Positron AI, a leading developer of semiconductor solutions for generative AI inference, has secured a whopping $51.6 million in funding from top investors like Valor Equity Partners and DFJ Growth. This latest injection brings the company’s total capital raised this year to over $75 million.

The influx of cash will be used to fuel the continued deployment of Positron’s first-generation product, Atlas, as well as accelerate the development of its next-gen products – including a custom silicon system set for release in 2026. The company’s innovative hardware solutions aim to address the growing pains of AI infrastructure, which is expected to reach $320 billion by 2025.

Currently, companies are facing escalating costs, power limitations, and a shortage of NVIDIA GPUs as they struggle to meet the demand for advanced AI solutions. Positron’s custom-built systems, on the other hand, offer specialized performance optimized specifically for generative AI – rather than relying on general-purpose GPUs.

What Sets Positron Apart

Randy Glein, co-founder and managing partner at DFJ Growth, explains that Positron’s approach is focused on improving the cost and energy efficiency of AI inference. By generating three times more tokens per watt than existing GPUs, Positron effectively multiplies the revenue potential of data centers.

“The early benefits of AI are coming at a very high cost,” Glein says. “Improving the cost and energy efficiency of AI inference is where the greatest market opportunity lies, and this is where Positron is focused.”

Scaling Up with New Funding

With its latest funding injection, Positron plans to accelerate product development and scale its operations. The company will focus on expanding its hardware lineup, including second-generation systems for even more powerful and efficient AI inference processing.

Funds will also be allocated towards hiring key personnel, further solidifying the company’s growth trajectory. According to Mitesh Agrawal, CEO of Positron AI, the company is committed to meeting the demands of modern AI – aiming to run frontier models at the lowest cost per token generation with the highest memory capacity available.

“Our highly optimized silicon and memory architecture allows for superintelligence to be run in a single system,” Agrawal explains. “We’re targeting running up to 16-trillion-parameter models per system on models with tens of millions of tokens of context length or memory-intensive video generation models.”

Inspired by

• https://justainews.com/companies/funding-news/positron-ai-raises-51-6m-to-advance-inference-hardware-for-ai-applications/