CuspAI Raises $450M at a $2.6B Valuation: NVIDIA, Meta and Tokyo Electron Join Its 45+ Partner AI Materials Foundry

CuspAI, a Cambridge, UK-based startup that uses AI to discover new materials, announced on July 20, 2026 that it has raised a $450 million Series B round at a $2.6 billion valuation¹². Kleiner Perkins and NEA co-led the round, with significant participation from Jeff Bezos’ investment fund, Bezos Expeditions¹.

Alongside the funding, the company launched the AI Materials Foundry, a materials discovery network with more than 45 founding partners including NVIDIA, Meta, and Tokyo Electron¹. The deal is a marker of how AI investment, so far concentrated in language models and coding tools, is expanding into the physical world of materials R&D.

What CuspAI Does: Designing Materials That Don’t Yet Exist

CuspAI combines frontier AI with domain expertise, exclusive data access, and customer partnerships to discover materials that do not yet exist, addressing what it sees as a critical bottleneck in industries such as semiconductors, energy, and climate¹. The company’s premise is that “the world needs materials that don’t yet exist” to advance semiconductors, clean energy, and advanced manufacturing².

Its platform, called MIRA, is designed to cover the full discovery cycle — from generative materials design through simulation, synthesis route planning, and coordinated experimental validation².

The company was founded by Dr. Chad Edwards and Prof. Max Welling¹. Its advisory board includes Abhi Talwalkar, an AMD board member, along with deep learning researchers Yann LeCun and Geoffrey Hinton². John Giannandrea, a former Apple and Google executive, has joined to establish the company’s foundry operations in the United States².

The AI Materials Foundry: 45+ Partners Pooling Data, Labs, and Compute

A centerpiece of the announcement is the AI Materials Foundry — a global network that combines data, laboratories, computing resources, and scientific expertise to accelerate the design of new materials, with more than 45 founding partners¹.

The partner list includes NVIDIA, Meta, Samsung, Hyundai Motor Group, Henkel, Applied Materials, Tokyo Electron, and Lam Research — major names across semiconductors and manufacturing¹. With semiconductor equipment maker Tokyo Electron on board and an office in Tokyo (alongside Singapore, Cambridge, Amsterdam, Berlin, and the United States)¹, the coalition also has meaningful ties to Japan’s manufacturing sector.

The Investors: Bezos Plus a Government-Backed Fund

Beyond the co-leads and Bezos Expeditions, the round included Glade Brook Capital Partners, Lux Capital, AMD Ventures, Tru Arrow Partners, StepStone, and veteran investor John Doerr. Notably, Britain’s government-backed Sovereign AI Venture Fund and Invest-NL of the Netherlands also participated¹². Existing investors Temasek, Basis Set Ventures, Giant Ventures, Touring Capital, Prosus, Phoenix Court, and Northzone also joined¹.

The funds will be used to expand across the US, Asia-Pacific, and Europe².

AI Money Moves From Models to Applied Infrastructure

July 2026 has seen a run of large AI financings. Data platform Databricks is raising at a $188 billion valuation, AI chipmaker SambaNova raised $1 billion with JPMorgan Chase as an on-prem inference customer, and a new financing structure emerged with a debt facility of up to $400 million collateralized by inference chips.

Against that backdrop, CuspAI’s raise shows AI investment reaching beyond language models and developer tools into scientific R&D — the “AI for science” wave. The participation of a government-backed fund suggests that materials discovery capability is starting to be framed as a matter of industrial competitiveness and economic security.

Ripple Effects for Materials and Manufacturing R&D

Materials discovery sits upstream of R&D in many industries — semiconductors, batteries, catalysts, and more. If AI-driven discovery cycles reach practical maturity, they could change how materials and manufacturing companies run R&D and make investment decisions. Tokyo Electron joining as a founding partner of a global coalition signals that this field is directly relevant to Japan’s industrial players as well.

The distance between AI and scientific research keeps shrinking. If your organization is evaluating AI for R&D, it is worth noting that the scope now extends beyond documents and code to simulation and experimental planning.