Positron AI has secured an oversubscribed $51.6 million Series A funding round, pushing its total capital raised this year to over $75 million. The round was led by Valor Equity Partners, Atreides Management, LP, and DFJ Growth, and will be used to speed up the deployment of Atlas, Positron’s advanced AI inference system designed to outperform NVIDIA’s H100 in cost, power efficiency, and performance.
Atlas, Positron’s flagship offering, is built on a memory-optimized FPGA architecture that achieves 93% bandwidth utilization, enabling it to deliver 3.5x better performance per dollar and 66% lower power consumption than the NVIDIA H100. “By generating 3x more tokens per watt than existing GPUs, Positron multiplies the revenue potential of data centres,” said Randy Glein, co-founder and managing partner at DFJ Growth.
The funding marks a significant milestone for Positron as it prepares to scale Atlas across various high-performance AI use cases such as LLM hosting, AI copilots, and generative agents. According to CEO Mitesh Agrawal, the company is focused on enabling large-scale AI deployments with extreme efficiency: “Our highly optimised silicon and memory architecture allows for superintelligence to be run in a single system,” he said. Positron systems are being designed to handle 16-trillion-parameter models and support long-context inference, making them well-suited for the future of advanced generative AI.
Looking ahead, Positron is already working on its next-generation inference platform, dubbed Titan. Titan will be powered by the company’s custom-built ‘Asimov’ chip and will feature up to two terabytes of high-speed memory per accelerator, setting new benchmarks for AI infrastructure performance.
With Atlas gaining early adoption and Titan on the horizon, Positron AI is positioning itself as a serious contender in the AI hardware race—offering a compelling alternative to GPU-based systems with a focus on energy efficiency, memory bandwidth, and scalability for next-gen AI workloads.
