Working out of Clarendon, Emerald AI announces launch, with $24.5M in seed funding
SUMMARY:
Emerald AI launches with $24.5M in seed funding to make AI data centers more energy-efficient and reduce strain on electric grids
The startup‘s software allows data centers to reduce energy consumption during peak grid demand without sacrificing AI computing performance
Backed by major investors like Nvidia and Google’s Jeff Dean, Emerald AI is ramping up efforts to test and demonstrate what its technology can do nationwide
As data center growth continues to explode across Virginia and the nation, so do fears that the centers are gobbling up too much energy and straining already overtaxed electric grids.
However, backed by Nvidia and other heavy hitters, a new AI startup with offices in Arlington County emerged from stealth Tuesday, stating that it aims to alleviate some of these concerns through software that will help data centers adjust energy consumption, making them more efficient in their energy usage during periods of peak power demand.
Emerald AI, which was incorporated last year in Washington, D.C., but has its primary offices in Clarendon, revealed Tuesday that it has secured $24.5 million in seed funding. The round was led by Canadian venture capital fund Radical Ventures, with participation from NVentures (Nvidia’s venture capital arm), AMPLO, CRV and Neotribe. The company’s individual backers include former Secretary of State John Kerry; Google Chief Scientist and Google AI head Jeff Dean; and Kleiner Perkins Chairman John Doerr.
Emerald AI’s founder and CEO is Varun Sivaram, formerly chief strategy and innovation officer for global offshore wind energy company Ørsted and also former chief technology officer for ReNew, the largest renewable energy company in India. Early in the Biden administration, Sivaram worked for Kerry, who was then special presidential climate envoy, as managing director and senior adviser for clean energy and innovation. Sivaram’s CV also includes a three-year stint as a consultant at McKinsey & Co. and a year as a senior research scholar at Columbia University’s SIPA Center for Global Energy Policy.
“We’re at a critical inflection point as exponential growth of AI computing pressures our electrical infrastructure,” Sivaram said in a statement. “To unshackle AI technology progress from power constraints, Emerald AI transforms data centers from grid liabilities into flexible assets, enabling grid operators to swiftly interconnect AI, bolster reliability and energy security, and more efficiently harness the massive spare capacity on today’s grids.”
A Joint Legislative Audit and Review Commission report on data centers from last year indicates that the unconstrained demand for power in Virginia is expected to double within the next 10 years, with the data center industry being the primary driver. The North American Electric Reliability Corp. (NERC) has warned that increased demand poses challenges to grid stability.
Furthermore, Emerald AI notes that substantial investments in building electricity infrastructure over the coming decades will likely lead to higher energy prices for American communities.
The startup’s solution is an AI conductor software platform that enables data centers to adjust and reduce their energy consumption during periods of peak grid demand to support grid stability while ensuring acceptable AI computing performance.
This method of managing energy consumption will prevent issues that now occur when data centers consume electricity that is needed elsewhere. The company says by allowing data centers to require less power to operate, its platform will help data centers bypass yearslong waits for grid interconnection.
Sivaram first conceived the idea for Emerald AI’s technology a few years ago while working for Ørsted.
Although the company was registered last year, Emerald AI’s first hire took place in January. The company now has nine full-time employees, including academics, researchers and Big Tech software engineers.
On Tuesday, Emerald AI released results of a demonstration that took place in May as part of the Electric Power Research Institute’s DCFlex Initiative in Phoenix. Nvdia reported that Emerald AI’s conductor platform demonstrated that it can reduce power consumption of AI workloads by 25% for 3 hours in response to a grid stress event, such as a hot summer day when people are using more energy, while maintaining computing service quality.
Emerald AI plans to conduct larger-scale demonstrations in Phoenix and around the country in the future.
“We have some projects planning in Virginia that we are over-the-moon excited about,” said Jack Megrue, Emerald AI’s chief of staff and business operations associate. However, he wasn’t able to offer more specifics.
In an emailed statement, Varun said Emerald AI is actively ramping up its activities in Pennsylvania, including major investments in data center flexibility. He said this would enable Pennsylvania to turbocharge its economic development and AI data center deployment while limiting cost increases for local communities and improving the reliability of the power grid for energy and national security.
The company has not yet announced any partnerships or collaborations with utilities like Dominion Energy.
Former Secretary of Commerce and Emerald AI Advisor Gina Raimondo stated that Emerald AI’s approach will accelerate AI deployment, enhance grid reliability and protect consumers from rising energy costs.
“Enabling faster innovation without putting additional strain on our existing power systems is critical for maintaining America’s competitive edge in AI development,” she said in a statement.
T