San Francisco‑based semiconductor startup Volantis announced Thursday that it has closed an $88 million venture round to accelerate its laser‑communication technology for artificial‑intelligence (AI) chips. The capital infusion, led by Stripe veteran Lachy Groom and Abstract Ventures, brings together a roster of seasoned backers including John Doerr, VXI Capital, Triatomic, Susa Ventures, and several prominent angels from the AI community.
Why the funding matters for AI hardware
Current AI processors from Nvidia and Advanced Micro Devices (AMD) are limited by the short reach of the tiny electrical wires that connect the compute core to the memory chips where AI models reside. Those companies mitigate the problem by surrounding each GPU with a handful of high‑bandwidth memory (HBM) packages, but even the most advanced designs can attach only eight memory stacks per GPU.
Volantis proposes a different approach: using vertical‑cavity surface‑emitting lasers (VCSELs) to transmit data between the compute die and memory via beams of light. VCSELs are already familiar to most consumers, powering facial‑recognition features in hundreds of millions of iPhones. By repurposing this proven component, Volantis believes it can sidestep the electrical‑wire bottleneck and potentially attach up to 220 memory chips around a single GPU.
From concept to a prototype chip
CEO and co‑founder Tapa Ghosh told Reuters that the company’s goal is to deliver a working AI chip by next year. “Advanced packaging is always to be respected – it’s never trivial – but it’s not necessarily a new thing to do,” Ghosh said. “No one is going to win a Nobel Prize if our project works, but the good news is, they won’t need to.” The statement underscores the pragmatic mindset driving Volantis: focus on solving a real engineering hurdle rather than chasing accolades.
The new funding will be used to scale up prototype production, secure additional VCSEL supply, and expand the engineering team needed to integrate the laser links into existing GPU architectures. By leveraging components that already exist in the consumer supply chain, Volantis hopes to avoid the severe shortages that have plagued the broader AI‑chip market.
Investor confidence in a critical technology
Among the investors, John Doerr’s participation signals strong confidence from the venture community. Doerr, who backed Google and Amazon in their early days, sees the laser‑link approach as a potential game‑changer for the AI hardware ecosystem. Other backers include AI podcaster Dwarkesh Patel, AI‑chip veteran Naveen Rao, and Anthropic researcher Sholto Douglas, all of whom bring deep domain expertise to the venture.
“The ability to dramatically increase memory bandwidth without the cost and complexity of traditional HBM stacks could unlock a new generation of AI workloads,” said a spokesperson for Abstract Ventures. “Volantis’s solution aligns with the industry’s need for scalable, cost‑effective performance gains.”
Implications for the broader tech landscape
If successful, Volantis’s laser‑based interconnect could reshape how data centers and edge‑computing devices handle AI inference and training. Faster, more efficient memory access would enable larger models to run on existing hardware, potentially lowering the barrier to entry for smaller firms and startups that cannot afford the most expensive GPU configurations.
While the technology is still in development, the substantial capital raise reflects a growing appetite among investors to fund innovative approaches that address the fundamental limits of today’s AI hardware. The coming months will be critical as Volantis moves from laboratory prototypes to silicon that can be evaluated by major OEMs and cloud providers.
Original reporting: Appleton, WI News Feed (HLL/CB) — read the source article.
