Jeff Bezos backs AI materials startup CuspAI in $400M round at $2.6 billion valuation

CuspAI, a UK startup using artificial intelligence to discover new materials, is reportedly raising $400 million in funding at a $2.6 billion valuation, according to the Financial Times. The round includes participation from Jeff Bezos’ family office, Bezos Expeditions, and venture capital firm Kleiner Perkins.

The reported financing would mark a major leap for the Cambridge-based startup. Less than a year ago, CuspAI raised $100 million. If the deal closes, the company’s valuation will have increased more than fivefold as investors place bigger bets on AI’s potential to accelerate scientific discovery.

“Jeff Bezos backs UK start-up CuspAI in a $400mn funding round. New financing will more than quadruple the two-year-old AI company’s valuation to $2.6bn,” The Financial Times reported.

Founded in 2024 by chemist Chad Edwards and AI researcher Max Welling, CuspAI is building what it describes as a search engine for materials. Its technology helps researchers discover new compounds for applications ranging from semiconductors and advanced manufacturing to clean energy and next-generation computing. The company’s goal is ambitious: to create AI systems capable of identifying the best possible material for virtually any application.

The Next AI Race May Be Materials Discovery as CuspAI Targets $2.6 Billion Valuation

That challenge matters far beyond academic research. New materials sit at the center of many industries, from semiconductors and clean energy to aerospace and advanced manufacturing. A chipmaker developing a new generation of processors may need materials that handle heat more efficiently. Battery manufacturers are constantly searching for compounds that improve energy density and lifespan. Finding those breakthroughs has traditionally required years of experimentation.

CuspAI wants to shorten that timeline.

The startup has developed a collection of AI models that generate candidate molecules with useful properties and then evaluate whether those materials are likely to perform as expected outside the simulation environment. The approach combines generative AI with scientific modeling, allowing researchers to explore far larger design spaces than conventional methods permit.

Its customer list reportedly includes major global companies such as ASML, Meta, and Hyundai, suggesting large enterprises are already testing AI-assisted materials discovery in real-world research programs.

The company has attracted an unusual concentration of scientific and AI talent. Geoffrey Hinton, widely known for his pioneering work in artificial intelligence and the recipient of the Nobel Prize in Physics, serves as an adviser. Yann LeCun, Meta’s chief AI scientist and founder of AMI Labs, is also involved with the company.

Edwards previously co-founded Cambridge Quantum, a company focused on quantum computing technologies. Welling is a professor of artificial intelligence at the University of Amsterdam and has spent years working on machine learning systems used across scientific disciplines.

CuspAI has been building technology infrastructure alongside its commercial products. In April, the company open-sourced a materials simulation toolkit called kUPS. According to CuspAI, the software can accelerate some simulation workloads by up to 49 times.

Material simulations play a central role in modern scientific research. Before a new material is manufactured and tested in a laboratory, researchers often use software to predict how it might behave. Those simulations can model everything from electrical conductivity and heat transfer to structural durability.

One challenge is that different scientific questions often require different simulation methods. Combining multiple simulation systems into a single workflow has historically been difficult. CuspAI says kUPS simplifies that process and improves performance by optimizing how simulation data is stored and processed on graphics processing units, or GPUs.

The company’s fundraising comes as interest in AI-driven scientific discovery continues to grow. Investors have poured billions into startups applying machine learning to biology, chemistry, pharmaceuticals, and materials science.

Large AI companies are moving into the same territory.

Earlier this month, Anthropic revealed research showing how Claude can assist chemists in analyzing data generated by nuclear magnetic resonance, or NMR, instruments. Those systems help researchers study the composition and structure of molecules and materials. Anthropic said AI can automate portions of a workflow that historically required significant manual effort.

That activity points to a broader trend across the AI industry. After years of focusing primarily on text, images, and software, many researchers are turning their attention to scientific discovery and the physical world.

CuspAI Team