Patents

Stanford v. Roche: Bayh-Dole Did Not Move Title Away From the Inventor

The Supreme Court held that the Bayh-Dole Act does not automatically vest patent rights in a federally funded university — title still begins with the inventor, and a stray 'do hereby assign' can send it elsewhere.

February 17, 2025
Researcher pipetting samples at a laboratory bench
A Stanford fellow's confidentiality agreement at a private company, not the federal funding statute, decided who owned the HIV-detection patents. Shutterstock
Educational content, not legal advice. This article explains general legal concepts. It does not create an attorney–client relationship. For your specific situation, consult a licensed attorney.

Board of Trustees of the Leland Stanford Junior University v. Roche Molecular Systems, Inc., 563 U.S. 776, No. 09-1159 (decided June 6, 2011), answered a deceptively simple question with enormous consequences for every research institution that takes federal money: when an employee invents something using federally funded research, who owns the patent? Stanford assumed the Bayh-Dole Act handed title to the university. Writing for a 7-2 majority, Chief Justice Roberts disagreed. Title, the Court held, begins where it has begun since the first patent statute — with the inventor — and Bayh-Dole does nothing to disturb that default. Because a Stanford researcher had earlier signed away his future rights to a private company in language that operated as a present assignment, the university’s competing paperwork came too late.

At a glance

  • Case: Bd. of Trustees of the Leland Stanford Junior Univ. v. Roche Molecular Systems, Inc., 563 U.S. 776 (2011)
  • Docket: No. 09-1159; decided June 6, 2011
  • Court: Supreme Court of the United States (on certiorari to the Federal Circuit)
  • Majority: Chief Justice Roberts (joined by Scalia, Kennedy, Thomas, Alito, and — except as to a footnote — Sotomayor); Sotomayor concurring; Breyer dissenting, joined by Ginsburg
  • Statute: Bayh-Dole Act, 35 U.S.C. §§ 200-212
  • Technology: PCR-based methods for quantifying HIV in blood, used to measure the efficacy of antiretroviral therapy
  • Holding: Bayh-Dole does not automatically vest title to federally funded inventions in the contractor; rights in an invention belong initially to the inventor, who may assign them
  • Status: Federal Circuit judgment affirmed

The two agreements that decided ownership

The dispute grew out of a chain of paperwork rather than a chain of discovery. Dr. Mark Holodniy joined Stanford’s Department of Infectious Diseases as a research fellow and, on arrival, signed a Copyright and Patent Agreement in which he stated that he “agree[d] to assign” to Stanford his right, title, and interest in inventions resulting from his employment. To develop the quantification technique, Holodniy needed access to the polymerase chain reaction (PCR), then held by the private company Cetus. As a condition of working at Cetus, he signed a Visitor’s Confidentiality Agreement providing that he “will assign and do[es] hereby assign” to Cetus his right, title, and interest in inventions made as a consequence of his access.

That difference in verb tense became the whole case. Cetus’s PCR-related assets, including its agreements with visiting researchers, were later acquired by Roche, which commercialized HIV-detection kits. Stanford obtained three patents naming Holodniy among the inventors and sued Roche for infringement. Roche answered that it already owned an interest in the very inventions Stanford was asserting, traced through Holodniy’s Cetus agreement.

Why “do hereby assign” beat “agree to assign”

The Federal Circuit had resolved the ownership clash on contract grounds drawn from its own precedent in FilmTec Corp. v. Allied-Signal, Inc. Under that line of cases, a promise to assign in the future (“agree to assign”) conveys only an equitable interest that requires a later act to perfect, whereas a present assignment of future inventions (“do hereby assign”) automatically transfers legal title the instant the invention comes into being. On that reading, Holodniy’s Stanford agreement created only an expectancy, while his later Cetus agreement immediately vested legal title in Cetus (and thus Roche) once the invention was conceived. Stanford was left holding a promise that the inventor was no longer free to honor.

The Supreme Court did not disturb that contract analysis. It granted certiorari on the larger question Stanford pressed: that the priority of the agreements should not matter because Bayh-Dole had already routed title to the university by operation of law. If Stanford were right, Holodniy never had assignable rights to give Cetus in the first place.

The Bayh-Dole argument, rejected

Stanford’s statutory theory rested on Bayh-Dole’s text and purpose. The Act gives a “contractor” — typically the university or business that performs federally funded work — the right to “elect to retain title” to any “subject invention,” defined as an invention “of the contractor.” Stanford read “of the contractor” to mean inventions that the contractor automatically owns by virtue of the funding relationship, with the inventor’s own rights extinguished or subordinated.

Chief Justice Roberts rejected the reading as contrary to two centuries of patent law. Since the Patent Act of 1790, the Court explained, rights in an invention have belonged in the first instance to the inventor — “the original and first inventor” — and only the inventor can convey them. Congress legislates against that backdrop, and when it wants to displace it, it does so unmistakably. Bayh-Dole, the Court observed, “speaks of disposition of rights in government funded inventions assuming that the contractor had title to begin with”; it never states that title vests in the contractor or authorizes a contractor to take title unilaterally. The phrase “invention of the contractor” most naturally means an invention the contractor already owns, not one it is entitled to seize.

The Court reinforced the point by contrast. When Congress has wanted to vest title in the Government or a contractor, it has said so directly, as in statutes governing inventions made by federal employees and atomic-energy inventions. Bayh-Dole’s machinery — the contractor’s right to “elect to retain title,” the Government’s “nonexclusive license,” and the agency’s limited “march-in rights” — all presuppose a contractor that holds title, not a statute that confers it. The result: a federally funded inventor who has validly assigned rights to a third party can leave the university with nothing under Bayh-Dole.

Open questions

The decision settles the threshold ownership rule but leaves the practical edges to be managed by drafting. The Court expressly declined to revisit the FilmTec distinction between present assignments and mere promises to assign, even as Justice Breyer’s dissent questioned whether so much should turn on a verb tense and warned that the rule rewards the second assignee who happens to use the magic words. Whether and how the Federal Circuit might soften FilmTec in a future case remains open. The opinion also leaves unaddressed how Bayh-Dole’s disclosure and election obligations operate when a contractor never obtained title because of a competing private assignment — a gap that affects the Government’s own license and march-in rights. Finally, the majority did not decide whether equitable doctrines might rescue an institution that holds only a promise to assign against a later present assignment taken with notice.

Implications

  • Get the assignment in present-tense, vesting language. Institutions and employers should use “hereby assigns” present assignments of future inventions, not “agrees to assign” promises, to capture legal title automatically at conception.
  • Bayh-Dole is not a title-grabbing statute. Federal funding does not, by itself, give a university or contractor ownership; the institution must still secure assignments from its inventors.
  • Audit collaborator and visitor agreements. Confidentiality and access agreements at partner companies can contain assignment clauses that quietly capture an employee’s inventions; conflicting commitments must be policed before collaboration begins.
  • First valid present assignment wins. Where two agreements compete, the one that effects an immediate transfer of legal title generally prevails over a mere expectancy, regardless of which was signed first.
  • Standing flows from ownership. A university that cannot prove it holds title may lack the right to sue; ownership defects surface as dispositive defenses, not technicalities.

Frequently asked questions

Did the Supreme Court invalidate Stanford’s patents? No. The Court did not hold the patents invalid. It held that Bayh-Dole did not give Stanford automatic title, leaving in place the Federal Circuit’s conclusion that Roche held an ownership interest through the inventor’s earlier present assignment to Cetus — which undermined Stanford’s ability to maintain its infringement suit.

Does this mean federally funded universities cannot own their researchers’ inventions? Not at all. Universities routinely and validly own such inventions — but they must obtain effective assignments from their inventors. Bayh-Dole governs what a contractor may do with title once it has it; it does not supply the title itself.

Why did the wording of two assignment clauses matter so much? Under the Federal Circuit’s FilmTec rule, “do hereby assign” transfers legal title the moment the invention exists, while “agree to assign” creates only a future promise. The inventor’s later present assignment therefore vested title before Stanford’s earlier promise could be perfected.

Authorities and sources