Founder Moves From Creator Economy to Data Center Power

Business Insider reports that Leonhard Soenke, who cofounded creator-economy startup Throne in 2021 with Patrice "Pat" Becker and Heiner Stinner, has cofounded a new company, TAR (Transformative American Resources), with Becker rejoining him as a cofounder. TAR raised a $27 million seed round at a $500 million valuation from an undisclosed strategic investor, according to Business Insider; the round was first reported by Forbes and confirmed by DataCenterDynamics. TAR builds modular, behind-the-meter power systems combining solar, wind, and battery storage with simple-cycle gas turbines for backup, "pre-wired, pre-assembled, pre-tested, and pre-commissioned" at a factory before shipping to a site, Becker told Forbes. The company says its systems can be operational in about three months, versus the five-to-seven-year grid interconnection queues Forbes and DataCenterDynamics report are now common for large power-hungry facilities. Soenke splits his time between San Francisco and Texas; in the Business Insider piece he said, "Our attitude is if ain't broke, don't fix it."

TAR's pilot project targets roughly 10 MW of continuous power at an undisclosed site, with its first customer deployment for an undisclosed "neocloud" offtaker planned at roughly double that capacity, per DataCenterDynamics. The company says it has a pipeline exceeding 200 MW for 2027 and multiple gigawatts by 2028. Becker told Forbes the approach is not necessarily cheaper than grid power on a per-megawatt-hour basis, but is priced below typical behind-the-meter power purchase agreements (which he put at $150-160/MWh), while avoiding interconnection queues, curtailment, and lengthy permitting. TAR's business model resembles Intersect Power, the co-located renewable-and-storage data center developer Google acquired for $475 million in late 2025, per DataCenterDynamics; other recent comparables include Aligned Data Centers' 31 MW battery project with Calibrant Energy and Redwood Materials' four-month solar-and-EV-battery microgrid deployment.

The reported financing and technical claims underscore that site-level power delivery, not just chip availability or algorithmic efficiency, increasingly gates how fast new AI training and inference capacity can come online. Teams evaluating colocation or private data center builds should treat interconnection-queue timelines as a first-order planning constraint alongside GPU procurement, and weigh emerging behind-the-meter providers against traditional utility-connected sites on speed-to-power rather than power cost alone.

Whether TAR discloses its first customer deployment and independently verifiable performance data (uptime, actual delivered megawatts versus nameplate capacity), how its natural-gas backup ratio evolves as renewable buildout scales, and whether other founders make similar pivots from consumer or creator-economy products into physical AI infrastructure.