Deeptech Investment Thesis: What It Must Answer in 2026

A deeptech investment thesis is a written filter. It tells a fund's partners, its limited partners and the founders who pitch it which companies the fund will back and, more usefully, which ones it will turn down before the first meeting. Search for the phrase and you will find a row of venture firms describing themselves. Each page is sincere, and each is a self-portrait. What none of them does is separate the questions every thesis must answer from the choices that make one fund different, or set those claims beside the data on how the asset class actually behaves.

This piece does both, and then turns the thesis around to show how a university research team or a young spinout should read one before asking it for money.

What a deeptech investment thesis is for

The category is no longer a niche bet. Deep tech now claims a stable 20% share of venture capital funding, up from about 10% a decade earlier, according to BCG's investor guide to deep tech. In Europe the shift is sharper: deep tech took 32% of all European venture funding in 2025, more than double its 15% share in 2015, as reported in Dealroom's European Deep Tech Report 2026.

When a fifth or a third of the market calls itself deep tech, the label stops doing any filtering on its own. A thesis has to do the work the label no longer does. That means defining the science it understands, the stage at which it can judge risk, and the length of time its capital can wait. If the distinction between genuine science risk and ordinary engineering risk is still fuzzy, EX Venture's explainer on deep tech versus high tech draws the line cleanly, and our own primer on what counts as deep tech investment covers the basics.

The five questions every thesis answers

Strip the branding from the published theses and the same five questions remain. Funds answer them differently. None of them skips one.

Is the science hard to copy

Every thesis starts with defensibility, because it is the only reason a long, expensive development cycle can pay. Octopus Ventures lists defensible, IP rich technology as one of the three traits behind its successful deep tech investments, alongside the team and the market. Inflexor Ventures turns that into a test anyone can apply: if an incumbent could replicate the solution in three to six months by assigning a few engineers to it, it does not qualify.

The test is useful precisely because it ignores patent counts. A portfolio of filings that a competitor can design around is weaker than a trade secret that took a laboratory a decade to reach.

How far is the technology from the lab

The second question is technical maturity. Juniper, which invests in bioindustrial companies, states that for physical products it typically invests at technology readiness level 3 or higher, where proof of concept has been demonstrated experimentally. Software and AI companies are judged instead on utility, differentiation and data moat.

A thesis that does not name its readiness threshold forces every founder to guess, and forces every partner meeting to relitigate the same point.

Why now

Juniper also splits timing into two kinds. Technology timing is an inflection that changes the cost, speed or scope of what is possible. Market timing is a window opened by supply chain constraints, regulation or another external shift. A good thesis names which kind of timing it is betting on, because the two fail differently: technology timing fails when the science stalls, market timing fails when the window closes before the product is ready.

Who buys it, and who validates that

Here deep tech departs most sharply from software. Revenue arrives late, so it cannot be the main evidence. Mela Ventures puts it bluntly: revenue is a lagging indicator of value, and value grows as the product crosses milestones that retire technical, manufacturing and adoption risk.

Ventures.eu defines traction for deep tech as design partners, pilot data or published validation that holds up outside a controlled environment. It routes both the technology and the demand through corporate partners who would eventually license, pilot or acquire it, and says that step compresses what would otherwise be 12 to 18 months of independent diligence into weeks. Whatever the mechanism, a thesis needs one: a named way of learning whether a real buyer wants the thing, before the fund is committed.

Can the fund stay in long enough

The last question is patience, and it is a question about fund structure rather than temperament. BCG found that deep tech investments take an average of 25% to 40% more time between funding stages, from seed through Series D. Id4 Ventures cites an even more specific penalty: roughly 18 months longer to reach Series A traction, with 35% more time and 48% more capital spent on R&D before early revenue.

A thesis that promises patience from a fund with a standard life and no reserves for follow-on rounds is making a promise its structure cannot keep.

The numbers a thesis has to be honest about

Returns

The reassuring headline is that deep tech does not cost returns. BCG's analysis of about 1,100 venture funds found a weighted average internal rate of return of 26% for deep tech focused funds against 21% for traditional venture investors over five years. The less quoted detail matters as much: on a nonweighted basis the figures were 25% and 26%. The honest reading is parity, with the best deep tech funds pulling the weighted average up. Selection, not the label, produces the return.

Two findings from the Hello Tomorrow and Deepbright Ventures study of more than 860 startups are worth building into any thesis. Hardware focused startups delivered a gross IRR of 27% against 13% for software. University spinoffs reached 26% against 23% for startups without an academic origin, though the authors note that success is not limited to academic founders.

Where the money is going

Capital is concentrating fast. Dealroom reports that US deep tech funding grew 92% to $144bn in 2025, with AI model makers accounting for more than half of that total, while Europe grew 21% to $20bn. Within Europe, the UK led with $5.2bn, ahead of France at $3.9bn and Germany at $3.2bn, and defence, security and resilience startups took 43% of deep tech venture funding, roughly twice their 2022 share.

For a thesis, the implication is to decide deliberately whether to follow that flow or to stand apart from it. A fund entering AI models or defence in 2026 competes for allocation. A fund outside those themes competes for attention.

The later stage gap

BCG notes that more than 80% of deep tech ventures build physical products, which brings engineering and unit economics risk, and that early investors must worry about securing later funding. Dealroom quantifies the European version of that worry: only 54% of funding for European deep tech rounds above $15m comes from within Europe, against 80% in the US. A seed thesis that does not say where Series B and C money will come from has left its hardest question unanswered. For a wider view of where this is heading, see our piece on where deep tech investment is heading.

Where theses actually differ

Once the five questions are answered, what distinguishes one fund from another is what it excludes.

• Stage. Id4 invests only at pre-seed and backs technical conviction before a finished business model. Ventures.eu invests at seed and Series A with tickets between €200,000 and €2,000,000.
• Sector. Juniper is bioindustrial. Ventures.eu names climate tech, hardware, ICT, cybersecurity and AI.
• Geography. Inflexor and Mela build their cases on Indian engineering talent and cost efficient R&D. Ventures.eu weights its portfolio roughly 60% toward Portugal.
• Founder model. Octopus writes about backing pioneers who have dedicated years to their research. Inflexor looks instead for a founding trinity of scientist, engineer and business strategist spread across two to four people, and names the lone founder as a warning sign.

None of these choices is right in general. Each is right for a particular pool of capital with a particular set of partners, and the thesis is where that fit is written down.

Reading a thesis from the research side

Most research teams read an investment thesis as a list of compliments to aim for. It is more useful to read it as a checklist of evidence the fund will ask for.

1. Answer the five questions before the meeting. State the replication time an incumbent would need, your readiness level, which kind of timing you are relying on, who has already tested the technology outside your lab, and how much capital reaches the next milestone.
2. Bring validation, not projections. A letter from a design partner or pilot data from a real site answers the buyer question. A revenue forecast five years out does not.
3. Filter funds by exclusion. If a fund invests at seed and you need pre-seed, or it names sectors that exclude yours, no pitch quality closes the gap.
4. Plan the later stage gap yourself. Grants, corporate partners and public co-investment can carry a spinout across the stretch where venture money thins out.

Our guide to turning research into a commercial company goes deeper on the spinout side of this process.

Testing a thesis in the room

Theses are written in partners' meetings and tested in conversations with the people they describe: researchers who know whether the science holds, and policymakers who decide whether the market will be allowed to exist. Those people are rarely in the same room as the investors.

EX Future Summit runs 18 to 20 November 2026 across three rooms at once: online, Las Palmas and Bali. Its program includes a Global Financial Raise track covering venture capital, private equity, national grants and Horizon Europe, and eight hosted meet-and-greet sessions matching universities with private partners. The event is built as a single thirty hour broadcast between two islands twelve hours apart, and online attendance is free for verified researchers and students. It is a first edition, so there is no track record to cite. What it offers is a room where a thesis meets the science and the regulation it depends on.

FAQ

How long does a deeptech investment take to pay back?

There is no single figure, but every source points the same way. BCG measured 25% to 40% more time between funding stages than other technology investments, and Id4 Ventures cites roughly 18 months longer to Series A traction. A thesis should assume a longer holding period and reserve capital for follow-on rounds.

Do deeptech funds return more than traditional venture funds?

Roughly on par. BCG found a 26% weighted IRR for deep tech focused funds against 21% for traditional venture funds, but 25% against 26% on a nonweighted basis. Manager selection matters more than the category.

Do investors prefer university spinouts?

Some data supports them. The Hello Tomorrow and Deepbright Ventures study found a 26% IRR for university spinoffs against 23% for other startups, while stressing that strong companies also come from outside academia.

What technology readiness level do deeptech investors expect?

It depends on stage. Juniper typically invests in physical products at TRL 3 or higher, once proof of concept is shown in the lab, while pre-seed specialists such as Id4 Ventures back technical conviction earlier.