Ask how venture studios build companies and most answers describe a funnel. Ideas go in, a few survive, a founder is hired, a company comes out. True, and it misses the mechanism. The thing a studio actually builds for most of the journey is not a company at all. It is a project: a technology or a thesis with people assigned to it, money spent on it, and no legal entity, no dedicated team and no funding round attached. The company is created late, on purpose, when evidence says the project has earned one.
That distinction is the whole trick. This piece explains what a project is allowed to be inside a studio, the gates that convert it into a company, what the new company inherits on the day it is born, how long the clock runs, and why the science version of the build runs differently. If you want the stage-by-stage walk through idea sourcing, founder matching and the cap table, read how venture studios work, step by step. This article is about the threshold.
A studio builds a project first, and a company only when it has to
Tim Keller runs Inventopia, a deep-tech incubator partnered with UC Davis, and has watched more than 40 companies come through it in eight years. His diagnosis of the standard path is blunt. The default model assumes every idea needs its own company, built through funding rounds tied to milestones, and each round costs the founders roughly 20 percent of ownership. Founders respond rationally: fewer, bigger raises, which often fail, instead of the small targeted experiments that would actually move the work forward. He calls it the solo-company ratchet.
The studio alternative, in Keller's words, is to develop a technology as a project rather than a full-time pursuit for any one person. Promising projects advance without a dedicated round or a dedicated team. Mature projects are matched with a full-time CEO and spun out as standalone companies, or sold directly. One line carries the philosophy: instead of building one company at a time, build a system for building many.
Everything below is what that system looks like from the inside.
What a project is allowed to be inside a studio
No entity, no dedicated team, no round
A project in a studio has three things a startup does not have, and lacks three things a startup is forced to have.
It has a shared team. In Keller's own studio, a shared technical group works across multiple projects at once, supported by centralized staff for functions like HR, accounting and sales. It has a budget line rather than a cap table. And it has an owner inside the studio who is accountable for the next decision, not for a five-year plan.
It lacks a legal entity, because registering one before the odds justify it creates administrative work for a team that is not ready to carry it. It lacks a dedicated full-time team, because the people it needs this month are not the people it needs in six months. And it lacks a priced round, which is the point: nobody has given up 20 percent of anything yet.
That is also why studios are stingy at the front door. Lehnert Ventures describes selection as a matter of thesis fit, a painful buyer problem and finite builder capacity, and argues that accepting every exciting idea produces polite neglect across too many builds. A studio that says no early is protecting the projects it already has. How that bench is organised, and where it breaks, is covered in the shared team model.
Who pays while it is a project
The studio does. One European studio operator is explicit that the risk and cost up to incorporation, including some fees for the co-founders, accumulate on the studio, and that his firm would never incorporate a company it did not believe had a real chance of significant success. In jurisdictions where winding down a company is slow or expensive, he adds, delaying incorporation is even more logical.
So the project phase is expensive for the studio and free for the idea. That asymmetry is what lets a studio run several projects in parallel and stop most of them without a bankruptcy, a down round or a founder's reputation attached.
The gates that turn a project into a company
GCV Labs puts the principle in one sentence: each stage of the process exists to produce evidence for a decision, not activity for its own sake. Their seven stages run ideate, evaluate, create, incubate, launch, accelerate, scale, and most ideas stop at evaluate, which they describe as the point of the stage.
Evidence thresholds, written before the work starts
The gates are only useful if they are written down in advance. Inverse Collective's playbook says it plainly: write the kill criteria before you start, because you will not be objective later, and ventures that should have stopped at month four and limped to month twenty consume the attention that the next project needed.
What a gate contains varies by studio, but the shape is consistent. Lionpeak Partners lists what it takes to greenlight a first product: five to ten qualified customer interviews confirming urgent pain, two to three signed pilot letters of intent with success criteria, and one quantifiable traction metric. Validation sprints compress that into a four to twelve week decision window, and the MVP phase that follows typically runs two to four months.
Read those as a template, not a law. The useful property is that the threshold exists before the work begins, so the decision at the gate is a comparison against a number, not a debate about a feeling.
The ninety-day proof plan
Between gates, the project needs an operating rhythm, or it drifts into branding work that feels busy and changes little. Lehnert's version is a ninety-day proof plan: name the first product slice, the first buyer conversations, the first hire or contractor needs, the weekly owners, and the rewrite triggers that define which customer signal forces a change to packaging or pricing. Then run it on a named cadence: a Monday decision review with open trade-offs logged, a midweek release plus at least one structured buyer conversation owned by a named person, and a Friday scoreboard on proof milestones, hiring status and risk language that would embarrass a slide deck.
The cadence is boring on purpose. It is also the thing a founder is actually buying. If the people who sold the partnership never appear in the weekly work, you have a logo relationship, not co-builders.
Incorporation is a decision, spin-out is another
Here is the step most explainers blur. A project becomes a company in two separate events, and a disciplined studio does not confuse them.
The first is incorporation. It happens when the gates say the odds justify a legal entity. The second is spin-out, which the same European operator defines by a different trigger: the company raises outside funding. Between those two events the studio is still financing most of the work, in his case as much as 300,000 to 500,000 after incorporation, and while the first round is being raised it hires the three to five people who will make the company a standalone team on day one. After the round, the core team moves from operating the company to supporting it, and a studio partner stays on the board alongside the CEO and CTO.
So the honest answer to "when does the project become a company" is: legally, at incorporation; operationally, at the first external round; and the gap between them is where the studio earns its stake.
What the new company inherits on day one
The reason a studio-built company starts further down the field than a garage startup is not the money. It is the assets the project accumulated while it was still a project, and the studio's standing assets it plugs into.
Inverse Collective frames the test well: if venture four starts from the same blank page as venture one, you do not have a studio, you have a portfolio with extra overhead. The inheritance they list is five playbooks: discovery (who to interview, what counts as a yes and what counts as a no), brand and onboarding, engineering and data with analytics shared so unit economics are visible from week one, growth motions that have worked before, and operator hiring run from a bench built before it is needed.
GCV's build stories show the same thing from the company's side. In their create stage, the founding team, brand, proposition, product design and first working technology are built in-house rather than contracted out, and capital arrives when there is something real to fund. Involvement then continues past launch, which M Accelerator describes as the graduation phase, where the studio's role shifts from operations to ownership.
The practical consequence for a founder stepping into a studio-built company: you inherit customers, a product, a brand and a set of tested processes. You also inherit decisions you did not make. Ask to see the decision log from the project phase before you sign.
How long it takes: the 12 to 24 month clock
Put the stages on a calendar and the shape of a software build is fairly consistent. The Indian Venture Studio Association publishes a five-stage version with months attached, and it lines up with the gate lengths above.
| Months | Stage | What has to be true to pass |
|---|---|---|
| 1 to 2 | Ideation and research | Market mapped, customer problem found, regulatory check done |
| 3 to 4 | Customer validation | Manual experiments, pricing tests, non-binding letters of intent |
| 5 to 8 | Prototype and pilot MVP | A working v1 with pilot design partners, retention measured |
| 9 to 12 | Incorporation and founder matching | Legal entity, IP transferred from studio to company, pre-seed capital in |
| 13 to 24 | Independent spin-out | Own core team hired, off shared resources, external seed raised |
Source for the stage months: IVSA. Note where incorporation sits. In this version the entity is created around month nine to twelve, after the pilot has produced retention data, and the IP moves from studio to company at that moment. Everything before that line is a project.
How the build changes when the project is science
Almost every page about how venture studios build companies assumes software, where a landing page can test demand and a prototype takes a sprint. A gasifier, a biomanufacturing process or a carbon removal method cannot be smoke tested, and the project phase runs on a different clock.
Risk comes off in a different order
Keller's data on deep-tech failure is the starting point. Across the companies he has worked with, most did not fail because the technology did not work. They failed for human and strategic reasons, often rookie mistakes by academic founders, or were crushed slowly by the capital needs of commercialization. Both, he argues, are consequences of the solo-company structure, and both are what the project model is designed to prevent. He cites studio models at roughly twice the internal rate of return of venture capital, four times the individual company win rate, and almost half the cost to bring a technology to market.
For a science project the gates therefore test different things. Instead of a signed letter of intent, the first gate is often a patent filed before a conference talk destroys novelty. Instead of a pricing test, it is a grant or public funding line secured so the next experiment does not require a priced round. Instead of a landing page, it is a pilot unit that runs in the field. The decision logic is identical: evidence first, entity later. The evidence is just heavier.
What EX EPIC builds before a company exists
This is the model EX EPIC runs, and the portfolio is the proof that projects do become companies. Zero-X, the waste-to-energy venture, has designed, financed and installed systems across 11 countries, with more than 200 units deployed. EX IX, the patent liquidity marketplace, launched on a base of 24 patent families filed with roughly 100 more validated in the pipeline. EX FI provides FCA-authorised payment and financial infrastructure to the rest of the group, and what that authorised payment institution status actually means is explained on its own site. The operators who commission and run those projects come through the EX EPIC Academy, which has placed more than 250 of them into the portfolio, and the capital track record behind all of it exceeds 160 million euros across four continents.
None of those started as a company with a round. Each started as a technology the group decided to finance, patent and deploy, carried on the shared bench until it had units in the ground or filings in the office, and only then given its own structure. For what that bench contains in a science studio, see deep-tech venture builder; for the founder's side of it, venture studio for scientists.
How to tell a real build from a brand on the cap table
Five checks, drawn from the operators quoted above. A studio that passes all five is building. One that fails the first two is marketing.
- Can they show you the gates in writing, with thresholds, before you commit? Lehnert's founder check is whether the studio can show the calendar, not only the ambition.
- Did they say no to a bad fit recently? A studio that accepts every deck is not protecting its capacity for yours.
- Who owns Monday review and Friday scoreboard? Named people. If the answer is "the partners drop in", the weekly work is not happening.
- What does the last company that spun out still use from the studio? Playbooks, pipeline, architecture. If the answer is nothing, there is no inheritance.
- Can they describe a clean stop? Lehnert treats kill clarity as a positive outcome. Partnerships that cannot discuss ending struggle when markets turn.
If you are still working out which flavour of studio you are talking to, the venture studio model covers the archetypes and venture studio examples walks through named ones. The definition itself lives in what a venture studio is.
FAQ
Do venture studios build companies from their own ideas or from founders' ideas? Both, and the door matters. In a studio-led build the studio researches the gap, validates demand and recruits a founder into the project. In a founder-led build an experienced operator arrives with a raw insight and the studio supplies engineering, design and regulatory muscle around it. The founder-led route usually leaves more equity with the founder, because less of the project-phase risk sat with the studio.
When does a venture studio incorporate the company? When the evidence justifies a legal entity, not at day zero. In a typical software build that falls around month nine to twelve, after a pilot has produced usage data. Spin-out is a separate, later event, most often defined as the first outside funding round.
Who owns the technology while it is still a project? The studio, or the university or corporate that licensed it to the studio. The intellectual property is assigned to the new company at incorporation, together with the pre-seed capital, which is why that sequence is handled as a single stage rather than an afterthought.
What happens to projects the studio decides not to turn into companies? They are stopped, and good studios treat that as a result rather than a failure. In deep tech a mature project that does not fit a standalone company can also be sold directly to an acquirer that needs the technology, which is one of the exits the project model keeps open and the solo-company model usually closes.