World’s First Off-grid, Hydrogen-powered AI Data Centre Unveiled
Data center start up, ECL has announced the roll-out of world’s first 1 GW hydrogen-based AI Data center named TerraSite-TX1 planned on an area of more than 600 acres near Houston in Texas.
Funding
The funding model for TerraSite-TX1 involves a combination of ECL’s capital and partnerships with financial investors. The first phase, which will provide 50 MW of capacity, is expected to cost $450 million, with the entire 1 GW project projected to cost around $8 billion. Portion of initial module will be off taken by Lambda, GPU cloud company, which has raised several hundred million dollars in venture funding as the first tenant. The data center will be scalable to as high as 2 GW based on market demand.
Hydrogen Supply
The Houston area is rapidly advancing in hydrogen production, with several significant projects underway funded with around $1.2 billion to produce low-carbon hydrogen using natural gas and renewable sources. Presently, there are about 1,600 miles of hydrogen pipelines in the U.S with the hydrogen production mostly tied to industries like refineries and heavy metals. ECL has tied up for supply of hydrogen for data center from three pipelines which merge at the project site.
Advantage
The off-grid data center is antecedent to the next generation data centers which were not even expected to evolve this early. Despite an early breakthrough, the data center is not only promising a vast suite of AI based applications but is also 100% sustainable with zero carbon emissions meeting the high energy demands of AI.
With this, the data center will be ten times more energy efficient with Power Usage Effectiveness (PUE) of 1.05 in contrast to the 1.2 to 1.4 range exhibited by conventional data centers due to operation and cooling inefficiency, and power losses in non-IT equipment. For cooling purpose, ECL data center will use water produced during hydrogen power generation reducing the environmental hazard. proprietary rear door heat exchanger technology for high-density rack cooling, while cutting off the reliance on local resources.
ECL has already experimented with hydrogen as the power source through its previous model named ECL-MV1 in Mountain View, California rolled out in June this year. The data center is known for zero emissions, minimal noise, and a negative water footprint, replenishing water to the community. It offered very high energy efficiency with a power usage effectiveness (PUE) of 1.05 and a 7x improvement in data density per rack, which is ideal for AI high-density demands. with PUE of 1.1 and high-density deployments of up to 75kw per rack. ECL’s cooling systems can currently support 75kW per rack and will reach 125kW per rack in the near future, addressing AI-driven demands.
ECL data centers are modular, allowing for easy expansion in 1MW increments, and are built to suit and delivered in less than 12 months, vs. the industry standard of 36 to 48. By eliminating the environmental impact associated with traditional data center development and utilization, they also give AI consumers assurance that their use of AI is not expanding their carbon footprints.
ECL data centers are optimized to support the massive amounts of computational power required by AI workloads, making them ideally suited for companies seeking to deliver AI applications or to leverage AI to accelerate their own innovation and operations, ECL cooling innovations, including utilization of water created from hydrogen-based power generation and
- ECL’s data centers produce zero emissions and are water-positive, often contributing excess water back to the community.
ECL’s AI infrastructure is enhanced by its advanced AI data center management system, ECL Lightning™, which offers real-time monitoring and micro adjustment control over every aspect of data center operations, from power generation to rack cooling. Its user-friendly interface provides comprehensive visibility and configuration capabilities, ensuring optimal performance for AI and other workloads. ECL’s proprietary cooling innovations include using water from hydrogen-based power generation, along with patent-pending quadruple loop and direct-to-chip cooling technologies for its extremely high-density racks.
- ECL’s cost to build is significantly lower than the average data center in the U.S. and worldwide, making it competitive.
Conclusion
It is Envisioned that the boom of such cutting-edge technologies shall revolutionize the data center economy where the growing need for AI based processing is expected to consume multitudes of power and exhaust the available energy sources in the coming future.
It is believed that ECL has made breakthrough in powerful and eco-conscious foundation for AI advancement by promising essential computational resources while drastically reducing the environmental impact of AI operations.
