- TAR, an Austin-based startup that builds modular off-grid power systems for data centers combining solar, batteries, and backup natural gas generators, raised $120 million in a Series A round led by Spark Capital (a key Anthropic investor) at a valuation of approximately $1 billion, with prior investors Buckley Ventures and Align Fund also participating.
- The company’s core differentiation is deployment speed: by using automation and robotics in its manufacturing and installation process, TAR says it can bring hundreds of megawatts of power online in months rather than the years required to secure a new grid interconnection — a critical advantage as data center developers face a multi-year backlog to connect to existing transmission infrastructure.
- TAR’s design is cleaner than most off-grid alternatives: natural gas serves only as emergency backup, with renewables and batteries providing the primary power supply — contrasting with many competing off-grid data center power proposals that use gas as the baseload source and face significant environmental and permitting scrutiny.
- The company is currently building a several-hundred-megawatt project for an undisclosed Texas data center customer, with another large development planned for next year, and will use the new funding to hire engineers and expand manufacturing capacity — positioning TAR at the intersection of two of the hottest infrastructure investment themes: AI compute and energy independence.
What Happened?
TAR announced Thursday that it closed a $120 million Series A funding round led by Spark Capital — the venture firm that is also a major investor in Anthropic — at a post-money valuation of approximately $1 billion. Prior investors Buckley Ventures and Align Fund participated. TAR, based in Austin and co-founded by Pat Becker, builds modular off-grid power systems that combine solar panels, battery storage, and natural gas backup generators to power data centers without any connection to the electric grid. The company uses robotics and automation to compress deployment timelines to months rather than years. TAR is currently building its first large-scale project — several hundred megawatts — for an unnamed Texas data center customer, and has a second major development planned for 2027.
Why It Matters?
The grid interconnection crisis for AI data centers is one of the most underappreciated infrastructure constraints on AI scaling. In major markets, new large load connections to the transmission grid now face queues of 5-7 years — meaning a hyperscaler or co-location developer that breaks ground today cannot reliably power their facility from the grid until 2031 or later. Off-grid power solves the interconnection problem entirely but historically has meant committing to diesel or natural gas baseload, which is expensive, politically fraught, and increasingly subject to emissions scrutiny. TAR’s renewables-primary, gas-backup design addresses the clean energy objection while preserving the speed advantage. The Spark Capital involvement matters because the same LP base funding Anthropic’s model development is now funding the power infrastructure those models will run on — a vertically integrated investment thesis across the AI stack.
What’s Next?
The Texas project is the proof-of-concept: if TAR delivers several hundred megawatts at the speed and cost it has promised, it immediately becomes a template for AI data center power procurement globally. The market opportunity is enormous — hyperscalers and AI companies are collectively seeking tens of gigawatts of new capacity, and any solution that cuts the delivery timeline from 7 years to 6 months will capture premium pricing. The key risks are execution (manufacturing and robotics at scale is hard) and the gas backup component (some jurisdictions will resist any natural gas on-site, regardless of how limited its role). Watch for TAR’s undisclosed Texas customer to be revealed — that name will tell investors whether this is a tier-1 hyperscaler relationship or a smaller co-location deal.
Source: Bloomberg
