Volantis raises $88M to bring iPhone’s AI memory trick to data centers

San Francisco semiconductor startup Volantis has raised $88 million to turn a memory approach already used in hundreds of millions of iPhones into a solution for AI hardware. The funding, disclosed Thursday, targets what the company describes as the inefficiency of moving data between processing units and separate memory chips.

The round stands out more for its approach than its size. Volantis is betting on a method that performs calculations closer to where data is stored, rather than relying on the standard way chips handle memory today. The idea isn’t new—Apple has used similar techniques in its chips to speed up machine-learning tasks—but Volantis wants to adapt it for AI workloads beyond phones. If it works, the startup could cut the power and delay problems that currently make large language models expensive to run at scale.

The risk is real. While the concept has been studied for years, it has rarely been used in real products because of software compatibility and manufacturing challenges. Volantis hasn’t named any customers or partners, and its $88 million funding looks small compared to the $231 million raised by Dutch competitor Euclyd last week. That gap shows how competitive the AI chip market has become, even as memory inefficiency remains a widely recognized problem.

The startup’s advantage is that its technology is already working in consumer devices. The real test will be convincing cloud providers and AI hardware makers to adopt a solution where every bit of speed and power matters.

Investors don’t seem worried. The round was led by existing backers, signaling strong confidence rather than a broad market bet. That’s common in chip funding—firms often commit more money before seeing real-world results. For Volantis, the next six months will decide whether its lab results can translate into actual deployments. If it lands even one major customer, it could move the conversation from “maybe” to “inevitable.”

The outcome matters beyond Volantis. Memory is becoming a key focus in AI hardware, with startups like SiMa.ai and Euclyd tackling the same issue in different ways. The results will affect not just chip costs but also whether large models can run efficiently on smaller devices. For now, Volantis has a working technology and a clear goal. Whether that’s enough to change an industry built around traditional memory remains unclear.

Sources: star-telegram.com