EPIC Microsystems Raises $21 Million Series A to Power AI Data Centers

EPIC Microsystems, a San Jose semiconductor startup building power-delivery systems for AI infrastructure, pulled in an oversubscribed $21 million Series A to commercialize its vertical power delivery architecture for next-generation AI compute platforms.

Seligman Ventures led the round, with Intel Capital, AICONIC Ventures and Cambium Capital also participating alongside existing seed investors A&E Investments, Assam Ventures and Nepenthe Capital. The financing brings EPIC's total funding to $26 million.

As part of the investment, Umesh Padval, Managing Partner at Seligman Ventures, will join EPIC Microsystems' Board of Directors. Padval framed the investment around a structural problem facing every major operator building out AI capacity. "Power density, efficiency and thermal management have become primary bottlenecks in scaling AI data center infrastructure," he said. "We believe their vertical power delivery architecture is well positioned to address the next generation of AI system demands."

The technical bet at the center of the raise is EPIC's decision to replace traditional inductor-style power systems. By switching out inductor-style systems for a hybrid switched-capacitor architecture, AI data centers can deliver higher density with improved thermal performance. The company's new power-delivery component design takes up less vertical space, making it easier to pack more compute and cooling into a dense rack, and lower-profile power stages leave more room for heatsinks, cold plates, manifolds, tubing and service clearances.

The company's technology is intended to give hyperscalers greater flexibility as AI infrastructure scales toward multi-megawatt environments, where power delivery is becoming as critical as compute performance itself.

EPIC Microsystems was founded by a team of semiconductor industry veterans with deep power design experience. Prior to launching the company, the team developed switched-capacitor solutions used in mobile devices, enabling fast charging across hundreds of millions of smartphones and laptops. That mobile-era work now forms the technical foundation for what the company is deploying at rack scale inside AI data centers.

Nvidia made a similar argument at the rack level, noting that moving certain power conversion elements out of the rack frees space for more compute and improves cooling efficiency, a signal that vertical power delivery is becoming a competitive differentiator rather than a niche engineering preference. With $21 million now secured and Intel Capital among its backers, EPIC enters commercialization with both the capital and the industry credibility to press that case directly to hyperscalers.