Yuval Bachar, a data center expert with experience at Meta, Microsoft, and Cisco, is now tackling the challenge of reducing carbon emissions in Silicon Valley’s data infrastructure through his startup, ECL. The company focuses on building hydrogen-powered data centers, offering an eco-friendly energy solution that accelerates deployment compared to traditional grid-connected facilities.
ECL has already launched a 1-megawatt hydrogen-powered data center next to its headquarters in Mountain View, California. Currently, hydrogen is delivered by diesel trucks from Southern California or northern Nevada, primarily derived from natural gas, the dominant source of electricity in the U.S.
The surge in AI technology, driven by platforms like OpenAI’s ChatGPT, has led tech giants such as Amazon, Google, Microsoft, and others to rapidly expand their data center capacity. These centers often rely on energy-intensive Nvidia GPUs to train and run advanced AI models. The growing demand for AI-ready infrastructure has made securing adequate power a top priority, and Bachar believes traditional utilities, particularly in regions like California and Virginia, cannot meet these increasing needs.
Founded in 2021, ECL has secured two paying customers, with several additional orders underway. Bachar highlights the urgent need for innovative infrastructure solutions, as conventional U.S. data centers are ill-equipped to handle the demands of AI.
While nuclear energy remains a promising carbon-free solution, Bachar acknowledges the time required to bring nuclear facilities online. Major tech companies, including Microsoft, Amazon, and Google, are investing in nuclear initiatives, but immediate solutions are needed to meet current demands.
ECL aims to construct a large-scale, 1-gigawatt hydrogen-powered data center in Texas within four years. This facility will eventually rely on green hydrogen produced via electrolysis, although high production costs remain a challenge.
The demand for AI data centers is expected to surge, with U.S. energy consumption projections reaching between 74 gigawatts and 132 gigawatts by 2028—potentially accounting for up to 12% of national energy use. Bachar stresses the urgency of scaling infrastructure quickly to meet the unprecedented energy demands of the AI era.