The commercialization of academic research is usually discussed as a pipeline problem: too few disclosures, too few patents, too few licences. The evidence points somewhere else. The science is rarely what fails. What fails is the machinery between a published result and a product someone pays for, and that machinery fails in five predictable places.
This page is the diagnosis. If you need the definition first, read what research commercialisation means. If you already have a technology and need to decide between licensing and spinning out, read which commercialization route to choose. What follows explains why so few of those routes end in a shipped product, and what changes the odds.
The yield problem in two numbers
Start with return on the public investment. In the 2021-22 academic year, UK universities earned GBP 244 million from licensing IP and GBP 86 million from selling spin-out shares, a combined return of just 2.1 percent of total research expenditure. The UK has one of the most mature university innovation systems in Europe. That is the good case.
Then look at the patents themselves. A study of research-based inventions at Politecnico di Torino cites earlier work finding that over 75 percent of university patents are never commercialized and deliver no value to society.
Those two figures describe inventions that cleared every early hurdle. Someone judged them novel, someone paid to protect them, and they still stopped. The reasons are structural, which is good news: structural problems have structural fixes.
Bottleneck one: the incentive system pays for papers
An academic career is built on publications, citations and grants. A licence or a spin-out counts for little in a promotion file, and time spent on customer discovery is time not spent writing.
The PoliTo researchers call this orientation asymmetry. Firms work to short horizons and measure success in revenue; academics work to long horizons and measure it in findings, models and methods. Because academics are evaluated on their publication record, they rarely have a personal incentive to prefer patenting over publishing, or licensing a technology over opening a new line of research.
The asymmetry has a second, quieter cost: fear. Between 2013 and 2017, around half of the PoliTo professors who applied for a university patent worried that an external company collaborating on the technology would steal the idea. A researcher who distrusts industry does not seek industry feedback, and a technology developed without that feedback drifts away from what any buyer wants.
Bottleneck two: transfer offices built for compliance
Technology transfer offices were created to solve this. In many institutions they have become a second bottleneck.
The Harvard policy analysis describes offices that operate as administrative compliance units rather than business development teams. Staff at public universities are incentivized not to give away state assets too cheaply, so negotiations stretch over every percentage point. The UK spin-out review found that 67 percent of spin-out deals took more than six months to complete. In software or AI, six months can make a technology obsolete before the company is formed.
Capacity is the other half. Statistics Canada data cited in the same analysis shows over 52 percent of institutions naming lack of expertise as a primary obstacle, and a university with a EUR 10 million research budget cannot generate enough deal flow to pay for specialist patent attorneys and business developers. Under that pressure, under-resourced universities abandon patents to save on renewal fees before a licensee is ever found, when renewing, abandoning or monetising a patent should be a commercial decision rather than a budget line.
Simply having an office does not fix it. A 2026 study in Scientometrics found that licensing outcomes show no robust positive association with technology licensing office affiliation; intermediation only creates value when it actually reduces uncertainty about a specific invention for a specific buyer.
Bottleneck three: the funding gap between grant and investor
This is the valley of death, and it has a precise shape. A research grant ends when the paper is written. A venture investor starts when there is a prototype and a business plan. Between those two points sits work that neither will pay for: proving the effect outside ideal lab conditions, building an engineering prototype, running the first pilot. Most European universities lack the proof-of-concept funds needed to cross it.
The gap is not only money. When researchers interviewed the directors of 16 US academic drug discovery programmes, the challenges they named were limited funding, lack of development know-how, and the absence of a regional drug development ecosystem. Cash without people who have done it before does not cross the valley either.
Grant programmes aimed squarely at this stage do exist; we cover the European ones in EU funding for deep tech startups.
Bottleneck four: the cap table breaks before the company exists
Suppose a technology survives all of that and a spin-out is formed. The next failure is written into its founding documents.
European universities have historically taken equity stakes in spin-outs averaging 15 to 30 percent, sometimes 50 percent or more, against a norm of 2 to 7 percent in the United States. Venture investors need founders to hold a meaningful share and need room on the cap table for their own returns. A company that starts with a blocking university stake is, in the Harvard analysis's words, uninvestable. The same terms discourage the most capable faculty from disclosing inventions at all.
Equity is negotiable, and a founder who knows the US norm negotiates from a different position. Know the benchmarks before you sit down: see how university spinout equity is split.
Bottleneck five: technology push without a buyer
The last bottleneck is the most common and the least discussed, because it looks like success. A lab has a genuine discovery, files a patent, and then goes looking for a problem to solve.
Deep Science Ventures puts a number on where that leads. The top five patent-producing universities in the US produce a third of new biotech companies, yet 47 percent of those companies have already failed. Tech push is not a guarantee of anything.
The Scientometrics study explains part of the mechanism. Basic research has a broad range of potential applications and an uncertain development path, which raises evaluation and integration costs for established firms. Universities weighted towards basic research achieve fewer licences on average, although basic research is associated with more startup formation, because a startup can experiment where an incumbent cannot.
The practical test is blunt. Strategic Spinouts argues that an invention usually needs to be an order of magnitude better, ten times cheaper, faster or more effective, before customers and investors accept the switching risk. And someone has to be willing to lead the company for the next five to ten years. A 20 percent improvement with a reluctant founder is a paper, not a company.
What actually moves research out of the lab
Each bottleneck has a counterweight that works. None of them is a new policy; all of them are about who carries the work after the paper.
Milestone money before equity money
Proof-of-concept funding is the direct answer to bottleneck three, and it works best tied to technical and commercial milestones rather than to publications. It also defuses bottleneck four: a technology that reaches investor readiness on non-dilutive money arrives at its first equity round with a cleaner cap table and a stronger negotiating position with its university.
Start from the outcome, not the patent
The alternative to tech push is to invert it. Deep Science Ventures works backwards from a desired outcome, identifies the technical and economic constraints that stand in the way, then recruits founders to the most promising approach. It has created 35 companies this way. The lesson for any lab is transferable: name the buyer and the constraint you remove for them before you name the invention.
Operators and deployment capital
The final gap is the one EX EPIC exists to close. Research funding buys experiments. Deployment needs a different instrument: capital that buys units, sites and certifications, and operators who have installed hardware in the field before.
We finance, patent and deploy breakthrough science on a EUR 160M+ capital track record across four continents, with 200+ waste-to-energy units deployed in 11 countries, 24 patent families filed with around 100 more validated in the pipeline, and 250+ trained operators placed across the portfolio. The pattern we see repeatedly is not weak science. It is a working prototype with nobody funded to spend two years turning it into a certified, installable product. That is the specific job of a deep tech venture builder, and it only works when the IP underneath is filed where the product will actually sell, which is why patent strategy for deep tech startups belongs in the first conversation, not the last.
A diagnostic for a stalled technology
If a technology you own, fund or manage has stopped moving, run it against the five bottlenecks in order.
- Incentives. Is anyone on the team rewarded for this reaching a market, or only for publishing it?
- Transfer office. Is the deal waiting on a negotiation that has run past six months, or on a renewal fee nobody has budgeted?
- Funding gap. Is there money, and people with development know-how, for the prototype and pilot stage specifically?
- Cap table. Would a venture investor accept the equity split as it stands today?
- Buyer. Can you name the customer, the constraint you remove, and why the improvement is ten times rather than 20 percent?
A single "no" is enough to explain a stall. Fix that one before spending on anything else.
FAQ
What percentage of university patents are commercialized? A minority. Research cited in the Politecnico di Torino study puts the share of university patents that are never commercialized at over 75 percent. The income side tells the same story: UK licensing and spin-out income together equalled only 2.1 percent of research spending in 2021-22.
Why do European universities commercialize less research than US universities? Three reasons recur in the evidence: universities have demanded spin-out equity stakes of 15 to 30 percent against 2 to 7 percent in the US, deals take longer, with 67 percent of UK spin-out deals running past six months, and many transfer offices are organized for compliance rather than business development.
What is the valley of death in academic research? It is the funding gap between the end of a research grant and the point at which a technology is ready for venture investment. Grants pay for discovery and investors pay for prototypes, so the engineering and pilot work in between often has no natural funder.
Is basic research harder to commercialize than applied research? It is harder to license, because its applications are broad and its development path uncertain, which makes it costly for established firms to evaluate. Basic research is, however, associated with more startup formation, since a new company can experiment with applications in a way an incumbent licensee rarely will.