Soma Energy launches to optimize AI data center energy use

The growing energy demand of AI

The rise of AI has driven a significant increase in data center energy consumption, and this trend will only grow as demand for AI workloads increases. This dramatic increase in energy use has placed immense pressure on an already strained grid infrastructure, highlighting the need for tools that can adapt to this new reality.

Caramanolis has spent nearly two decades in electricity trading and energy optimization. His experience managing large-scale energy portfolios and time at AWS gave him a firsthand view of the inefficiencies in data center energy management.

"The scale of energy demand has changed dramatically," Caramanolis said. "In the past, the grid wasn't growing; it was relatively flat. But now we're seeing data center demand go from 3% of the U.S. grid to maybe 15% in the next three or five years."

The Environmental and Energy Study Institute reports that data centers are projected to use 12% of electricity demand by 2030, and that percentage is only expected to grow.

A focus on efficiency over expansion

Soma Energy focuses on optimizing existing grid infrastructure. Caramanolis emphasized the importance of unlocking underutilized grid capacity, which can often be as much as 20 to 40% in parts of the U.S. "And by making better use of that for these large loads, we can untap a lot more electricity," he said.

According to Stephanie Wu, a senior analyst on the energy and smart infrastructure team at Omdia, a division of Informa TechTarget, the U.S. grid maintains a generation reserve margin of about 15 to 26% for emergencies.

"However, at the transmission level, the 'idle' capacity is often far greater because lines must be built to survive the single hottest hour of the year, not the average weekdays," Wu said. "This conservative approach ensures reliability but results in significant underutilization for the remaining 99% of the time."

Optimizing energy use alleviates grid strain and enables the development of more AI and cloud data centers, which are now essential to the technology industry's growth.

Optimizing energy use alleviates grid strain and enables the development of more AI and cloud data centers, which are now essential to the technology industry's growth.

Soma Energy's platform uses AI software to optimize energy delivery, creating a unified control layer that integrates data centers and power producers. Traditionally, these entities have operated separately, but Soma Energy envisions a future where they work together seamlessly.

"In the past, from our experience, they've been sort of separated, and we believe that it could be one control layer that integrates all those resources for a system solution or large energy users like data centers," Caramanolis explained.

Using AI to optimize energy use

While AI is a major driver of increased energy demand, it also offers tools for managing the challenges it creates. Soma Energy's platform uses AI to enhance grid efficiency, enabling data centers and power plants to access electricity more effectively.

According to the vendor, AI is used "to coordinate generation, storage, and large energy loads in real time, optimizing how power is produced and consumed across the system." The platform adapts to changing market conditions and grid constraints, making better use of available capacity and improving system efficiency. This leads to quicker access to power, lower costs and greater flexibility.

This approach not only enhances operational efficiency but also addresses significant challenges in energy distribution.

"AI provides the solution to unlock the 'stranded' 20 to 40% underutilized [grid] capacity," Wu said. "Rather than waiting five to 10 years for new transmission lines, AI software can optimize the existing grid in real time with the spare capacity."

The platform is designed to adapt to a wide range of scenarios, from small data centers to hyperscale facilities. By optimizing battery dispatch and integrating renewable energy sources, such as wind and solar, Soma Energy's software can deliver 10 to 30% cost savings, depending on the size and location of the data center. Cost savings are relative to the standard method of relying only on grid power, according to the vendor.

"There's so many factors that go into the savings and the efficiency opportunities that it's really depending on the situation, the region, location and size," Caramanolis said.

Addressing grid bottlenecks

Soma Energy's AI software untangles grid bottlenecks, which are becoming increasingly common as energy demand grows. The platform also allows customers to use any energy resources they prefer.

The platform's adaptability to multiple energy resources is particularly important as the energy landscape becomes more complex, with a mix of traditional and renewable energy sources. Soma Energy's platform also supports the integration of battery systems, which can provide additional capacity and flexibility for data centers.

"We can unlock existing capacity that exists on the grid today," Caramanolis said. "But we can also incorporate restore systems that enable a lot more additional capacity for data centers and a lot more flexibility that they need."

A sustainable future for AI and energy

Soma Energy's emergence from stealth comes amid rising energy demand and sustainability challenges. The company's ability to work with a variety of energy sources brings hope to data center energy adaptability.

Ultimately, Soma Energy aims to support the AI-driven economy by making the grid more efficient and scalable.

"We've experienced the problems and challenges in the grid for decades," Caramanolis said. "Our solution will make it more efficient to access electricity at scale and support the AI economy."

Kelly Richardson is the site editor for Informa TechTarget's data center site.