Ricursive Intelligence is betting on AI-designed chips to help solve the compute crisis
The founders detailed how their startup is developing AI that can design chips, learn from each iteration, and improve subsequent designs, all with the goal of automating and accelerating the chip design process.
Key points
- Ricursive Intelligence founders Anna Goldie and Azalia Mirhoseini presented their AI chip design startup at TechCrunch Disrupt on Wednesday.
- Goldie said compute demand could grow at least 10,000x in five years, requiring more efficient chips.
- Their previous AI system AlphaChip produced a donut-shaped chip layout that surprised engineers but proved successful.
At TechCrunch Disrupt on Wednesday, Ricursive Intelligence founders Anna Goldie and Azalia Mirhoseini introduced their startup's approach to using AI for chip design. They said the goal is to automate and speed up chip design by having AI learn from each iteration and improve subsequent designs.
Goldie said there is an understanding that in the next five years demand for compute will grow at least 10,000 times, and that this may be conservative. She said the company does not want to build 10,000 times more data centers, but instead wants to help make chips more efficient in using bandwidth, wafers, and fabs.
AI-designed chip layouts
The founders said AI is challenging traditional chip design by finding unconventional layouts that outperform traditional ones. They cited AlphaChip, an AI system they previously developed, which produced a circular, donut-shaped arrangement of chip components that initially surprised engineers but ultimately proved successful.
They explained that they optimize entire chips rather than partitioning them into dozens of smaller blocks. This allows AI to create intricate patterns, including wave-like designs, which they said they plan to share more about in the future.
There is this understanding that in the next five years we are going to have at least 10,000x more demand for compute, and that’s somewhat potentially conservative
— Anna Goldie
Human role and future chips
Goldie argued that letting AI handle chip design will allow humans to focus on developing applications and exploring new possibilities. She said the shift is toward humans bringing creative ideas while AI handles execution, and that this is happening in chip design, software development, and other areas.
The founders also said AI-designed chips will make it possible to create all sorts of chips that might not have been possible before. Goldie noted that if designing a new chip takes six months or a year or two years, companies must be extremely selective, but if a chip can be designed in about a week, custom chips for applications in space, AR, VR, and biological computing could make financial sense and be more energy efficient.
