Volantis, an optical interconnect chip company, said on X that it has completed an $88 million Series A financing round.
The company said it is using an optical approach to tackle AI memory bottlenecks, adding large amounts of high-speed memory to chips and raising per-chip memory bandwidth and capacity by several orders of magnitude. Volantis said the system is intended to support high-speed inference for models with more than 10 trillion parameters, with speeds of up to 10,000 tokens per second for a single user.
Initial use case targets AI agents
Volantis said the technology will first be used to accelerate agents, reducing programming-related tasks from hours to minutes or seconds.
The company also said an optical architecture can improve memory systems by several orders of magnitude from one generation to the next, rather than by about 2x per year.
Team background and product progress
According to Volantis, members of its team previously worked on the first CoWoS product, early HBM, the first silicon photonics CPO system, the first tunable VCSELs shipped at large production scale, and some of the first processors to communicate directly with light.
The company said it has already achieved transmission distances inside chip packaging that are more than 10x those of equivalent fine electrical wiring. Its next-generation product has taped out, and it is aiming to set a record for bandwidth density at those distances.