The fundraising was led by Andreessen Horowitz and includes investment from new backers Arm Holdings Plc and M12, Microsoft Corp’s venture fund, Gimlet chief executive officer Zain Asgar said in an interview. The startup is also collaborating with Arm to make its software compatible with various forms of that company’s chip technology.

The round comes just six months after Gimlet raised US$80 million. It came together so rapidly because the company received several unsolicited term sheets from interested investors, Asgar said.

Gimlet’s customers include AI labs and companies in the financial services industry, he said, while declining to provide specific names. Gimlet started by focusing on software to carve up AI tasks and send them to different kinds of chips, but the company is now helping customers set up data centres and is also working on data centres of its own, Asgar said.

As the business worked to deploy its software, it quickly realised most customers didn’t know how to arrange data centre hardware to run these kinds of complicated multi-chip setups. For example, different types of chips often need varying kinds of cooling systems or temperatures. 

“We figured out all this like really cool tech around how to distribute workloads to different types of chips,” he said. “But one of the challenges that we kind of realised is that nobody builds data centres in a heterogeneous manner. So we kind of started peeling the onion one layer at a time.”

The company also has been focusing on one very specific AI task — running AI models superfast — because that is where a combination of different chips is yielding compelling results, he said. 

Less than a year ago, the company raised a US$12 million seed round backed by venture firm Factory and angel investors including Intel CEO Lip-Bu Tan, Figma Inc head Dylan Field and Andreessen general partner Raghu Raghuram, who is also leading that firm’s funding of Gimlet.

The area is getting more attention. Last month, newer rival Callosum raised US$100 million in early financing from backers including the UK’s public AI fund.

The “craziest thing” about the AI boom so far is that it has been built largely on a single technology stack anchored by Nvidia Corp chips, Raghuram said. That’s changing, and the market needs technology to help companies navigate that. 

“The history of computing is never that way, especially a market that's going to be trillions and trillions of dollars and so many different use cases,” he said. “So, fundamentally, we think the world will be multi-silicon, and we think the world will have multiple different types of workloads with very, very different demands.”

Uploaded by Liza Shireen Koshy