What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Volantis announced an $88 million Series A on October 1, 2026, to develop and commercialize A-1, an AI inference system designed around photonic links between compute chips and memory. The company says the architecture is intended to address memory limits in large AI models; it plans its first integrated inference-engine deliveries to customers in 2027. A-1’s published capacity and speed figures are design targets, not independently verified results.
What Volantis is building
Volantis describes A-1 as an inference system that uses a photonic fabric to connect compute chips with a larger pool of memory. In AI inference—the process of running a trained model to generate outputs—the system must move model data between compute and memory. Volantis’ stated rationale is that increasing memory capacity alone is not enough if the links cannot move data fast enough; its design aims to increase capacity and bandwidth together.
Founder Tapa Ghosh said the company is “building a system for AI inference that uses photonics to break the memory wall.” That is Volantis’ description of its goal, not independent evidence that A-1 has achieved the claimed performance. The company says its optical fabric uses integrated micro-VCSELs, a type of laser component, to link compute and pooled memory. Volantis’ account of its architecture provides the company’s rationale; its funding announcement sets out the system targets.
What the announced specifications do—and do not—show
Volantis says A-1 is designed for models exceeding 20 trillion parameters and for up to 10,000 tokens per second per user. Both numbers are company-stated design targets. The announcement does not provide an independently verified A-1 system benchmark establishing those figures as achieved operating performance.
#1 Best Overall
SiliconANGLE’s contemporaneous report also presents technical specifications attributed to the company, rather than independent test results. A meaningful assessment will require delivered-system measurements with the model, workload, hardware configuration, and measurement method identified. For the speed-per-user figure in particular, results need to clarify the workload and operating conditions; a target by itself does not show what customers will experience. SiliconANGLE’s report provides secondary coverage, but does not establish an independent matched benchmark.
What the $88 million round is for
The Series A was announced October 1, 2026, and was co-led by Lachy Groom and Abstract Ventures. The company named John Doerr, VXI Capital, Triatomic, Susa Ventures, and angel investors Dwarkesh Patel, Naveen Rao, and Sholto Douglas as additional participants.
Volantis says it will use the financing to develop and commercialize A-1 and its photonic memory architecture, expand engineering, and progress toward customer deployments. The raise therefore funds development and a planned move toward deployment; it is not itself evidence that the system is commercially available or that customers have received it.
When customers may see A-1
Volantis says it plans to deliver its first integrated inference engines to customers in 2027. This is a forward-looking schedule announced by the company, not a report of completed delivery. Whether that schedule is met—and what the delivered systems can do—will be important practical tests of the architecture.
Recommended Free Tools
What to look for when evaluating the claims
Once systems are delivered, the most useful evidence will connect the architecture to measured results rather than repeating targets. Look for:
Quick Recap
- Independent or clearly documented measurements of end-to-end inference performance, including the model, workload, system configuration, and test conditions.
- Evidence that memory capacity and bandwidth scale together in actual deployments, and an explanation of how the optical fabric contributes to the measured outcome.
- Customer deployment details that distinguish a planned delivery from a system in operation.
- Comparable measurements against other inference systems using the same workload and a transparent measurement basis.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
