Why We Invested in Emerald AI

‍Energize Capital is proud to co-lead the $150M Series A investment in Emerald AI, the software platform transforming AI data centers into flexible assets for the power grid. DCVC joins the round as our co-lead, along with other leading global financial and strategic investors. Energize Managing Partner John Tough will join the Emerald AI board. This marks our first direct investment into the data center space and second in demand flexibility after our investment in Axle Energy.

The Data Center Power Crunch

When Energize was founded in 2017, Data centers were rarely headlining news. Hyperscaler facilities typically topped out around 30 megawatts (~25,000 homes), and when the “world’s biggest” data center was opened in Las Vegas, it had a power capacity of 130 megawatts.  Compare that to last week’s news that OpenAI is planning a new data center in Ohio with 10 GW of power capacity – 333 times larger than those “large” projects ten years ago.

Today’s data center build-out is redefining the scale of computational infrastructure. In July, BNEF projected that data centers will add 118 GW of power demand to the U.S. grid by 2030 - 52% larger compared to their December 2025 forecast and 2.5 times current data center power demand. McKinsey reported that approximately $7 trillion in capital expenditure will be required to meet global projections. This is a structural change to our infrastructure and one that, without the right tools, could cause major challenges for national infrastructure, utilities, and consumers.

The grid is already feeling the strain. The average transmission line is 40 years old, with many lines and transformers already operating past the end of their design life. Congested wires and capacity constraints cause long interconnection delays reaching up to 5-7 years, and grid expansion requires large utility CapEx budgets – a cost that is transferred both to interconnectors and to consumers. These build-outs often reach far beyond what typical usage levels require, and with inherently low risk tolerance, utilities overbuild infrastructure to have buffer room above the highest demand peak, even if this peak only happens about 0.5% of the time. As a result, hyperscalers face major time delays and CapEx costs, while power lines run at 50% of their peak capacity.

Data centers need a better way to coordinate with the grid so that both hyperscalers and utilities can intelligently and efficiently utilize their capacity without risking power outages or service interruptions. This takes improved visibility into energy and computational workload patterns, and it takes orchestrating those compute patterns to offset peak events. When data centers are treated as flexible energy partners, rather than fixed load, those gigawatt-sized projects can be grid assets instead of liabilities.

‍

Introducing Emerald AI

Emerald AI transforms data centers into flexible grid assets that dynamically adjust power consumption in response to power grid conditions. The company does this through its suite of products, including its Emerald Conductor software platform, which intelligently orchestrates computational workloads and onsite energy resources to respond to grid events, reducing power consumption from data centers when the grid is under strain (think A/C units running during a heat wave, or service disruptions from storms).

This model is already in practice in other sectors: today, when utilities anticipate a grid event, they might call upon local industrial companies to curb energy usage during certain times of the day to protect the grid; our portfolio company Axle Energy does this on the residential scale by aggregating and modulating distributed energy resources like EVs and heat pumps to help reduce strain during peak events. Emerald AI applies this logic to the data center segment, stepping in to orchestrate the relationship between utilities and data center operators to intelligently curb compute during peak events without sacrificing workload performance. This is a win-win for everyone: utilities earn more revenue by serving more load with lower infrastructure requirements, hyperscalers can build larger projects without over-stressing the grid, and consumers are protected from having to pay the price for costly grid upgrades. Existing data center customers also benefit; in exchange for making their compute more flexible, they can secure preferable terms and faster access to more capacity. As Emerald Conductor flexes power draw without compromising the performance of mission-critical AI workloads, the effect is often invisible to the end user.

Emerald AI has already completed five demonstrations at data centers across the globe in partnership with NVIDIA, EPRI, Oracle, Nebius, National Grid, and other regional utilities and grid operators, and has now moved into commercialization. This year, Emerald AI and NVIDIA launched the first commercial, multi-megawatt deployment of NVIDIA DSX Flex, with Emerald Conductor integration to dynamically manage data center power consumption in response to grid conditions. The company has been named a 2026 Technology Pioneer by the World Economic Forum and was included on the TIME100 Most Influential Companies list.

The Energize Angle

As AI solutions continue to scale, we see the data center build-out as a durable tailwind for energy innovation, and we are bullish on solutions that serve both the “stock” and “flow” of these infrastructure assets. To meet demand, hyperscalers and grid operators alike will lean on a robust tech stack that can make power capacity go further, faster. Energy flexibility will be table stakes for data center deployment, and software solutions across the value chain will be essential to making that happen safely. As with every technology cycle before it, the AI build-out will hinge on efficiency; this pressure is already coming from consumers, hyperscalers, infrastructure, and the market. Emerald AI is positioned to be a core enabler of that efficiency and flexibility.

The Team

Emerald AI was founded by CEO Dr. Varun Sivaram in 2024 and has grown to over 35 employees across the U.S. Sivaram brings deep policy expertise as well as his experience as Chief Strategy Officer at Orsted to tackle the complex landscape of policy, energy, and innovation stakeholders.

This capital infusion allows Emerald AI to grow its commercial deployments, expand its engineering, research, and commercial teams, and advance the utility programs and industry standards that establish verified flexible load as a dependable resource for grid operators. You can explore job opportunities here: https://www.emeraldai.co/careers

‍