AI Eating The World? Sima.ai Is Selling Teeth ... And Raised $150 Million To Sharpen Them

“It’s a $5 trillion Nvidia and then it’s us,” Sima.ai founder and CEO Krishna Rangasayee told me last week. “My world is very simple.”

Sima.ai is a physical AI chipmaker, and now is a very good time to be in physical AI. The space is exploding right now with hundreds of billions of dollars in investment. Haomiao Huang, founding partner at Matter Venture Partners, explained why in a recent conversation, saying that “if AI is gonna eat the world, hardware will be its teeth.”

The news today is that Sima.ai has raised a $150 million Series C at a $1.45 billion valuation, bringing its total funding to $500 million. The story behind the story is that the 200-employee seven-year-old startup is just now hitting an inflection curve where its chips will power medium-sized-to-large drones (growth industry), smart-ish to near or full self-driving cars (growth industry) and all kinds of robots from quadrupeds to humanoids (also a growth industry).

And, the company says, its chips will do more than Nvidia for less cost using less power and generating less waste heat, because its chips are optimized for physical AI, not AI inference. (That’s where Nvidia reigns supreme for now, but it’s only a fraction of what a robot needs in its compute stack.)

“Our primary competition has been Jetson Orin class of products,” Rangasayee says, referencing Nvidia’s family of chips and kits for next-generation robots that offer up to 275 trillion operations per second of performance. “And so we are typically twice the performance of Nvidia on tokens per second or frames per second, but we are 1/5th the power.”

A robotics CEO once told me that compute in a humanoid robot can demand up to 40% of the robot’s power draw, so that’s fairly significant.

Of course, these are Sima.ai’s numbers, not independent benchmarks. But they speak to a real problem. Nvidia’s newer Jetson Thor platform is powerful, and plenty of the humanoid robot companies I talk to use it. It’s also expensive, and it’s power-hungry.