The Bayh-Dole Tripwire: What the Harvard March-In Threat Means for Every University, Nonprofit, and Small Business Holding a Federally Funded Patent

For 45 years, the federal march-in right under the Bayh-Dole Act was the dog that did not bark. Every administration since Carter's signed the law in December 1980 has had the authority. None had used it. The authority sat in 35 U.S.C. § 203, cited in tech-transfer training decks and footnoted in every university IP policy, but treated by patent counsel and grant administrators as a theoretical backstop — the kind of provision lawyers explain to clients with the caveat "this has never been invoked."

On August 8, 2025, Commerce Secretary Howard Lutnick sent a letter to Harvard President Alan Garber announcing that the Department was initiating a march-in proceeding against the university's federally funded patents and giving Harvard until September 5, 2025 to produce a complete list of every patent stemming from federally funded research, along with disclosure dates, licensing terms, and commercialization status. The letter cited three categories of alleged non-compliance: failures of timely invention disclosure and election of title, violations of the U.S.-industry preference in licensing, and failure to take "effective steps to achieve practical application." It was the first invocation of march-in authority in the statute's history.

Nine months later, the immediate Harvard dispute has receded from the front page. The policy precedent has not. Every university, hospital, nonprofit research institute, and small business holding a Bayh-Dole patent — that is, every entity that ever elected title to an invention conceived under a federal research grant — is now operating under a regulatory regime in which march-in is no longer theoretical. The compliance posture that was sufficient on August 7, 2025 is no longer sufficient on May 22, 2026, and the gap is wider than most institutions realize.

What Bayh-Dole Actually Promised

The bargain Congress struck in 1980 was straightforward. Before Bayh-Dole, inventions made under federal research grants were owned by the government, which sat on roughly 28,000 patents and had licensed fewer than five percent of them. Universities, small businesses, and nonprofits could not commercialize what they had invented because they did not own it, and industry would not invest in inventions that the government could license to anyone at any time. Bayh-Dole reversed the default: grantees could elect title to inventions conceived under federal funding, the federal government retained a non-exclusive license for its own use, and a narrow set of conditions — outlined in 35 U.S.C. § 203 — allowed the government to "march in" and grant licenses to third parties if the grantee was failing to commercialize, failing to meet public-health or safety needs, or failing to preference U.S. industry in licensing.

The statute lists four march-in conditions. The grantee has not taken effective steps to achieve practical application of the invention. The grantee has not satisfied health or safety needs reasonably required by the invention. The grantee has not met requirements for public use under federal regulations. Or the grantee has breached the preference-for-U.S.-industry clause in 35 U.S.C. § 204. The first condition — the "practical application" trigger — is the broadest, and historically the most contested, because the statute does not define how much commercialization is enough, how long a grantee has to achieve it, or what evidence the agency considers.

For 45 years, every petition to invoke march-in authority — including the high-profile petitions filed against AbbVie's Norvir HIV drug in 2004, Pfizer's Xalatan glaucoma drug in 2010, and Astellas/Pfizer's Xtandi prostate-cancer drug in 2016 and 2023 — was rejected. NIH, which received most march-in petitions because most federally funded patents originate in biomedical research, consistently held that high drug prices alone did not satisfy the "practical application" trigger, because the drugs in question were widely available on the market. The Biden administration's December 2023 draft framework opened the door to considering price in the practical-application analysis, but did not finalize the framework, and the door remained mostly closed.

The August 2025 Harvard letter walked through it.

What the Harvard Letter Actually Did

The legal mechanics of the Harvard proceeding matter, because they translate directly into what every other Bayh-Dole grantee should be planning around. The Commerce Department did not invoke march-in by certifying that a specific commercialization failure had occurred on a specific Harvard patent. It opened a compliance review across Harvard's entire federally funded patent portfolio and demanded the university produce the disclosure, election, and licensing record for every patent in that portfolio.

This is procedurally distinct from the way the practical-application march-in mechanism has been described for four decades. The framework most tech-transfer offices internalized assumed that march-in would arrive, if it arrived, as a petition challenging a specific commercialization failure on a specific patent — Norvir or Xalatan or Xtandi — and the institution would respond with evidence on that one invention. The Harvard procedure inverted the model. It treated the absence of a clean, auditable compliance trail across the full portfolio as itself a march-in trigger. The implicit theory is that any institution unable to produce timely disclosure records, election-of-title records, and "practical application" evidence for every federally funded patent is, by definition, failing to meet Bayh-Dole's obligations.

Wilson Sonsini's legal alert on the proceeding framed the operational consequence directly: the threshold for march-in exposure is no longer "are you commercializing this specific invention" but "can you document that you have complied, across your entire portfolio, with every disclosure and election obligation Bayh-Dole imposes." Few institutions can. The compliance gap between what Bayh-Dole technically requires and what tech-transfer offices have historically maintained is substantial.

The Compliance Stack Most Institutions Actually Have

Bayh-Dole's reporting machinery is the iEdison system — now part of NIH's Interagency Edison platform — through which every federal funding recipient is supposed to disclose subject inventions, elect title, file patent applications, report continued utilization, and disclose licensing terms. The compliance touchpoints include the initial invention disclosure (within two months of the inventor's report to the institution), the election of title (within two years of disclosure to the agency), the patent application filing (within one year of election), and ongoing annual utilization reports.

In practice, what most institutions maintain is a comprehensive record of invention disclosures and patent filings, and a much thinner record of utilization reporting, U.S.-manufacture compliance, and "practical application" documentation. The annual utilization report is widely treated as a formality; the U.S.-manufacture preference under § 204 is widely treated as a licensing-template clause rather than an enforced obligation; and the "effective steps to achieve practical application" requirement is essentially never documented contemporaneously, because nobody used to ask.

The Harvard letter asks. The other implication of the Commerce Department's portfolio-wide approach is that institutions with stale or incomplete iEdison records — patents disclosed but not elected, patents elected but with no utilization reporting, patents licensed without documented U.S.-manufacture compliance — are now sitting on the same compliance posture that triggered the Harvard proceeding. The administrative cost of catching up is substantial: an audit across a portfolio of 200 to 2,000 patents, depending on the institution, requires retrieving twenty to forty years of disclosure history, reconstructing election timelines, verifying licensing terms, and contacting industry licensees for current utilization data. Few tech-transfer offices have the staffing to do this proactively.

Who Is Actually Exposed

The march-in authority under § 203 applies to every "subject invention" — defined under § 201(e) as any invention conceived or first actually reduced to practice in performance of work under a federal funding agreement. The funding agreement does not have to be a research grant. It includes cooperative agreements, contracts, subcontracts, subgrants, and consortium memberships in which federal funding flowed to the institution. The class of exposed institutions is therefore broader than research-intensive universities. It includes:

• Every R1 and R2 research university with NIH, NSF, DOE, DOD, or NASA grant awards.
• Every academic medical center with NIH cooperative agreements.
• Every independent research institute (Salk, Whitehead, Broad, HHMI affiliates) with federal funding flowing through subaccount mechanisms.
• Every nonprofit research organization with federal pass-through funding.
• Every small business that has elected title to a SBIR or STTR-derived invention, under the small-business carve-out in Bayh-Dole.
• Every contractor or subcontractor on a federal R&D contract where invention rights were retained by the contractor.

The small-business exposure is especially relevant for SBIR and STTR firms, because Bayh-Dole's election-of-title and disclosure obligations apply with the same force to a five-person startup that won a Phase II contract as to Harvard. The small business that lets a Phase II SBIR patent disclosure lapse — that fails to file the iEdison report, fails to elect title in time, fails to document practical application — has, under the precedent the Commerce Department asserted in August 2025, the same march-in exposure.

What Has Actually Changed Operationally

Three things have changed since August 8, 2025, regardless of how the Harvard-specific proceeding is ultimately resolved.

First, the burden of proof has shifted. Before 2025, a grantee defending against a march-in petition produced evidence on the specific invention in question. After 2025, the implicit standard is that the institution must be able to produce a clean compliance record across the full portfolio on demand. Institutions without a current iEdison audit, current utilization tracking, and current U.S.-manufacture documentation are exposed to a compliance-review demand they cannot answer quickly.

Second, the policy rationale has shifted. The Biden 2023 draft framework focused on whether high drug prices satisfied the practical-application trigger — a narrow theory that affected mostly pharmaceutical patents. The 2025 Commerce theory is much broader. It treats systemic non-compliance with disclosure and election obligations as itself a march-in trigger, decoupled from any specific commercialization complaint. This expands the universe of exposed patents from "marketed drugs with contested pricing" to "every patent in the portfolio."

Third, the institutional risk calculus has shifted. Tech-transfer offices that historically treated Bayh-Dole compliance as a routine administrative function — handled by a portfolio manager and a paralegal — are now being asked by general counsel, by university presidents, and by board audit committees to demonstrate that the compliance record is clean enough to survive a Commerce Department demand on five weeks' notice. The internal staffing implication is significant. Most institutions are understaffed for this posture.

What Grantees Should Do This Quarter

Counsel guidance from the major university tech-transfer firms — Wilson Sonsini, Kilpatrick Townsend, Foley Hoag, Morrison Foerster — has converged on roughly the same five-point remediation playbook for the next two quarters.

Pull a full iEdison audit across every active and inactive federally funded patent. Identify gaps in disclosure timing, election timing, patent-application filing, and utilization reporting. Reconstruct the timelines where records are incomplete. The exercise will surface patents that were never properly elected, patents that lapsed without notice, and patents whose utilization reports have not been filed for multiple years. None of these gaps is fatal in isolation, but their cumulative weight is what creates march-in exposure.

Document contemporaneous "effective steps to achieve practical application" for every patent the institution intends to retain. The Bayh-Dole regulations at 37 C.F.R. § 401 do not specify the form, but the documentation should include licensing inquiries, marketing efforts, prototype development, clinical-trial activity where applicable, and any other evidence of bona fide commercialization effort. The institution that has spent five years quietly holding a patent without any commercialization activity is the institution most exposed under the practical-application trigger.

Audit licensing agreements for § 204 U.S.-manufacture compliance. Bayh-Dole requires that any product that will be sold in the United States and embodies the subject invention be substantially manufactured in the United States, unless the agency has granted a waiver. Many licensing templates include the clause; few institutions verify the licensee's actual compliance with it. The Harvard letter cited U.S.-industry preference violations as one of the three categories. Licensees that have offshored manufacturing without a waiver create direct § 204 exposure.

For SBIR and STTR small businesses, conduct the same audit on a smaller scale. The compliance obligations are identical; the staffing constraints are tighter. Most small businesses do not have a dedicated tech-transfer office. The CEO, CTO, or external patent counsel should personally verify that every Phase II election was filed on time and every utilization report is current. The cost of catching up is small; the cost of an uncured lapse, under the new enforcement posture, is potentially the loss of the patent.

Review the broader federal grant compliance environment. The march-in posture is one element of a wider shift in how the federal government is using its existing grant-compliance authorities — anti-fraud enforcement, audit demands, suspension and debarment, OMB Uniform Guidance terminations — to assert leverage that prior administrations chose not to use. The institutions that build clean, defensible compliance records across the whole stack are the institutions that survive the new posture without losing patents, awards, or institutional standing.

The Wider Direction

The Bayh-Dole bargain was not designed to be lightly disturbed. The reason march-in sat dormant for 45 years was not bureaucratic neglect; it was deliberate restraint. Every administration recognized that aggressive use of the authority would chill the innovation pipeline that the statute exists to support — that startups, pharmaceutical investors, and corporate licensees would not put capital behind federally seeded inventions if the government could pull the patent at will. The Carter-era compromise was that the federal government would have the tool but would not use it.

The August 2025 Harvard letter ended that compromise. Whether that proves to be a one-time political action against a specific institution or the beginning of a sustained enforcement posture is, at this point, unknowable. What is knowable is that the legal authority is now demonstrably live, the procedural model is now demonstrably portfolio-wide, and the operational compliance gap at most grantee institutions is demonstrably substantial. The institutions that act in the next two quarters to close that gap will have a much easier time in the next four. The institutions that wait will not.

For coverage of related federal grant-policy shifts in 2026, see Granted's analysis of the White House anti-fraud task force, the Schedule Policy/Career reclassification of federal grant-makers, and the NEH grant terminations ruled unconstitutional. The throughline across all of these stories is the same: existing grant-compliance authorities, dormant or lightly used for decades, are now being exercised. The compliance posture that worked yesterday is not the compliance posture that works today.