Berkeley*IEOR v. Grainger: A Technical Improvement Must Be in the Patent Claim

A software patent may describe a genuine technical advantage and still fail patent-eligibility review if the patent claim does not require the feature that produces that advantage. The Federal Circuit applied that principle on September 21, 2026, in Berkeley*IEOR v. W.W. Grainger Inc.

The court held claims covering computer-based calculations of object-level profitability ineligible under 35 U.S.C. § 101. The patent owner relied on parallel processing as the inventive concept, but the representative claim could also be practiced without parallel processing. Because the claimed invention did not require the asserted technical improvement, the improvement could not supply the “significantly more” needed at step two of the Alice analysis.

The decision is nonprecedential, so it does not create a new binding rule. It is nevertheless a useful warning for software inventors, patent applicants, patent owners, and accused infringers: eligibility analysis follows the claim language, not the best technical story that can be found elsewhere in the patent.

The profitability patents and the district court’s ruling

Berkeley*IEOR owned U.S. Patent Nos. 7,596,521, 7,882,137, and 8,612,316. The patents concerned methods for calculating object-level profitability. Berkeley sued Teradata Operations and several Teradata customers in the Northern District of Illinois, alleging direct and indirect infringement.

For eligibility, the parties treated claim 1 of the ’521 patent as representative. The claim described a process performed in a computer using a relational database management system. Among other steps, the system would prepare information, establish processing rules, calculate marginal profit values and a fully absorbed profit adjustment value, and combine the results into an object-level profitability measure.

The district court found the claim directed to the abstract idea of calculating profitability. It nevertheless found an inventive concept at Alice step two. In the district court’s view, an unconventional combination improved computer operation by allowing faster parallel computations.

That conclusion had a problem. The claim did not require the calculations to occur in parallel. Under the district court’s construction, the relevant “independently calculate” language permitted sequential calculations and, in some circumstances, no calculation for certain objects. Parallel processing appeared as an implementation described in the patent, but it was not a mandatory part of the claim.

The Federal Circuit focused on what the claim required

In its September 21, 2026 opinion, the Federal Circuit agreed that the representative claim was directed to an abstract idea. Read as a whole, it covered mathematical calculations for determining profitability, with a computer used to collect, analyze, and combine information. Performing the calculations faster on a computer did not, by itself, make the idea nonabstract.

The decisive analysis came at step two. The court explained that an alleged inventive concept cannot save a claim when the claim is broad enough to be practiced without that concept. Put differently, a technological improvement cannot make the claim eligible if the claimed invention does not require the steps that produce the improvement.

The parties and the district court agreed that parallel processing was optional under the claim language. Berkeley’s other proposed inventive concept—calculating marginal profit values inside the relational database—ultimately relied on the same asserted parallel-processing benefit. The Federal Circuit therefore found no genuine dispute of material fact that the representative claim lacked an inventive concept.

The court reversed the district court’s summary judgment of eligibility and affirmed judgment for the defendants on the alternative ground of patent ineligibility. Because that ground resolved all asserted counts, the court did not reach Berkeley’s challenges concerning the customer-suit stay, indirect infringement, or infringement under 35 U.S.C. § 271(g).

A strong specification cannot substitute for a claim limitation

A patent specification and its claims do different work. The specification teaches how the invention operates, supplies written-description and enablement support, and may disclose alternatives and embodiments. The claims define the legal boundaries of the invention. Tucker Law’s overview of patent claim construction explains why the wording at the end of a patent can control both validity and infringement disputes.

Berkeley*IEOR does not make the specification unimportant. A carefully drafted specification can identify the technical problem, explain why the solution differs from conventional systems, support meaningful claim limitations, and preserve options for prosecution. But a court generally will not import an optional embodiment into a claim merely because that embodiment supplies the best eligibility argument.

This creates a familiar drafting tension. An applicant often wants broad claims that are difficult for competitors to design around. Broad functional language, however, may omit the very mechanism that distinguishes the invention from an abstract result. The goal is not to add technical detail indiscriminately. It is to identify which features actually produce the asserted improvement and decide deliberately where those features belong in the claim set.

Drafting software claims around the mechanism, not only the result

Software inventions are often first described in terms of a business outcome: faster pricing, better matching, more accurate forecasting, reduced fraud, or improved scheduling. Those outcomes can be commercially important, but they may not explain a technological solution. A useful patent-development process pushes one level deeper.

Connect the problem, mechanism, and measurable effect

The application should explain what technical limitation existed in prior systems, what components or operations solve it, and what technical effect follows. If parallel execution reduces latency, for example, the disclosure should explain the architecture, data dependencies, scheduling, or other features that make the parallel operation possible. Saying only that the system is “faster” may describe a benefit without identifying the invention.

Audit each independent claim for the asserted improvement

For every proposed independent claim, ask whether the claim can be performed while omitting the feature said to make the technology better. If the answer is yes, Berkeley*IEOR shows the risk of relying on that feature later as the inventive concept. This is a claim-scope question, not a request to copy every implementation detail into every claim.

Use a claim ladder deliberately

A claim set may include independent claims directed to different statutory forms and dependent claims that add implementation details. That structure can preserve multiple positions, but only if the claims actually capture the relevant distinctions. The broader claim and the narrower claim may receive different eligibility treatment. Tucker Law’s analysis of Constellation Designs v. LG illustrates how related claims can produce different Section 101 outcomes.

Keep the prosecution record consistent

An applicant responding to a Section 101 rejection should identify where the claimed language—not merely the background or preferred embodiment—recites the practical application or technical improvement. The USPTO’s MPEP § 2106 eligibility framework likewise directs examiners to evaluate the claim as a whole and determine whether additional limitations integrate an exception into a practical application or amount to significantly more.

Eligibility is only one part of patentability. Added limitations can affect novelty, obviousness, written description, enablement, infringement coverage, and business value. A sound patent-application strategy considers those issues together rather than optimizing for one doctrine in isolation.

What issued-patent owners should review

For an issued software patent, the first eligibility review should compare the asserted technical advantage against the actual elements of each potentially important claim. Marketing materials, expert declarations, and specification passages cannot create a claim requirement that is absent from the text.

That review should remain claim specific. One claim may require a concrete data structure, processing sequence, network arrangement, or resource-management technique while another reaches only a desired result. Owners should also consider the governing claim construction and the prosecution history before adopting an eligibility position. An interpretation offered to show a technical limitation may affect infringement or other validity arguments.

Timing matters as well. Section 101 can arise during prosecution, in pre-suit diligence, on a motion to dismiss, at summary judgment, or later in a case. Tucker Law’s patent-litigation practice evaluates eligibility alongside claim construction, infringement, prior art, damages, and the commercial objective of the dispute.

What accused businesses can learn from the decision

An accused business should not accept a patent owner’s description of the “invention” without mapping that description to the claims. A patent may discuss sophisticated technical features that an asserted claim never requires. The eligibility analysis should separate mandatory claim elements from optional embodiments, advantages, and attorney argument.

That does not mean every broad software claim is ineligible or that an eligibility challenge is always the best first move. Some claims recite concrete technological improvements, and some disputes require claim construction or factual development. A complete defense assessment should also consider noninfringement, prior art, ownership, standing, venue, damages exposure, and available post-grant options.

How Berkeley*IEOR fits the recent eligibility decisions

Constellation Designs emphasized that eligibility can differ across claims in related patents. Berkeley*IEOR presents the complementary drafting lesson in unusually direct form: an unclaimed technical feature cannot supply the inventive concept simply because the specification describes it and the feature would improve performance.

The opinion is nonprecedential and should be cited with that limitation. It also does not hold that parallel processing is inherently abstract or incapable of supporting eligibility. The narrower point is that parallel processing could not do the work in this case because the representative claim did not require it.

The practical takeaway

Before filing a software patent application—or asserting an issued software patent—identify the exact feature said to make the technology better and locate that feature in the claim language. If the claim can be practiced without it, the specification’s technical narrative may not carry the eligibility argument.

Tucker Law helps inventors and technology companies connect technical disclosure, claim strategy, prosecution, and enforcement. To discuss a software invention, an Office Action, or an issued patent under Section 101, contact Tucker Law.

This article provides general legal information and is not legal advice. Patent eligibility depends on the particular claims, specification, prosecution history, procedural posture, and controlling authority.