Patent Infringement Damages Calculation: A 2026 Guide to Georgia-Pacific Factors in AI & Hardware

1. The Legal Foundation: 35 U.S.C. § 284

The statutory basis for patent damages in the United States is 35 U.S.C. § 284, which states that courts shall award damages “adequate to compensate for the infringement, but in no event less than a reasonable royalty for the use made of the invention by the infringer.”

This language establishes two critical principles. First, damages are compensatory—they aim to restore the patent holder to the position they would have occupied absent the infringement. Second, a reasonable royalty is the statutory floor: even if the patent holder cannot prove any actual lost sales, they are still entitled to compensation.

The statute also grants courts discretion to enhance damages up to three times the compensatory amount in cases of willful infringement, and to award interest and costs. The May 2025 en banc decision in EcoFactor v. Googlefurther clarified how rigorously courts must scrutinize damages evidence under Daubert and Rule 702—raising the bar for expert testimony and underscoring that approximately 40% of damages-focused appeals now result in reversals or remands.

2. Two Frameworks: Lost Profits vs. Reasonable Royalties

U.S. courts recognize two primary methods for calculating patent infringement damages. The choice between them—or the decision to apply both in combination—depends on the patent holder’s market position and the evidence available. With 95–97% of patent cases settling before trial, the framework applied during early valuation often dictates the negotiation range.

The Panduit Test for Lost Profits

To recover lost profits, the patent holder must satisfy all four prongs of the Panduit test, established in Panduit Corp. v. Stahlin Bros. Fibre Works (1978):

1. Demand for the patented product existed during the infringement period.
2. Absence of acceptable non-infringing substitutes in the market.
3. Manufacturing and marketing capability to exploit the demand.
4. Quantifiable amount of profit the patent holder would have earned.

Failing any single prong typically forecloses lost profits recovery, pushing the analysis to the reasonable royalty floor. In practice, the second prong—proving no acceptable substitutes existed—is the most frequently litigated. The median per-case damages award from 2020 to 2024 stood at $1.8 million, though that figure masks enormous variation: the mean jury award in 2024 alone was approximately $83 million.

3. The 15 Georgia-Pacific Factors Explained

The Georgia-Pacific framework, from Georgia-Pacific Corp. v. United States Plywood Corp. (1970), remains the dominant method for determining a reasonable royalty rate. It asks: what would a willing licensor and a willing licensee have agreed to in a hypothetical negotiation at the time infringement began?

The court identified 15 factors to guide this analysis. While all 15 are relevant, courts and practitioners consistently emphasize a subset as most influential in modern technology cases. The Federal Circuit’s 2025 en banc ruling in EcoFactor v. Google tightened requirements for Factors 1 and 2, holding 8–2 that comparable licenses must account for “differences in technologies and economic circumstances” before they can anchor a royalty rate:

The remaining factors (3–7, 10, 12, 14) address the scope of the license, the commercial relationship between the parties, industry profit margins, and expert testimony. While they contribute to the overall analysis, they are less frequently dispositive in reported decisions.

4. Recent High-Stakes Cases (2024–2026)

The 2024–2025 period has produced some of the largest patent damages verdicts in U.S. history, alongside landmark appellate rulings that are reshaping how damages are calculated:

5. Willful Infringement & Treble Damages

Under 35 U.S.C. § 284, courts may enhance damages up to three times the compensatory amount when infringement is found to be willful. The Supreme Court’s 2016 decision in Halo Electronics, Inc. v. Pulse Electronics, Inc. overhauled the standard, replacing the rigid two-part Seagate test with a flexible totality-of-the-circumstances approach.

Recent data reveals how frequently willfulness succeeds at trial: 72% of juries find willful infringement when the question is presented to them. Once willfulness is established, courts enhance damages in 54.8% of cases (40 of 73 decisions studied), with an average multiplier of 2.2x. Full treble damages are awarded in approximately 40% of enhancement cases.

Courts consider:

• Whether the infringer had actual knowledge of the patent and that its conduct constituted infringement.
• The infringer’s subjective good faith, including whether they obtained a competent opinion of counsel.
• Whether the infringer made reasonable efforts to design around the patent after learning of it.
• The closeness of the case—whether infringement and validity were hotly contested issues.

Venue matters significantly for enhancement. The Central District of California grants enhancement in 85.7% of willfulness findings, while the District of Delaware does so in only 9.1% of cases—a disparity that makes forum selection a strategic factor in any willfulness-dependent damages theory.

6. The Apportionment Challenge in Multi-Component Products

One of the most difficult issues in modern patent damages is apportionment: determining what share of a product’s value is attributable to the patented feature versus unpatented components.

Consider a smartphone that practices a single patent covering a noise-cancellation algorithm. The phone contains thousands of other patented and unpatented technologies, from display panels to antenna designs. Awarding royalties based on the entire phone’s sale price would vastly overcompensate the patent holder.

The Federal Circuit has recently refined its apportionment doctrines through several key rulings:

• Caltech v. Broadcom — Supply-Chain Consistency: The Federal Circuit vacated a $1.185 billion award, holding that when a patent holder asserts the same patent against entities at different levels of the supply chain (chipmaker and device assembler), the same royalty rate must apply. A higher rate for the downstream assembler implies double-counting.
• Brumfield v. IBG — Foreign Economic Activity: The Federal Circuit held that foreign economic activity may be included in a reasonable royalty calculation, expanding the potential damages base for patents practiced in products with global revenue streams.
• Vectura v. GSK — Built-in Apportionment: Introduced the “built-in apportionment” doctrine, recognizing that certain royalty structures inherently account for the patented feature’s contribution without requiring a separate apportionment step—particularly in pharmaceutical and medical device contexts.
• Smallest Salable Patent-Practicing Unit (SSPPU): Where possible, the royalty base should be the smallest component that embodies the patented invention, not the entire end product. Established in LaserDynamics v. Quanta (2012) and still foundational.

In AI and semiconductor cases, apportionment is especially complex. A patented neural network architecture may improve inference speed by 15%, but isolating that improvement’s dollar value from the broader system—including software optimizations, memory bandwidth, and cooling design—requires sophisticated economic modeling.

7. How AI Is Changing Damages Analysis in 2026

Artificial intelligence is reshaping patent damages analysis on multiple fronts. In a recent survey, 43% of patent litigation professionals identified AI as the most impactful trend in their practice, with 38% already using data-driven tools for damages calculation and FRAND rate determination.

• Comparable license identification: AI-powered patent analytics platforms can now scan thousands of licensing agreements and court decisions to identify genuinely comparable royalty rates, reducing the subjectivity in Georgia-Pacific Factor 1 and 2 analyses. Platforms such as Lex Machina have expanded their Protege AI Assistant to provide predictive litigation analytics, including venue-specific damages outcomes.
• Market simulation modeling: Machine learning models can simulate “but for” scenarios—what would the market have looked like without the infringement—with greater granularity than traditional econometric methods. These tools are particularly valuable for modeling the complex multi-product markets in cloud computing and semiconductors.
• Daubert compliance: With EcoFactor v. Googleraising the bar for expert testimony reliability, AI tools help damages experts stress-test their methodologies before depositions—flagging comparable license gaps and apportionment weaknesses that could trigger exclusion under heightened Rule 702 scrutiny.
• AI as the subject of patent disputes: A growing number of cases involve patents covering AI algorithms themselves—transformer architectures, attention mechanisms, reinforcement learning techniques. These cases present novel apportionment questions: how do you value a patented training methodology when its impact varies across model sizes and applications?

8. Practical Steps for Patent Holders

Whether you are evaluating the strength of a potential infringement claim or preparing for litigation, the following steps can help position your damages case:

1. Document your licensing history. Existing licenses for the patent-in-suit are the strongest evidence for Georgia-Pacific Factor 1. After EcoFactor, keep detailed records of not just the license terms but also the technological and economic context of each deal.
2. Identify the smallest salable unit. Before trial, determine which component of the accused product practices your patent. This sets the foundation for a defensible royalty base and is critical for surviving Caltech-style challenges on supply-chain consistency.
3. Quantify the patented feature’s contribution. Gather technical evidence (benchmarks, A/B tests, customer surveys) showing the performance improvement or cost savings your invention provides. Consider whether the Vectura built-in apportionment doctrine applies to your royalty structure.
4. Preserve evidence of willfulness. With juries finding willfulness 72% of the time it is presented, and enhancement yielding a 2.2x average multiplier, documenting the infringer’s knowledge can significantly amplify recovery. Choose your venue carefully—enhancement rates range from 9.1% to 85.7% depending on the district.
5. Engage a damages expert early. The complexity of modern damages analysis requires economic expertise. The EcoFactordecision’s tightened Daubert standards mean that expert reports face greater scrutiny than ever—engaging an expert before filing can help shape your litigation strategy and avoid costly exclusions.