The Startup That Became a Lawsuit
TexasLDPC was founded in late 2014 by Annapurna Yarlagadda and her husband, Dr. Kiran Gunnam, to commercialize error-correction technology for data storage and communications, the class of patents that keep hard drives and fiber links from corrupting data. The technology came from Gunnam's doctoral work at Texas A&M University, which granted the company an exclusive license in June 2015 to make, use, sell and sublicense the patented inventions, along with the exclusive right to sue infringers and collect damages in exchange for equity and a share of any recovery.
The company never found customers. It exhausted its capital and, by its own description, refocused solely on enforcing its rights. In December 2018 it sued Broadcom for patent and copyright infringement in the District of Delaware, later adding Avago Technologies and LSI Corporation, where Gunnam had once worked and urged the company to license the technology. Broadcom, Avago and LSI moved to dismiss, arguing that Texas A&M was a necessary party the suit could not proceed without.
The district court agreed and dismissed the case in September 2023 on two grounds: the license had automatically terminated when TexasLDPC ceased business operations by pivoting to enforcement, and even if it survived, TexasLDPC lacked all substantial rights in the patents and could not sue alone. Last week, the Federal Circuit reversed both findings.
What the Court Actually Decided
The appellate ruling is a study in how contract language outlives business strategy. The license defined commercially reasonable efforts expressly to include enforcing the patents and copyrights, and a separate provision barred termination for failing to record sales or sublicensing revenue by 2021 as long as TexasLDPC was exercising commercially reasonable efforts to enforce its rights or seek collections.
Under that language, the Federal Circuit held, a pivot to litigation was not a pivot away from the agreement. It was performance of the agreement. The district court had given too much weight to the agreement's recital clauses, which described A&M's interest in commercializing the technology, because under Texas law recitals cannot override operative terms. TexasLDPC's enforcement activity remained a business operation, and the license had not terminated.
On the second question, whether the licensee held enough rights to sue in its own name, the court applied the all substantial rights doctrine, under which a patent owner who transfers essentially everything short of title has made an assignment, conferring standing on the transferee alone. The panel examined the totality of the agreement, including the exclusivity of the license, the right to sue and collect damages, and the limits A&M retained, which included a research and education carve-out, the sole right to sue Marvell, and approval rights over assignment. The court concluded the district court had misapplied the doctrine and that the case could proceed.
The Technology at the Center
The patents cover LDPC error correction, a mathematical technique that lets data storage and communication systems recover information corrupted by noise. LDPC codes underpin much of modern storage and networking, and they are one of those technologies that became foundational without ever becoming famous. A dispute over who owns the right to practice a foundational method, licensed from a university to a startup, is a familiar shape in patent litigation, and the stakes are the royalty streams that flow through storage chips for years.
The Two Threshold Questions, Answered
The case turned on two doctrines that function as gates for every patent suit.
The first is termination. The district court read the license as ending when TexasLDPC stopped trying to sell products and started trying to enforce patents, on the theory that the company had ceased business operations. The Federal Circuit read the same agreement's operative language and found that enforcement was expressly included in the definition of commercially reasonable efforts, and that a separate provision barred termination for lacking sales as long as enforcement continued. Under Texas law, the court held, the recitals describing commercialization aspirations could not override those operative terms, so the agreement never terminated.
The second is standing. A patent can be enforced only by its owner or by a licensee holding all substantial rights, a standard the Federal Circuit has refined over decades. A&M retained real rights, the right to sue Marvell, approval over assignment, a research carve-out, but the panel found that what was transferred, exclusivity, the right to sue and collect damages, the practical substance of the patent, added up to all substantial rights under the doctrine's totality test. The distinction between retaining some rights and retaining substantial rights is where most of these cases are decided, and the panel walked through the agreement provision by provision to reach its answer.
What the Reversal Does Not Decide
The remand means the case proceeds, not that TexasLDPC wins. Infringement has not been adjudicated, claim construction has not begun, and the damages questions, whether Broadcom's chips practice the patented methods, whether the copyright claims survive, how long the alleged infringement ran, remain entirely open. The Federal Circuit decided who may sue, not who is right.
The parties' next moves are predictable. Broadcom can seek rehearing, and the district court will now schedule the litigation that dismissal avoided. TexasLDPC's recovery, if any, flows through its agreement with A&M, which takes equity, a share of sublicense royalties and a share of enforcement recoveries. The university's technology transfer office will be watching the economics of its most consequential license.
The Drafting Lessons Buried in the Opinion
The opinion reads, in places, like a treatise on how university technology licenses go wrong and how they go right. Texas A&M's agreement did three things the Federal Circuit rewarded: it defined commercially reasonable efforts to include enforcement, it barred termination while enforcement continued, and it made the assignment of enforcement rights explicit rather than implied. Each of those clauses did its work in this case, and each exists in some version in most university licenses, which is why the ruling will circulate far beyond patent litigators.
The district court's error, in the panel's account, was reading the agreement's purpose clauses as if they were conditions. Recitals say what the parties hoped for; operative terms say what they promised. Texas law, like contract law generally, enforces the promises. The practical advice that follows for anyone drafting or inheriting a license is unglamorous and specific: check what the definition of effort includes, check what the termination provision forbids, and never assume a recital can undo either.
The Longer Arc of the Dispute
The timeline is worth seeing whole. The patents emerged from doctoral research in the 2000s. The license was signed in 2015. The suit was filed in 2018. The dismissal came in 2023. The reversal arrived last week. The infringement questions have not been briefed on the merits, which means the technology that was cutting-edge when the research began may be standard by the time the case is finally heard.
That pace is a structural feature of patent litigation rather than an accident of this case, and it is the background fact behind every argument about the patent system's costs. A dispute that takes this long is expensive for everyone, and the parties will now negotiate and litigate in the shadow of that arithmetic, with settlement value determined by how much of the remaining runway each side can afford. The court has ruled on who may sue. The rest is endurance.
Primary sources
- IPWatchdog on the Federal Circuit reversal for the procedural history, the license terms and the court's reasoning.
- TexasLDPC Inc. v. Broadcom Inc., Federal Circuit opinion for the precedential panel opinion and the district court dismissal it reversed.