AI inference technology startup Positron has raised US$23.5 million in funding from investors groups such as Flume Ventures, Valor Equity Partners, Atreides Management and Resilience Reserve. The company aims to expand production of US-made AI chips, offering a cost-effective, energy-efficient alternative to NVIDIA’s AI hardware.
Also Read: Fluidstack Builds AI Supercomputers in France with 10 billion Euro investment
ATLAS System: High-Performance AI Calculation
The company said it has already shipped products to data centers and neoclats across the US, calling it a feat, “with over hundreds of millions of people backing, a chip startup cannot overcome.”
Positron claims that ATLAS systems offer 3.5x more performance and 3.5x more power efficiency than NVIDIA’s H100 GPUs for AI inference. Its memory-optimized architecture achieves over 93% bandwidth utilization, significantly outperforming GPUs. These systems achieve 70% faster inference with 66% lower power consumption than the H100/H200 setup, reducing data center capital expenditures by 50%.
“Positron’s FPGA-powered server supports the Trillion parameter model while providing plug-and-play compatibility that embraces the face and Openai API,” the official release states.
Made in America ai chip
Beyond specifications, the company emphasized that the first generation Atlas systems are fully designed, manufactured and assembled in the United States.
“This funding has led to AI hardware expanding at a pace we’ve never seen before. With the expansion of first-generation products shipping, we’ll bring second-generation accelerators to the market in 2026,” Positron said. said Mitesh Agrawal, CEO of the company. “Our solutions are growing rapidly to provide AI hardware that outperforms traditional GPUs in both cost and energy efficiency while also eliminating dependency on foreign supply chains.”
Also Read: Government plans to revive top Indian and Origin AI experts for local innovation: Report
Domestic alternatives to global AI hardware
Unlike many semiconductor companies that rely on offshore manufacturing, Positron has built a completely American supply chain by ensuring that AI hardware is designed, manufactured and assembled within the US. I’ve said that.
“Investing in domestic AI hardware is a strategic obligation when it comes to ensuring America’s global AI stance,” said Scott McNealy, operating partner at Flume Ventures. “Positron has proven that world-class AI computing doesn’t need to come from overseas, and is excited to support their mission to make the United States a leader in AI hardware manufacturing.”
Beyond cost savings, Positron said its technology tackles the growing power constraints of AI infrastructure. “Many legacy data centers struggle to support high-power GPUs that consume 10,000 watts per server, which exceeds the capacity of traditional data centers. Positron’s energy-efficient architecture This allows these facilities to participate in AI computing without the need for large infrastructure upgrades. “AI chip startups said.
Also Read: LG and TenStorrent Strengthen Partnership to Strengthen AI Chip Functions
“The demand for AI calculations is surged, and companies are looking for viable alternatives that are energy and cost-effective in the long term,” said Rob Reid, co-founder of Resilience Reserve. “What sets Positron apart is its ability to sell AI hardware at unprecedented speeds and deliver high performance per watt. An innovative approach is cost-cost and cost-specific It’s about allowing AI workloads to be expanded without barriers. Power consumption.”
Subscribe
