Chip startup Positron has raised $23.5 million to support the production of its AI inferencing chips.
The funding round saw participation from Flume Ventures, Valor Equity Partners, Atreides Management, and Resilience Reserve.
Positron describes its Atlas chips as a “cost-effective alternative to Nvidia’s AI hardware,” claiming that its silicon can currently achieve 3.5x better performance per dollar and 3.5x greater power efficiency than Nvidia H100 GPUs for inference.
Its FPGA-powered servers are also compatible with Hugging Face and OpenAI and support trillion-parameter models, with Positron saying the systems are capable of 70 percent faster inference and 66 percent lower power consumption than Nvidia H100/H200 setups, which could cut data center capex by 50 percent.
The company is based in Reno, Nevada, and designs, fabricates, and assembles its chips in the US, with the manufacturing taking place in Arizona - likely TSMC's new fab. In January 2025, Positron appointed former Lambda Labs COO Mitesh Agrawal as chief executive.
“With this funding, we’re scaling at a pace that AI hardware has never seen before–from expanding shipments of our first-generation products to bringing our second-generation accelerators to market in 2026,” said Agrawal. “Our solution is growing rapidly because it outperforms conventional GPUs in both cost and energy efficiency, while delivering AI hardware that eliminates reliance on foreign supply chains.”
