Volantis raises $88 million in Series A funding to develop photonic inference systems
October 6, 2026 · 2 min read
Volantis has raised $88 million in a Series A funding round led by Lachy Groom and Abstract Ventures, with participation from more than six other investors including John Doerr of Kleiner Perkins and Naveen Rao, former head of Intel's artificial intelligence product group.
The funding will support the development of photonic inference systems built on a specialized memory interconnect architecture. The company claims its technology delivers over 30 times the memory bandwidth of current accelerators, addressing a key performance bottleneck for AI models.
Volantis' chip design features an interconnect span exceeding 200 millimeters, allowing more than 220 AI accelerator chips to be placed alongside memory chiplets. This expanded reach overcomes limitations imposed by traditional wiring, which restricts memory modules to within 5 millimeters of the GPU core.
The photonic interconnect uses vertical-cavity surface-emitting lasers (VCSELs) to transmit data as light. These devices convert electrical signals into light using quantum wells and mirrors, and can be manufactured from gallium arsenide, a material more readily available than those used in conventional lasers.
Volantis plans to ship its A-1 inference appliance next year. The system occupies roughly one-third the size of a standard server rack, includes 10 terabytes of memory, and provides 250 terabits per second of memory bandwidth. The company estimates the system can process up to 10,000 tokens per second when running a 20-trillion-parameter model.
Tapa Ghosh, co-founder and CEO of Volantis, wrote in a blog post that the technology enables real-time frontier inference, restores scaling laws, and allows entire codebases to be managed within a single context window. He added that an AI coding agent could complete tasks in 30 seconds instead of 30 minutes.