Soma Energy, a US-based energy software startup focused on optimizing electricity use for data centers, has raised $7 million in seed funding, led by Category Ventures.

The company intends to use the proceeds of the funding to support the development of its AI-based platform.

Founded by several former Amazon Web Services (AWS) energy and machine learning specialists, including CEO Ath Caramanolis, CTO Mario Souto, and chief AI scientist Henrique Hoeltgebaum, Soma claims that its platform is designed to manage electricity supply and demand in real time, helping data centers secure power more quickly without waiting for new grid infrastructure.

“We saw and solved these problems at AWS 10 years ago, the grid constraints, the interconnection delays, the complexity of managing power at scale,” said Caramanolis. “We built the company because we knew there was a better way. The answer is not simply more infrastructure, but better orchestration. By applying AI to power plants and large energy loads in real time, we unlock flexibility that already exists and accelerate time to power today.”

Soma Energy’s platform works by connecting on-site generation, storage, and electricity demand into a single control layer, which it claims enables data centers to operate more flexibly and access additional capacity from existing infrastructure. For power producers, it provides real-time guidance on when to generate, store, or trade electricity across renewable and storage assets.

“Soma Energy is changing what’s possible for data center growth,” said Josh Simms, CEO, H5 Data Centers. “By coordinating existing resources, we were able to access capacity significantly sooner than expected, accelerating our time to power and removing a critical constraint on expansion.”

The company said it is currently managing around 2GW of electricity for power producers and working with five data center customers. It plans to use the new funding to expand its engineering and commercial teams and scale deployments across North America.

Sona is the latest firm seeking to expedite the connection of data centers to the grid using AI-powered tools. Recent examples include GridCARE, which emerged from stealth last year, aiming to leverage generative AI-based analysis to detect pockets with geographic and temporal capacity on the existing grid, which it said could reduce the time-to-power of data centers to between six and 12 months.