Term sheet paradox: why DeepTech founders aren't getting AI's leverage

DeepTech is now 45% of UK venture capital investment value, up from 14% in 2021. The 2026 HSBC Innovation Banking Term Sheet Guide reports 71% of UK VCs optimistic about the year ahead, with the same proportion expressing strong conviction in DeepTech specifically, a thirty point year-on-year jump. The headlines say it clearly enough: DeepTech is in favour.

So why are term sheets getting tougher?

That is the question buried inside HSBC's 2026 analysis, drawn from 711 UK term sheets and £11.2bn of investment value, and the question almost no public commentary on the report has surfaced. The data describes a market that is simultaneously more optimistic in sentiment and tougher in the terms attached to that sentiment. Investor optimism is real. So is the cost of accepting it, paid in the small print of the term sheet.

For UK DeepTech founders raising in 2026, the most useful thing to understand is not the headline mood music but the three axes along which the market has quietly bifurcated underneath it. Stage. Sector. Geography. On each axis, most early-stage founders are sitting on the wrong side of the line.

HSBC's own data names the disconnect

The Investor Pulse survey embedded in the 2026 report covers 41 VCs and is unusually candid. 71% are optimistic about the year ahead. 41% are prioritising new investments, up from 20% a year earlier. Conviction is concentrated in AI (78%) and DeepTech (71%, up thirty points year on year).

In the same survey, 53% expect founders to push for more aggressive valuations, while 71% expect the rest of the term sheet to remain broadly unchanged: the preferences, the protections, the consent rights, the clauses that govern what happens at exit. HSBC describes the result, in its own words, as "a growing disconnect between pricing expectations and structural discipline". The bank's CEO Emily Turner goes further, writing that negotiations in 2026 "are likely to become more polarised".

This is unusual language for a corporate banking publication. It is also exactly right. Investors are willing to deploy more capital. They are not willing to soften the terms that come with it. The interesting question, which the report stops short of answering, is what this polarisation looks like in practice for DeepTech founders specifically, because the deeper findings get more uncomfortable from here.

Tighter at Seed, looser at the top

HSBC's data shows participating preference shares at Seed doubled year on year, from 7% to 14% of deals. Participating preferences pay the investor their cheque back and a pro-rata share of remaining proceeds at exit; non-participating pays one or the other. In modest exits, where total exit value sits within a few multiples of capital invested, the difference materially reduces founder proceeds. In blockbuster exits, the difference fades. Seed syndication also rose from 26% to 31%, investor consent rights from 76% to 85%, and founder warranties from 46% to 62%.

At the other end of the market, Series B participating preferences fell from 17% to 10%, Series C+ participating preferences more than halved from 15% to 6%, and median Series C+ post-money valuations rose 35% in twelve months, from £191.3m to £258.4m.

The market has not become more or less founder-friendly in 2026. It has done both, at different stages. Capital is concentrating in fewer, larger, proven companies where competition forces investors to soften their terms. Elsewhere, the terms are getting tougher. The decision rule is uncomfortable: the protections in your term sheet depend less on the quality of your company than on which stage of the bifurcated market you happen to be raising in. This connects to the dilution paradox we examined in November.

DeepTech term sheets are not AI term sheets

This is the finding most easily missed in the headlines, and the one that should most concern early-stage DeepTech founders. Within the HSBC dataset, pure AI term sheets feature non-participating preferences 97% of the time. DeepTech term sheets, excluding pure AI, feature them only 83% of the time. Translated, participating preferences appear in 17% of DeepTech rounds versus 3% of pure AI rounds. Investor consent rights are higher in DeepTech (86%) than in AI (74%). Founder leaver provisions are higher (56% versus 46%). Syndicate deals are more common (42% versus 31%). HSBC's own commentary describes DeepTech term sheets as "more protection-led than pure AI", signalling "greater investor caution and a longer-risk investment horizon".

This is the empirical answer to a question we asked in April, about what happens when investors say they want DeepTech but apply SaaS-era criteria. The answer is sharper than expected. The criteria are not simply SaaS-era. They are specifically tougher than the criteria applied to AI.

The "DeepTech is in favour" narrative does not translate into uniformly favourable terms. Pure AI founders are extracting genuinely loose terms in 2026. DeepTech founders are receiving the optimism without the term sheet to match.

Within Series A, two different markets

The headline Series A story in 2026 sounds like a triumph. AI and DeepTech rounds are reaching sizes once reserved for Series B. HSBC reports Isomorphic Labs completing a £450m Series A, the UK's largest of the year, and estimates 10-15% of UK deals are now "super-sized" Series As above £30m, where founders "typically have greater negotiating power over terms".

The trouble is that this top decile is dragging the narrative upward while the actual market median is moving the other way. The median UK Series A post-money valuation in 2025 was £17.1m. That is down 5% on the £17.9m median a year earlier. Participating preferences at Series A did fall, but only from 12% to 9%: a real improvement rather than a transformation.

The implication is unflattering for founders who benchmark themselves against the breakouts. Unless your company sits in the narrow super-sized cohort, the Series A you are negotiating in 2026 is at a lower valuation, with marginally tighter terms, than the equivalent round a year earlier. Two markets, one stage, one widely misread headline.

The London standard is not your standard

HSBC's data shows 51% of UK Seed rounds in 2025 took place outside London, with the regional concentration heaviest precisely in DeepTech-aligned sectors: CleanTech (61% in regions), Life Sciences (60%), Hardware (57%), Energy (56%).

In London, 93% of term sheets include non-participating preference shares. In the Midlands, only 73% do. In the North of England, 77%. The investor mix is different too. London is 71% led by traditional VC and LP funds, with VCT and EIS exposure at around 5% combined. The Midlands is 33% VC/LP, with 28% VCT and 28% EIS by deal count.

HSBC's own framing on this is unusually direct: founders raising outside London "should expect a wider mix of capital sources and more negotiation on downside protection and structure, rather than a single 'London-standard' template". This matters more for DeepTech than for any other sector, because DeepTech concentrates precisely where the structural pressure on terms is highest. The clean, non-participating, simple-cap-table London standard is not what most DeepTech founders raising outside London should expect to negotiate against.

The handover gap is where DeepTech founders lose value

The underlying fact running through all three axes is the gap between the capital pool that funds early-stage DeepTech and the pool that scales it. HSBC puts it bluntly: the UK is "well-capitalised at Seed and Series A" but has "less domestic capital available for £30m+ cheques, with many UK funds capping out at Series B". UK investors led 72% of Seed and 54% of Series A in 2025. At Series C+, that figure collapses to 21%, with US investors leading 37% and contributing £3.1bn of inbound investment, up from £1.5bn a year earlier. 67% of UK investment value came from foreign-led deals.

The 2026 European DeepTech Report confirms the same pattern at European level, extending the scale-up gap we covered last September.

The strategic implication is uncomfortable. Consider the DeepTech founder raising in the regions, at Seed, in a non-pure-AI category: they are accepting tighter terms today from a pool of domestic investors whose follow-on capacity and return models differ materially from the US growth funds they will need to convert at Series B. The compounding cost of that gap is only visible at exit, when it is too late to renegotiate.

The terms you accept today are the terms you live with at exit

The 2026 HSBC data is revealing for anyone willing to read past the headlines. It describes a market genuinely optimistic about DeepTech while tightening every term it has leverage over, at the stages, sectors, and regions where the bifurcation lands most heavily.

The sophisticated DeepTech founder reads the terms, not the optimism. The participating preference accepted at Seed will look expensive to a US growth fund at Series B. The syndicate composition agreed today determines who can lead the next round.

We wrote in April that the negotiation can be lost before the term sheet arrives. The 2026 data extends the point. The terms that determine your exit are not the ones you negotiated hardest for. They are the ones you let pass. Which side of the bifurcation you sat on will be visible then, whether or not it feels visible now.

 
Let's talk.