The sector is debating commercialisation. It has not yet recognised what it holds.

Faraz Rizvi × Foundry · 24 June 2026 · 9 min read · Markdown

Faraz Rizvi is a UK operator-practitioner writing about the work between a research breakthrough and a fundable company. He runs SpinUp Forge. Foundry is SpinUp Forge’s custom agentic harness.

The UK is debating how to get more companies out of its universities — equity terms, proof-of-concept funding, spinout throughput — and skipping the stage that sits before all of them: most institutions cannot see the assets they are trying to commercialise. The government’s own review put a figure on it — £150 billion of public-sector knowledge assets, with a recommendation that institutions simply register what they hold (HM Treasury, 2018). If you fund, run, or back a university spinout, you are working at the wrong end of the problem.

In February 2025 I sat on a panel marking ten years of ICURe — Innovate UK’s programme for getting academic teams out of the lab and in front of a market. The room in Manchester held most of the ecosystem: founders, investors, incubators, technology transfer offices, funders. I said something I thought was obvious. That research is, in a broad sense, the act of generating intellectual property — not in the narrow legal sense of a patent, but in the plain sense that if your work is not producing new knowledge, it is hard to say what it is for.

The room inhaled. I have been thinking about it ever since.

I have catalogued an industry’s assets before

The value was never the missing music; it was the missing systems. I spent fifteen years learning that in one industry, and now meet it again in another.

I did not come to this from academia. I spent the better part of fifteen years in the music business, during the years it was dragged (late, and against its will) into the digital age. At Entertainment UK, then Europe’s largest physical distributor, I proposed and built a lab to show the board what was coming. We defined the data standards for digital products, the encoding, the metadata; working with a young music discovery startup called Shazam we captured and fingerprinted the catalogue. By 2003 we had assembled what we believed was the largest digital music catalogue in the world: approaching two million tracks, at a time when iTunes (now Apple Music) launched with two hundred thousand.

An industry can be sitting on enormous value and be unable to find it. Later, at Universal Music, I helped architect and implement the supply chain that carried a release from metadata to rights to assets to distribution. The hardest shift was conceptual: moving from delivering a product — send the album to iTunes — to delivering an asset, once, with its rights attached: where it can stream, who can claim it, whether someone’s birthday video can use it. At one point I found that Universal was releasing its digital catalogue into only a modest fraction of the territories where it actually held the rights. Not due to missing music, it was missing systems. The value was real and locked up by plumbing.

I tell you this because I now spend my weeks inside a university, helping academics move research toward commercialisation. And I keep recognising the same problem I spent a career solving somewhere else — only here, almost no one has built the system to solve it.

The paper is the brochure

Academics are measured on the paper. The commercial value usually sits in the assets it was built on — the code, the models, the methods the system was never taught to see.

Academics are trained, rewarded and promoted for one output: the paper. But the paper is rarely the commercialisable thing. It is the verified, glossy brochure that narrates a body of work. The commercial value is more often in the assets built along the way — the framework, the method, the model, the dataset, the code.

The cleanest example I know. An astrophysicist modelling galaxies needed a faster way to run fluid-dynamics simulations, and built one. You will not find an industrial market for modelling galaxies. But a genuinely faster method for computational fluid dynamics touches aerospace, automotive, energy, flood defence — markets measured in billions. The galaxies were the research. The solver was the asset. It spun out as Morpheus Fluid.

I should be careful here, because a TTO professional will rightly push back. For a therapeutic or a new material, the patent is the asset, and the timing of disclosure is the whole game — publish too early and you have given it away. And most research code is not a product; it is a proof of concept that would need real engineering before anyone could use it. One study of nine thousand research scripts found only a quarter ran without error as published (Trisovic et al., 2022). So the honest claim is narrower than the slogan: the value is often in assets the institution does not recognise as assets — the software, the methods, the datasets, the know-how the system was never built to see — and almost never knows it holds.

The thing the sector is not arguing about

The sector has stopped arguing about equity and started arguing about throughput. Both sit downstream of the stage no one names: you cannot pipeline an asset you have never recognised.

The UK spinout debate has matured. We have stopped fighting about equity terms. The average university stake has fallen to around sixteen per cent in 2024, down from twenty-two the year before, and the guidelines have converged (RAEng, 2026). The argument has moved, correctly, to throughput and capacity: new spinouts fell from 202 in 2020-21 to 141 in 2024-25 (UKRI register, via THE); nearly half of English universities are forecast into deficit (Office for Students); technology transfer offices are cutting staff and narrowing to triage.

But there is a stage upstream of all of that, and we are not discussing it. You cannot pipeline, fund, or put through a route to market an asset you have never recognised or catalogued. The throughput debate assumes the asset is on the table. Often it is on a personal laptop, an unlabelled git repository, a departmental drive — invisible to the office whose job is to commercialise it. Discovery currently leans on the academic spotting the opportunity and coming forward. That is not a pipeline. It is luck, and it does not scale.

This is not only my observation. In 2018 the Treasury estimated the public sector’s knowledge assets at £150 billion and told institutions to register what they hold; in 2022 the government stood up a Government Office for Technology Transfer to do exactly that. By late 2024 it had catalogued more than 200 public-sector knowledge assets and funded over 130 proof-of-concept projects (GOTT, 2024). The doctrine exists, it is funded, and it works — across government departments and public labs. It has barely reached the university estate, where so much of the IP is generated.

Why “build a marketplace” is the wrong answer

Every open marketplace for university IP has been tried, and quietly failed. Music’s lesson is the opposite of a storefront: lay the boring rails first — the next thing reading your catalogue is a machine, not a person.

The instinct, having named the problem, is to build a platform — a discovery marketplace where industry can browse academic IP and pull what it needs. I understand the instinct, and have put that forward myself. Now I would gently warn against the naïve version of it, because it has been tried, repeatedly, and it has not worked. The most serious attempt at an open IP exchange, IPXI, closed in 2015 (IAM). Flintbox, Tynax, yet2, IN-PART — the survivors quietly stopped being marketplaces and became curated matchmaking services instead (Research Money). Open, two-sided markets for heterogeneous IP have not reached liquidity.

The music industry is instructive because of the order it learned the hard way. The distribution came first — the open internet, then Napster — and the devices and storefront followed fast: the iPod, then the iTunes Store. What the industry had to build in response was the unglamorous part — the supply-chain rails that let it legitimise and scale supply: standard identifiers for recordings and works, a metadata-exchange standard (DDEX), a fingerprinting system to know what was being used. And even now, with all of that, the body that collects US streaming mechanicals inherited over four hundred million dollars of royalties it could not match to an owner (The MLC). The rails are necessary, hard, and never finished. Universities have barely begun them. They have DOIs for datasets and citations for software, but no standard, commercial way to catalogue a method, a model, or the right to license it.

There is a forward reason to do the boring work now. The discovery layer is already shifting from people browsing to machines retrieving — industry is beginning to point AI agents at the literature, the patents, the repositories to find what it needs. An agent will read a catalogue no human would ever scroll. But an agent can only surface what has been described: an asset with no identifier, no metadata, no statement of who may license it is invisible to a model exactly as it is to a person. Catalogue now and you are findable when that shift lands. Wait, and you are unseen by the machines too. AI does not make the metadata work redundant — it is the precondition for it.

So the order matters. Recognise the asset. Catalogue it internally, so the commercialisation office can see what the institution holds. Standardise enough metadata that it can be described. Then stage controlled, rights-cleared discovery to industry — open the paper, gate the asset. The platform is the last ten per cent. The catalogue is the work.

Where this leaves the funders, and the rest of us

The best funders now back people over plans, but none of them recognise the asset upstream. Until an institution can see what it holds, the work falls to whoever will do it — asset by asset, founder by founder.

Some of this is a funding-culture problem. To their credit, the best funders have moved — ARIA backs people over plans (ARIA), ICURe buys out a researcher’s time rather than a milestone chart, Enterprise Fellowships fund roughly a hundred researchers a year to pursue the pathway, not the paperwork. The residual gap is not the plan format. It is that none of these instruments recognise the asset upstream — they fund an asset’s path to market, but only once someone has, by luck, already surfaced it.

I do not think the answer is to wait for a national platform. It may come; the policy logic is sound. But until an institution can see and steward its own assets, the discipline has to be supplied the hard way — asset by asset, founder by founder, from the operator’s side of the table. That is unglamorous. It is the work. The gap is not the science.

← Back to SpinUp Forge