Volantis raises $88M for photonic inference

Volantis raised an $88 million Series A to build A-1, a photonic inference system designed for models exceeding 20 trillion parameters at up to 10,000 tokens per second per user. In the financing announcement, the company says its chip-to-memory fabric can increase capacity and bandwidth together by nearly two orders of magnitude, with first integrated inference engines planned for 2027; those are pre-commercial targets, not shipped benchmarks. Hardware teams should file the round as a bet on memory bandwidth as the next inference bottleneck, but wait for measured throughput, power and utilization data before redesigning a serving stack; the archive’s inference goodput analysis provides the measurement standard.