Positron AI Secures $51.6M Series A for AI Inference Hardware

Positron AI, a leader in American-made semiconductors and AI inference hardware, recently closed an oversubscribed $51.6 million Series A funding round. Valor Equity Partners, Atreides Management, and DFJ Growth led the investment. This round brings Positron AI's total capital raised this year to over $75 million.

The new capital will support the continued deployment of Positron’s first-generation product, Atlas. Furthermore, it accelerates the rollout of its second-generation products, slated for 2026. Global tech firms anticipate spending over $320 billion on AI infrastructure in 2025.

Positron’s Atlas product delivers 3.5x better performance-per-dollar and up to 66% lower power consumption than Nvidia H100 GPUs. Unlike general-purpose GPUs from companies like AMD, Atlas is purpose-built for accelerating and serving generative AI applications. Moreover, its memory-optimized FPGA-based architecture achieves 93% bandwidth utilization.

Advancing Next-Gen AI Inference

Atlas supports up to half a trillion-parameter models in a single 2-kilowatt server. Additionally, it is fully compatible with Hugging Face transformer models and serves inference requests via an OpenAI API compatible endpoint. Chips for Atlas are fabricated in the U.S. and are already deployed in production environments.

Positron AI was co-founded in 2023 by CTO Thomas Sohmers and Chief Scientist Edward Kmett. Former Lambda COO Mitesh Agrawal joined as CEO to scale commercial operations. The team brought Atlas to market in just 18 months with only $12.5 million in seed funding.

The company's upcoming Titan system, powered by ‘Asimov’ custom silicon, will feature up to two terabytes of high-speed memory per accelerator. This allows for up to 16-trillion-parameter models on a single system. Furthermore, Titan supports parallel hosting of multiple agents or models within data centers.

Early enterprise customers include Parasail and Cloudflare. Positron AI is based in Reno, Nevada.