Volantis

The Volantis A-1 integrates proven compute IP with proprietary photonics for datacenter-ready inference. 15x tok/$ and 6x tok/W vs GPU baselines on trillion-parameter models.

Optical waveguides extend memory reach 100x beyond electrical limits. Integrated micro-VCSEL arrays deliver 200+ TB/s bandwidth with sub-5ns latency—no external lasers.

Led by Lachy Groom and Abstract, totalling $97M raised to date, to break the memory wall with photonics.

Join a small, high-ownership team solving the memory wall in AI. Open roles in photonics, packaging, and systems engineering. Apply to work on frontier hardware.

Founded by engineers from NVIDIA, Broadcom, AMD, Ayar Labs, and Micron who pioneered HBM and co-packaged optics. Volantis scales memory capacity and bandwidth by an order of magnitude.

Get in touch with Volantis to request early access, reserve inference capacity, or learn about photonic AI infrastructure for large-model serving.

Book a meeting with the Volantis team to discuss early access, capacity reservations, or partnership opportunities for photonic AI inference.

Run giant AI models at low latency. Volantis photonic interconnect delivers SRAM-level bandwidth with rack-scale memory—1,000+ tok/s/user on frontier models.

Run giant AI models at low latency. Volantis photonic interconnect delivers SRAM-level bandwidth with rack-scale memory—1,000+ tok/s/user on frontier models.

Volantis raises $88 million in Series A funding co-led by Lachy Groom and Abstract Ventures. The company is developing a new photonic architecture for AI inference, aiming to build a photonic memory layer for next-generation AI. The funding supports that development effort.

Oct 2026 · Tech Funding NewsNews article

Volantis raises $88 million in a Series A funding round to propel development of its photonic AI inference architecture.

Oct 2026 · IndustrialBriefsNews article

Volantis raises $88 million in a Series A led by angel investor Lachy Groom and Abstract Ventures to develop an inference-optimized chip architecture. Its optical interconnect uses VCSELs, and the company says it provides more than 30 times the memory bandwidth of current accelerators. The A-1 appliance, planned for shipment next year, has 10 terabytes of memory and 250 terabits per second of bandwidth; Volantis estimates it can process up to 10,000 tokens per second on a 20-trillion-parameter model.

Oct 2026 · SiliconAngleNews article

Volantis raises $88 million in Series A funding to develop and commercialize a photonic architecture for AI inference.

Oct 2026 · Pulse 2.0News article