Volantis bags $88M to fix AI’s memory bottleneck

Volantis has raised $88 million in a Series A round led by angel investor Lachy Groom and Abstract Ventures, with participation from John Doerr, former Intel AI chief Naveen Rao, and more than half a dozen other backers. The San Francisco startup aims to address the so-called “memory wall” that limits today’s AI accelerators, proposing an approach that could reduce latency and power consumption in data movement.

The funding round stands out for its size, reflecting investor curiosity about alternative solutions to AI hardware challenges. Volantis’s concept involves leveraging photonics—an area that has gained attention for its potential to improve data transfer efficiency. If successful, such a method might offer advantages over traditional electrical interconnects, though the technology remains unproven at scale.

Photonics has long been used in networking, but adapting it for server-level applications is still an emerging field. Volantis suggests it is building on prior work in the space, though details about its progress or differentiation remain limited. The team includes engineers with experience at major tech companies, which may lend credibility to its development efforts.

The startup faces competition from other players exploring optical or advanced interconnect solutions, each pursuing different technical approaches. Volantis has not yet shared benchmarks or customer deployments, leaving its performance claims untested. The $88 million will likely support further development, with prototypes expected in the coming years.

For those tracking AI hardware, the round highlights ongoing efforts to tackle memory bottlenecks—a persistent challenge in scaling AI systems. Whether Volantis’s approach will deliver meaningful improvements remains to be seen, but the funding underscores investor willingness to explore unconventional solutions in this space.

Sources: siliconangle.com