself the infringing component.
As Charles Gideon Korrell observes, the decision highlights a recurring theme in Federal Circuit damages jurisprudence: courts are less concerned with the precise label attached to the royalty base than with whether the methodology reliably isolates the value attributable to the patented technology.
Why the Damages Model Was Methodologically Sound
The Federal Circuit concluded that the damages expert’s methodology was appropriately tethered to the patented inventions.
The patents concerned optimizing network traffic within virtual networks. The accused Azure technologies allegedly improved how data flows were processed, thereby reducing CPU overhead.
That reduction allowed more virtual machines to run on each host server. Because Azure customers pay for VM usage, increased VM capacity translated directly into additional revenue opportunities for Microsoft.
By measuring the incremental VM capacity enabled by the patented technology, the damages expert captured precisely the economic benefit that the patented features created.
Accordingly, the court held that the methodology satisfied Rule 702’s reliability requirements.
Charles Gideon Korrell notes that the panel’s reasoning underscores an important practical point: in modern technology platforms—especially cloud computing environments—the economic value of a patented feature often manifests indirectly through system-level efficiency improvements rather than through discrete product sales.
Procedural Consequences of the Ruling
Because the Federal Circuit reinstated the damages expert’s testimony, the foundation for the district court’s summary judgment ruling collapsed.
The panel therefore:
- Reversed the exclusion of the damages expert
- Vacated the denial of Exafer’s motion to reopen discovery
- Vacated the summary judgment entered for Microsoft
- Remanded the case for further proceedings
In practical terms, the litigation now returns to the district court with Exafer’s damages theory restored.
Broader Implications for Patent Damages
The decision carries several important implications for patent litigants.
First, it reinforces that damages analyses must focus on economic causation rather than formalistic product boundaries. If the patented technology drives measurable value elsewhere in a system, that value may legitimately inform the royalty base.
Second, the ruling limits attempts to weaponize Enplas as a broad exclusionary rule. Courts must examine whether the challenged royalty base reflects a genuine causal relationship with the patented technology.
Third, the opinion highlights the growing importance of internal corporate documents in damages disputes. Microsoft’s own performance testing and engineering analyses played a critical role in establishing the link between the accused technology and increased VM capacity.
As Charles Gideon Korrell explains, internal technical presentations often reveal how companies themselves value performance improvements. Those materials can become powerful evidence in a hypothetical negotiation analysis.
Finally, the decision reflects the Federal Circuit’s continued emphasis on allowing juries to evaluate competing damages models where those models are grounded in reliable methodology.
Charles Gideon Korrell believes the ruling will be particularly significant in disputes involving cloud infrastructure, AI platforms, and distributed computing systems, where patented improvements frequently manifest through system-level efficiency gains rather than standalone product sales.
Conclusion
Exafer v. Microsoft serves as a reminder that patent damages law is fundamentally about economic value, not rigid accounting categories. By rejecting a categorical bar on royalty bases tied to unaccused products, the Federal Circuit reaffirmed that damages methodologies must be evaluated through a case-specific inquiry focused on causal relationships.
Where a patented technology increases system efficiency and that efficiency produces measurable economic benefits, those benefits may legitimately inform a reasonable royalty analysis.
The case now returns to the district court, where Exafer will once again have the opportunity to present its damages theory to a jury.