AI for our most pressing crises in silicon
Complexity of chips is exploding. In turn, time to design and test chips is exploding1. Silicon supremacy is critical for national security2.
So we need to rethink how we predictably uncover drastically better chips, and with the help of AI, before we hit Moore's Wall3.
Normal is rethinking semiconductors from the ground up
We are building mathematical and physical intelligence to accelerate hardware innovation.
Fortune 500 companies use our AI systems to scale new hardware and enable faster time-to-market (case study).
Founded in 2022 by engineers at Google Brain and Google X pioneering AI for the physical world in the USA.
Ctrl+R: refreshing hardware
Prod. 2025: As a first step, we are building AI that can design and verify hardware logic from specifications alone. We are the first team to bridge auto-formalizing AI4 to silicon.
Prod. 2027: As a second step, we are widening the conventional design space, with thermodynamic computing (Nature). This means enabling custom silicon at the energy limits of artificial reasoning (Wired).
Prod. 2028: "Ctrl+R" refers to the safe recursive self-improvement step5: lastly, AI revises its own design specifications.
About us
We are a diverse, lean team of the world's best engineers, scientists, and operators, with offices in New York City, San Francisco, London, and Copenhagen.
We deliver our software and hardware with the largest semiconductor design and manufacturing institutions, a responsibility which entails individual ownership and deep partnership. We say "respect the opportunity" and "collaborate".
We've raised $35M in seed capital supported by Eric Schmidt (FSV), Celesta Capital, Drive Capital, the UK Advanced Research + Innovation Agency, Samsung Next, Intel Ignite, and the National Security Innovation Network.
The race is on to secure silicon supremacy, and to provide 1000x more intelligence per unit. Normal computing.
Faris Sbahi, Antonio Martinez, Yui-Hong "Matthias" Tan
March 24, 2025
Join us.
More updates coming soon.