Jeff Bezos and the UK Government Just Bet $450 Million on a Startup That Hunts for New Materials Using AI

Key points

• CuspAI, a two-year-old Cambridge startup, raised $450 million in a funding round reported by The Guardian, valuing the company at $2.6 billion as of 2025.
• Investors include Amazon founder Jeff Bezos and the UK government's sovereign AI fund, a state-backed pot of money set aside to support strategically important AI companies.
• CuspAI's software uses artificial intelligence to predict which combinations of materials could work for new technologies, without requiring years of lab testing.
• The company is targeting rare metals, the hard-to-source minerals that sit inside every modern chip and battery, aiming to reduce how many are needed.

Finding a new material used to take decades. Scientists would mix compounds, run experiments and try again, like searching a continent for a single buried coin.

CuspAI thinks AI can draw the map.

The Cambridge-based startup builds software that scans the near-infinite number of possible material combinations and predicts which ones are worth testing in a real lab. Think of it as a recommendation engine, the kind that suggests your next series to watch, but pointing at molecules that could end up inside the next generation of chips instead of TV shows.

The company just raised $450 million to push that idea forward. Jeff Bezos put in money. So did the UK government, through its sovereign AI fund, a pot of public cash set aside to back AI companies Britain considers strategically important. The round values CuspAI, which is only two years old, at $2.6 billion. We first covered the company's ambitions on 20 July 2026, in a piece on how AI labs are pulling in huge capital before they've shipped a product.

Every piece of technology you own depends on rare metals, minerals extracted from a small number of places on Earth and processed through fragile supply chains. The chips in your phone and the battery in an electric car both depend on getting the right materials in the right quantities.

That supply chain is a genuine vulnerability. One political dispute or one mine closure sends prices spiking. CuspAI's pitch is that smarter software could find alternative materials, or better combinations, that cut how much of the rare stuff manufacturers actually need.

Could this change what ends up inside your devices?

Not immediately, but the direction is real. If CuspAI's models correctly predict which new materials perform well, manufacturers could skip years of expensive trial-and-error. That means cheaper chips, batteries that last longer, and less dependence on minerals from politically unstable regions. It's also worth watching whether the UK government's involvement accelerates domestic supply-chain policy alongside the science, because the funding bet only pays off if the regulatory environment catches up.

With $450 million banked and some very heavyweight backers, CuspAI now has the runway to find out if they're right.