Key Points
CuspAI raised $450M at $2.6B valuation on July 20, backed by Bezos, Nvidia, Meta.
AI platform compresses material discovery from years to weeks using machine learning simulations.
Data center gallium demand could exceed supply by 2030, with China controlling 99% of refined gallium.
Coalition of 45+ partners including Nvidia, Meta, Samsung pools resources to accelerate material testing.
Jeff Bezos has joined a $450 million funding round in CuspAI, an AI startup that uses machine learning to discover new materials for computer chips. The July 20 round valued the two-year-old Cambridge company at $2.6 billion and brought total funding past $580 million. Bezos Expeditions backed the deal alongside Kleiner Perkins, NEA, AMD Ventures, and the UK Government’s Sovereign AI Venture Fund. CuspAI’s platform compresses material discovery from years of lab work to weeks or months, addressing shortages in gallium, iridium, and tantalum that are constraining AI infrastructure growth.
Why chip makers are hunting for new materials
Data centers powering AI systems depend on rare elements that are becoming scarce and expensive. The International Energy Agency projects that data center demand for gallium could exceed current global supply by 2030. China controls 99% of refined gallium production, exposing tech companies to geopolitical risk. Iridium and tantalum face similar supply pressures as AI computing expands worldwide.
How CuspAI’s AI platform works
CuspAI uses generative AI and advanced simulations to predict which molecular combinations will produce the properties engineers need. Instead of scientists mixing compounds by hand over months or years, the platform identifies candidate materials virtually in weeks. The company cannot produce metals directly but can search for alternative materials with similar properties or help scientists develop new compounds that were previously difficult to synthesize.
The AI Materials Foundry coalition launches
CuspAI unveiled the AI Materials Foundry, a coalition of over 45 founding partners including Nvidia, Meta, Samsung, Applied Materials, and Hyundai. The coalition pools computing power and lab data so teams can test and verify new materials faster than any single company could manage alone. Co-founder Chad Edwards said chipmakers are “frantically searching for new materials” to keep pace with AI infrastructure demands.
Bezos’s bet on materials science as the AI bottleneck
Bezos Expeditions has also invested in Prometheus, an AI startup helping engineers design and manufacture complex physical products. These investments signal that Bezos believes AI’s next major opportunity extends beyond software and chatbots into engineering and materials discovery. CuspAI was founded in 2024 by Dr. Chad Edwards and Prof. Max Welling and initially focused on carbon capture and water purification before pivoting to semiconductor materials as AI demand surged.
Final Thoughts
CuspAI’s $2.6 billion valuation and backing from Bezos, Nvidia, and Meta reflects growing urgency around chip material shortages. If the AI platform can compress material discovery timelines, it could unlock supply chains that are currently constraining AI infrastructure growth.
FAQs
Data centers need rare elements like gallium and iridium that are in short supply. CuspAI uses AI to discover new materials faster, potentially solving bottlenecks slowing AI infrastructure expansion.
CuspAI closed a $450 million Series B round on July 20, 2026, valuing the company at $2.6 billion with total funding exceeding $580 million.
The AI Materials Foundry includes over 45 partners: Nvidia, Meta, Samsung, Applied Materials, Hyundai, and others pooling computing power and lab data to test new materials.
No. CuspAI cannot produce metals directly. It identifies alternative materials with similar properties or helps scientists develop new compounds that are easier to synthesize.
Disclaimer:
The content shared by Meyka AI PTY LTD is solely for research and informational purposes. Meyka is not a financial advisory service, and the information provided should not be considered investment or trading advice.
About Author

Huzaifa Zahoor
Co FounderHuzaifa Zahoor is the engineer who built Meyka. He has spent years writing Python, training AI models, and building data pipelines specifically for financial markets. His technical articles have reached over 30,000 readers on Medium, so he knows how to make complex things easy to follow. If this article touches on how the tools work, he is the person who actually built them.
What brings you to Meyka?
Pick what interests you most and we will get you started.
I'm here to read news
Find more articles like this one
I'm here to research stocks
Ask Meyka Analyst about any stock
I'm here to track my Portfolio
Get daily updates and alerts (coming March 2026)