From University Lab to Startup: How Medical Discoveries Become Products

Suppose you are a postdoctoral researcher, and after two years of failed attempts your experiment finally works. A molecule shrinks tumors in mice, or a sensor picks up a signal nobody could measure before. Your lab head is delighted. A journal paper is the obvious next step. But a colleague asks a different question over coffee: “Should you talk to the tech transfer office before you publish?”

That moment, often unglamorous and easy to miss, is where many medical products begin. This article follows the path from a university bench to a company that can put something in front of patients, and explains why most promising discoveries never complete the journey.

Step one: telling the university you invented something

Most research universities have a technology transfer office (TTO), a team whose job is to protect and license discoveries made by their researchers. The first formal step is usually an invention disclosure: a confidential form describing what was found, who found it, and who paid for the work.

Timing matters because of patents. Publicly describing an invention, in a paper, a poster or even a conference talk, can make it harder or impossible to patent in many countries. So researchers are often advised to disclose before they present. This is not about hiding science; the patent application itself is eventually published. It is about preserving the option of commercial development, which, for a drug that may cost a company a great deal to develop, investors generally expect.

Who owns a publicly funded discovery?

In the United States, much of that research is paid for by federal agencies such as the National Institutes of Health. Under the Bayh-Dole Act of 1980, universities and small businesses can keep ownership of inventions made with federal funding, in exchange for obligations: they must report the inventions, the government keeps a licence to use them, there is a preference for US manufacturing, and agencies hold “march-in” rights to license an invention to others in certain circumstances.

A Government Accountability Office report published in April 2026 found that funding recipients kept ownership of most federally funded inventions. When they declined, the most common reason was that the invention had low commercial potential: nobody could be found to develop it.

License it, or start a company?

Once a patent application is filed, the TTO looks for someone to develop the invention. There are two broad routes:

• License to an established company. A pharmaceutical or device firm takes on the development, paying the university in some combination of upfront fees, milestone payments and royalties.
• Form a spinout. A new company is created around the technology, often with the inventors as founders or advisers. The university typically licenses the patent to the startup, sometimes in exchange for a share of the company, known as equity.

A startup makes sense when the technology is too early or too unusual for a big company to take on, or when it could become a platform for many products. Our biotech startup map covers what happens after the company exists, from seed funding to partnerships.

A discovery proves something is possible. A product proves it can be made, repeatedly, safely, and for someone who needs it.

Crossing the “valley of death”

People in translational research talk about a valley of death: the funding gap between an exciting academic result and something investors can back. University grants reward new knowledge; investors want reproducible data, a manufacturable product, a clear route through regulators and a market. The work in between, repeating experiments in independent labs, testing safety, developing a stable formulation, is expensive and not very publishable.

In the US, one bridge is the Small Business Innovation Research (SBIR) program and its sister, Small Business Technology Transfer (STTR), which is built around partnerships with research institutions. NIH describes them as non-dilutive funding for early-stage research and development, meaning founders do not give up ownership in exchange. The programs lapsed and were then reauthorized for five years on 14 April 2026, a reminder that even long-standing support can pause.

Money is not the only gap. Many scientist-founders have never run a business. NIH’s I-Corps program offers funded small-business teams an eight-week course in customer discovery, which means interviewing the people who would actually use, buy or pay for a product. Teams are expected to talk to at least 100 stakeholders, from clinicians to hospital buyers. A common outcome is a pivot: learning that the problem doctors most want solved is not the one the lab set out to solve.

The regulatory road shapes the business

What a startup must prove, and so how much money and time it needs, depends heavily on what kind of product it is.

Medicines

The FDA describes five steps for a new drug: discovery, preclinical research in the lab and in animals, clinical research in people, FDA review, and safety monitoring after approval. The clinical stage alone runs through several phases of trials, which we explain in inside a clinical trial. You can walk through each stage in our Drug Development Journey.

Medical devices

Devices are sorted by risk. According to the FDA, most lower- and moderate-risk devices reach the market through a 510(k) submission, showing they are “substantially equivalent” to a device already on sale. Novel devices without such a predecessor may use the De Novo pathway, and the highest-risk devices, such as many implants, need premarket approval, the most stringent route. A startup building a new type of diagnostic sensor faces a very different road, and cost, than one improving an existing instrument.

A case study: from Penn labs to the first CAR-T approval

CAR-T therapy re-engineers a patient’s own immune T cells so they recognize and attack cancer cells. One version was pioneered at the University of Pennsylvania, where Carl June played a leading role. In 2012, Penn and Novartis formed an alliance to develop the therapy, managed on Penn’s side by its technology transfer center, which describes coordinating research agreements, manufacturing arrangements and material sharing across institutions.

On 30 August 2017 the FDA approved Kymriah (tisagenlecleucel), which Novartis describes as the first CAR-T therapy approved, for children and young adults up to age 25 with a form of acute lymphoblastic leukemia that had not responded to treatment or had relapsed at least twice. The approval was for a specific blood cancer in a specific group, not cancer in general.

The lesson is that “lab to product” does not always mean a startup. Here, a university partnered directly with a large company that could fund manufacturing and trials at scale. For the wider set of cancer tools CAR-T sits among, see the new cancer toolkit.

Why most discoveries stop short

There is no shame in a discovery that never becomes a product. The reasons are usually mundane:

1. The biology does not hold up. The result does not reproduce, or works in mice but not in human tissue.
2. It cannot be made reliably. A therapy that is difficult to manufacture consistently is hard to test and harder to sell.
3. The need is not there. Clinicians already have an adequate option, or no one will pay for the new one.
4. The patent is weak. Without protection, investors may be unwilling to fund expensive trials.
5. Money and time run out before the data are convincing.

Understanding these hurdles is also useful when reading company news; our guide to researching a biotech stock explains how to separate a strong story from strong evidence.

What to watch

• How the renewed SBIR and STTR programs change support for early-stage biomedical startups.
• Whether NIST’s draft framework on march-in rights, and changes to Bayh-Dole reporting, alter how universities license inventions.
• Growth of university-run “proof of concept” funds that aim to shrink the valley of death before a company forms.