Nicolas Sauvage believes it takes four years for the best bets in AI to look obvious. He shared this insight at StrictlyVC's San Francisco event last week.

Since founding TDK Ventures' corporate venture arm in 2019, he’s focused on what he calls 'the boring parts of AI.' The highest-profile example is Groq, an AI chip startup valued at $6.9 billion, which bet big on inference before it became trendy.

Sauvage looks for technologies that have become more interesting to VCs in the last year: solid-state grid transformers and sodium-ion batteries for data centers, among others. He’s particularly interested in physical AI with a clear purpose—robots performing specific tasks like moving goods or working in hazardous environments.

He argues CPUs are poised for a comeback as inference chips like Groq reshape how models speak, and is keeping an eye on China's rapid manufacturing advancements driven by AI-assisted prototyping. According to him, dexterity remains the unsolved problem that could tip the balance of technological advantage.