Business wire
AI Hardware Industry Veteran Mitesh Agrawal Joins Positron as CEO—A Startup Aiming to Challenge Nvidia’s Grip on AI Infrastructure
For media inquiries or to schedule interviews with Mitesh Agrawal or Thomas Sohmers, please contact:
Helen Cho
Bonfire Partners
helen@bonfirepartners.io
Positron, a rising force in AI inference technology for artificial intelligence computing, today announced that Mitesh Agrawal, former Chief Operating Officer of Lambda, has joined the company as its new Chief Executive Officer. Agrawal’s decision to leave Lambda—an AI cloud computing company deeply integrated with Nvidia’s ecosystem—for Positron signals the growing opportunity in the silicon space to provide high-performance chips that outperform Nvidia and put pressure on their market dominance.
Agrawal brings an extraordinary track record to Positron. During his time at Lambda, he helped scale the company from $500,000 to nearly $500 million in annualized revenue run rate and played a pivotal role in securing their recent funding rounds totaling hundreds of millions of dollars. Over his career, Agrawal has helped raise more than $1 billion in capital, cementing his reputation as a leading figure in AI infrastructure. His intimate understanding of Nvidia’s product offerings, gained from years of collaboration, makes his transition to Positron even more compelling.
"Joining Positron is an opportunity to disrupt an industry that’s exciting and looking for alternatives. The AI hardware market needs to continue to evolve beyond the existing offerings in the market and I’ve seen firsthand that Positron offers a much-needed alternative with superior performance, better energy efficiency, and freedom from vendor lock-in,” said Agrawal. “Positron’s Atlas systems that is available to purchase in the market today provides 3.5x performance/dollar improvement for transformers’ inference over Nvidia’s H100 systems and a 3x performance/watt already with a roadmap to increase this further and the system just works out of the box using HuggingFace transformer libraries with ZERO code changes. I’m excited to work with Positron’s incredible engineering at this critical moment and help increase the momentum that the team (with experience of taping out more than 100 chips) has created."
Positron’s systems have positioned the company as a disruptive force in the AI hardware space. Delivering three times the performance of Nvidia GPUs while consuming just one-third of the power, Positron’s first-generation Atlas systems are designed, fabricated, and assembled entirely in the United States. These attributes provide a strategic edge for data centers and enterprises looking to reduce their dependency on Nvidia and address rising concerns about cost, energy consumption, and geopolitical stability.
Thomas Sohmers, founder and former CEO of Positron, underscored the significance of Agrawal’s leadership. "Mitesh’s expertise and leadership come at a pivotal time for Positron and the AI hardware industry," said Sohmers. "Mitesh understands what it takes to build and rapidly grow one of the largest AI Neoclouds and the demands of the leading AI customers better than almost anyone. His decision to join Positron highlights his confidence in our ability to reshape this market. With his track record of scaling companies and driving innovation, Mitesh will be instrumental in helping Positron meet the growing demand for alternatives to Nvidia."
Agrawal’s appointment reflects broader trends in the AI hardware market where data centers and enterprises are urgently seeking alternatives to Nvidia, especially as the inference applications are on an exponential growth curve— and on the heels of DeepSeek’s disruptive launch of their low-cost R1 model.
“DeepSeek’s success has demonstrated that inference-time compute is the new scaling factor delivering real results for AI models, rather than simply training larger and larger systems. As user adoption skyrockets, the demand for efficient inference will only grow—not just to serve end users, but also as a key component in building, refining, and expanding the next generation of AI,” said Mitesh. “This is where performance per dollar and performance per watt become absolutely critical. At Positron, we’ve always believed that making inference more cost-effective and energy-efficient is fundamental to making the best AI models accessible to everyone. By focusing on specialized hardware that optimizes these metrics, we’re able to power advanced AI while drastically reducing both infrastructure costs and environmental impact.”
Positron’s cutting-edge technology offers these organizations a scalable, efficient, and secure solution to meet the growing demands of AI workloads.
"The AI hardware industry is at an inflection point," Agrawal added. "At Positron, we’re focused on delivering highly efficient, simple-to-use, and easily accessible chips that put customers back in control of their infrastructure and help carry forward the progress made in AI at a faster pace and cheaper costs."
This announcement coincides with Positron’s growing momentum, including major deals with leading cloud computing and inference providers and increasing interest from enterprises eager to diversify their hardware infrastructure.
About Positron
Positron is redefining AI hardware with its advanced inference systems that deliver unmatched performance, energy efficiency, and scalability. Designed and manufactured entirely in the United States, Positron empowers data centers and enterprises to unlock the full potential of their AI workloads while reducing dependency on Nvidia and achieving greater operational freedom.
View source version on businesswire.com: https://www.businesswire.com/news/home/20250203502454/en/
