# Throne's founders leave creator gifting for AI's power bottleneck > Source: > Published: 2026-06-29 11:26:34+00:00 Leonhard Soenke and Patrice Becker have moved from creator wishlists to data center power, with their new company [TAR](https://www.tar.com/?ref=runtimewire) saying it has raised a $27 million seed round to build energy [infrastructure for AI](/article/seltz-lands-12-5-million-to-build-search-infrastructure-for-ai-agents) compute. The switch is not a brand extension. It is a founder reset. In an [as-told-to essay published by Business Insider](https://www.businessinsider.com/why-creator-economy-startup-founder-pivot-data-center-power-tar-2026-6?ref=runtimewire), Soenke said TAR, short for Transformative American Resources, recently raised the seed round at a $500 million valuation from an undisclosed strategic investor. Because the investor is unnamed and the valuation is disclosed by Soenke rather than by filings or a named backer, the clean read is narrower: TAR has told the market it has substantial seed capital, but the counterparty underwriting that price remains hidden. That matters because TAR is not selling a software workflow into a familiar SaaS budget. Soenke and Becker are leaving [Throne](https://throne.com/landing?code=ama-96693&ref=runtimewire), the creator gifting company they founded in 2021, for a business that requires land, equipment, permitting, contractors, grid strategy and capital discipline. The founders are making a bet that the next scarce input in AI is not another wrapper, app or model interface. It is power delivered quickly enough to turn GPUs into useful tokens. Soenke framed the move as a search for a harder problem after Throne had become stable. He told Business Insider that he and Becker began discussing a move away from Throne in the summer of 2025, then left New York for San Francisco to spend time with AI labs and major compute providers. Those conversations led them to TAR. The founders have since moved to Austin while keeping a San Francisco office for engineering work, a split that mirrors the company they are trying to build: part software-era speed, part Texas energy and construction market. ### From privacy-first gifting to power infrastructure Throne was a good software-founder training ground in one specific way: it forced Soenke and Becker to build around real-world logistics while presenting creators and fans with a simple interface. The product lets creators on platforms including YouTube, TikTok and Instagram receive gifts from fans without exposing private addresses. Soenke told Business Insider that Throne had grown into a large organization, worked closely with Amazon and had partnerships with major vendors. He also said he and Becker handed Throne to trusted team members over roughly six months, while remaining available when needed. That handoff is the quiet operational detail behind the launch: TAR is not a weekend pivot. It is a deliberate founder transition out of one market and into another. ### TAR is selling deployment speed, not a new electron TAR's pitch is built around behind-the-meter energy systems. In Business Insider's account, Soenke said the systems will be modular and scalable, using a mix of solar, batteries, wind energy and natural gas. He was explicit that TAR is not trying to invent a new way to produce energy. It is trying to deploy existing technologies faster. That distinction is central. The AI infrastructure market is crowded with companies claiming to solve pieces of the compute stack. TAR is choosing a less glamorous but more binding layer: whether a data center can get reliable energy on a timeline that matches demand for training and inference. Becker gave a more concrete description in a [Forbes interview earlier this month](https://www.forbes.com/sites/johnkoetsier/2026/06/15/startup-raises-27-million-to-solve-two-massive-data-center-problems/?ref=runtimewire), saying TAR is building modular, scalable, behind-the-meter systems consisting of solar, batteries, wind and, for emergencies or long periods of unfavorable weather, simple-cycle gas turbines. The company's own site is spare. [TAR's homepage](https://www.tar.com/?ref=runtimewire) says it builds the energy infrastructure America needs and links to the Forbes announcement of its $27 million seed round. Its [careers page](https://www.tar.com/careers?ref=runtimewire) uses the same framing. There is no public list of customers, no disclosed project locations, no named investor, no revenue figure and no signed megawatt pipeline published on TAR's site. That does not make the company unserious. It makes TAR an infrastructure startup at the stage where the difference between ambition and company value will be measured in energized projects, not launch copy. ### Why the market is receptive now The AI industry has spent the past several years treating compute as a procurement and chip-allocation problem. That problem has not gone away, but power has moved from footnote to gating factor. The [International Energy Agency](https://www.iea.org/reports/energy-and-ai/executive-summary?ref=runtimewire) has highlighted rapidly rising data center electricity use and warned that grid bottlenecks and multi-year transmission build times can delay projects. That is the opening TAR is trying to exploit. If grid interconnection queues and transmission timelines are too slow, behind-the-meter power becomes not just a sustainability story but a speed-to-market strategy. If a compute provider can power a data center faster without waiting entirely on grid upgrades, the value is not only cheaper electricity. It is earlier GPU utilization. ### The unanswered questions are the company TAR's open questions are not peripheral. They are the business. The first is whether TAR can actually build faster than utilities, independent power producers and hyperscalers already attacking the same bottleneck. Behind-the-meter power can reduce some grid exposure, but it still faces land, permitting, equipment supply, gas availability, weather variability, interconnection choices and operating risk. Supply chains for turbines and other heavy equipment can also be tight. The second is whether TAR's mix of renewables, batteries and natural gas can meet the reliability profile AI data centers require. Data centers do not merely need energy over a year. They need power at specific moments, with tight uptime expectations and increasingly volatile loads from AI workloads. That makes batteries and dispatchable generation more than accessories. The third is customer proof. Business Insider reports that Soenke and Becker spent time with AI labs and major compute providers before landing on TAR, but neither Business Insider nor TAR's site names a customer. The undisclosed strategic investor could be a meaningful market signal, but without a name it cannot be weighed. Still, the founder logic is clear. At Throne, Soenke and Becker built a product around a trust and logistics problem that creators could not solve with generic tools. At TAR, they are applying the same founder pattern to a much larger and harder market: AI companies need compute, compute needs power, and the power system moves at a different speed than the model cycle. That is why this pivot is more than a curiosity about creator economy founders entering energy. It is a marker of where AI value is moving. The software layer still matters, but the founders chasing the next bottleneck are increasingly being pulled into the physical world, where the work is slower, heavier and harder to fake.