22JUL

CuspAI raising $400m on Bezos money

4 min read

09:12UTC

CuspAI, a 2024 Cambridge AI-materials company, has signed term sheets for a reported $400m round at a $2.6bn valuation with Bezos Expeditions among the backers. The round is not yet closed.

← Back to CuspAI pools 45 partners' labs and compute Jump to analysis ↓

TechnologyDeveloping

Key takeaway

The fortnight's biggest UK round is foreign-led and not yet closed.

CuspAI, a Cambridge AI-materials company founded in 2024, is raising a reported $400m at a $2.6bn valuation, with Bezos Expeditions and Kleiner Perkins among the backers, according to the Financial Times ¹. Term sheets are signed; the round is still being raised and has not closed ². If it completes, the valuation lands at more than four times the company's September 2025 mark. Geoffrey Hinton and Yann LeCun, two of the people who built modern deep learning, sit on its advisory board.

The company solves a narrow problem. A customer specifies the properties it needs, a heat-resistant coating, a catalyst, a filter, and the platform generates candidate molecular structures and simulates them in silico before anyone touches a lab bench, compressing a discovery cycle that historically ran for years of wet-lab iteration. Meta is a customer for carbon-capture materials, Hyundai is another, and CuspAI worked with the Finnish chemicals group Kemira on materials to strip PFAS (per- and polyfluoroalkyl substances, the so-called forever chemicals) from water ³.

PhysicsX took its $300m Series C from Singapore's Temasek , and Orbital Industries closed an earlier Series B with no UK lead investor at all , the same shape the CuspAI round now repeats. British seed money is being rebuilt at the bottom of the market while the largest rounds are underwritten from Seattle and Singapore, which in practice means the founders stay in Cambridge while the upside accrues to investors abroad. AI-materials is the newest vertical to mint a near-unicorn this way, and it did so without a domestic lead.

Deep Analysis

In plain English

When a scientist or engineer needs a new material with specific properties ; say, a coating that resists extreme heat, or a filter that removes toxic chemicals from water ; the traditional approach is to mix and test compounds in a laboratory over months or years. CuspAI, a Cambridge startup founded in 2024, uses AI to do that work computationally: the computer generates and tests thousands of candidate molecules before anyone touches a lab bench. On or around 20 June 2026, the Financial Times reported that CuspAI is raising $400m from investors including Jeff Bezos's personal investment fund and the Silicon Valley firm Kleiner Perkins, at a reported valuation of $2.6bn. Term sheets have been signed, but the deal has not yet formally closed, so these figures are not confirmed. If the round does close at that valuation, it would make CuspAI worth more than four times what investors valued it at in September 2025. PFAS ; per- and polyfluoroalkyl substances, sometimes called 'forever chemicals' ; are a class of industrial compounds that persist in the environment and have been linked to health problems. CuspAI is working with the Finnish chemicals company Kemira to find materials that can remove PFAS from water, which is one reason industrial customers are interested even at this early stage.

Deep Analysis

Root Causes

AI-materials discovery requires a specific combination of frontier compute (for running molecular simulations at scale), large curated materials datasets, and the ability to hire researchers at the intersection of computational chemistry and machine learning.

UK universities produce the researchers; the compute and dataset infrastructure is concentrated in the US. Kleiner Perkins and Bezos Expeditions have direct relationships with the hyperscaler infrastructure that makes the product work, making them strategically useful LPs beyond the capital itself.

The valuation step-up mechanism ; more than four times the September 2025 mark ; reflects a commercial pull from industrial customers that was not visible at the earlier round. Meta's carbon-capture materials work, Hyundai's materials R&D, and Kemira's PFAS-removal brief are enterprise contracts large enough to reset the growth-capital case, and US growth investors are faster at repricing when enterprise traction signals product-market fit.

What could happen next?

Consequence
A successful CuspAI close at $2.6bn resets UK AI-materials valuation benchmarks and is likely to attract further foreign capital into Cambridge's materials-science cluster within six to twelve months.
Short term · Reported
Risk
The round remains open and unconfirmed; a failure to close at the reported valuation, or at all, would damage the materials-AI narrative and may chill the wider UK deeptech valuation environment given how closely this round has been covered.
Immediate · Reported
Opportunity
PFAS-removal materials face a compliance-driven commercial demand curve from EU Water Framework Directive enforcement beginning in 2025–26; CuspAI's Kemira work positions it ahead of several years of regulation-mandated purchasing.
Medium term · Assessed

Source Landscape

This story draws on neutral-leaning sources

Primary parallel: ARM Holdings in the 1990s established the model that British deep-Science IP scales on foreign capital while remaining headquartered in the UK.

ARM's Japanese SoftBank acquisition in 2016 and its eventual Nasdaq listing in 2023 traced a path where the scientific and engineering talent stayed in Cambridge, the returns accrued to successive foreign owners, and the UK government's response was limited to Listing Rules reform rather than ownership stakes. CuspAI's reported Bezos-led raise follows the same mechanism at an earlier stage: UK-invented AI-materials Science, Cambridge-based team, US-accruing equity upside.

Counter-parallel: DeepMind's 2010–2014 trajectory offers a different read. DeepMind raised its first rounds entirely from UK and European backers ; Founders Fund, Horizons Ventures ; before Google acquired it in 2014.

The acquisition produced the founder talent pool that built Isomorphic Labs, Wayve, PhysicsX and Conduct. A foreign-led round at the growth stage does not preclude the same talent-recycling effect if the company eventually spins out a generation of founders who build domestically.

A $2.6bn reported valuation on a company founded in 2024 implies that the round is priced almost entirely on future revenue potential, not current earnings.

AI-materials compresses a wet-lab discovery cycle that historically cost £5–20m and took two to four years of iteration per candidate compound, into a model-generation-plus-simulation pipeline that can screen thousands of candidates in days. The commercial margin, if the product delivers, is extremely high: industrial customers pay for R&D outcomes, not the compute hours that produce them.

Kemira's PFAS-removal contract sits inside a compliance-driven demand curve. EU member states face a PFAS remediation legal deadline under the Water Framework Directive revisions, with national regulators beginning enforcement action in 2025; any validated PFAS-removal material reaching market in 2026–27 enters a compliance-driven demand curve rather than an elective one. That regulatory pull may justify the valuation step-up more than the carbon-capture or automotive applications.

Consensus view: Dealroom's European AI investment tracker and the ScaleUp Institute's 2025 deeptech brief both document a structural pattern where UK AI deeptech at the growth stage is led by US and Asian capital: among rounds above $100m in UK AI since 2024, fewer than 15% have a UK-domiciled lead investor.

CuspAI's reported Bezos Expeditions and Kleiner Perkins lead continues that pattern. Dealroom notes that materials-AI ; generative models for molecular discovery ; is accelerating fastest among UK deeptech verticals, with CuspAI alongside Orbital Industries ($50m, no UK lead, and PhysicsX ($300m, Temasek-led, confirming that foreign capital leads UK Science at scale.

Counter-view: Nesta's 2025 deep-tech commercialisation review argues that foreign-led rounds are not a market failure but a rational market signal: UK public research produces frontier Science, but UK institutional investors historically underweight patient, deep-Science exposure at the growth stage, so US capital fills the gap at commercial terms that British pension funds and family offices have not yet matched.

Nesta's position is that the founders keep the IP, build in Cambridge, and hire in the UK regardless of where the equity accrues.

Key tension: Whether the foreign-capital-leads-UK-deeptech pattern represents a permanent structural feature of the UK innovation economy, or a transitional state that the British Business Bank's expanding direct-investment mandate and new Growth Capital Initiative can eventually displace.

First Reported In

Update #9 · Private money rebuilds Britain's seed tier

Sifted· 24 Jun 2026

Read original →

Causes and effects

Caused by

PhysicsX hits $2.4bn on Temasek cash

PhysicsX's $300m Temasek-led round established the pattern of US/Asian capital leading the UK's largest AI deeptech rounds, which CuspAI's reported raise continues.

Occurred 8 Jun 2026

Read story →

Orbital's $50m raise has no UK lead

Orbital Industries' $50m with no UK lead investor is part of the foreign-capital-leads-UK-top-of-barbell pattern that CuspAI's reported Bezos/Kleiner round reinforces.

Occurred 28 May 2026

Read story →

This Event

CuspAI raising $400m on Bezos money

The fortnight's largest UK round is led from abroad, extending the pattern of US and Asian capital underwriting the top of the British market.

Led to

CuspAI closes at $2.6bn, EU fund circles

CuspAI's late-June close confirms the $2.6bn valuation set when term sheets were signed on 20 June.

Occurred 2 Jul 2026

Read story →

CuspAI names 45 partners for lab access

June's $400m term sheets at the same $2.6bn mark were formalised in the 20 July close and partner announcement.

Occurred 20 Jul 2026

Read story →

Different Perspectives

EQT

EQT was reported on 3 July to be in advanced talks for a further CuspAI stake on behalf of the EU's Scaleup Europe Fund, but does not appear on the round CuspAI closed. Whether Europe's own sovereign vehicle secured a position, was outbid, or withdrew is unresolved.

Invest-NL

Invest-NL joined CuspAI's $450m round as a new state investor alongside Britain's Sovereign AI Unit, named in the company's own 20 July announcement. Any framing of British state capital arriving must also credit Dutch capital with the same access, since Invest-NL took the opportunity too.

Kleiner Perkins and Bezos Expeditions

Kleiner Perkins and Bezos Expeditions co-led CuspAI's $450m Series B at a $2.6bn valuation, a figure Lowdown has now reported three times since June. Silicon Valley and a family office, not a UK investor, set the price of Britain's highest-profile materials-AI company.

British Business Bank

British Business Bank proposed a £100m debt fund for aerospace suppliers at Farnborough on 20 July, co-designed with Airbus, Rolls-Royce and GKN Aerospace. It is the Bank's first non-equity instrument this cycle, aimed at suppliers who can service a loan but have no equity story, though the structure remains unfinalised.

Institute of Physics

The Institute has long argued STFC's national-laboratory infrastructure, not its grant programmes, is the binding constraint on UK physics output, and warns mothballing capacity like Clara removes capability that cannot be rebuilt on a four-year cycle. It represents the discovery-science community absorbing the reallocation the Bank's equity cheques do not touch.

Helsing

The Munich-headquartered defence-AI firm chose Plymouth over Continental sites for a £350m manufacturing plant building underwater surveillance gliders, alongside its record raise. Its choice of postcode signals confidence in UK manufacturing capacity for defence hardware even as it looks abroad for the capital financing that hardware.