AI needs a different kind of power plant [partner content]

Episode Summary

Executive Summary: Brad Hartwig, Arbor Energy’s CEO, explains how his rocket-engineering background led him to build modular, low-emission power turbines for an AI-driven market facing severe turbine shortages. He argues that 3D-printed, supercritical CO2 turbines can deliver faster, cheaper, and cleaner firm power than traditional gas plants while better addressing community concerns around water use and local pollution.

Main Topics: From rockets to power generation (Priority: 5/5): Hartwig traces a path from childhood building projects and rocket obsession to USC’s Rocket Propulsion Lab, SpaceX, test piloting, wildfire rescue, and finally energy infrastructure on Earth. Arbor’s turbine technology (Priority: 5/5): Arbor applies aerospace tools—oxycombustion, high-pressure turbomachinery, and 3D printing—to build a fuel-flexible supercritical CO2 turbine that can run on natural gas or biomass syngas and capture CO2. AI-driven power demand and turbine shortages (Priority: 5/5): The conversation centers on exploding demand for firm power from data centers, while traditional turbine manufacturers face multi-year backlogs and long build timelines. Speed, modularity, and manufacturing model (Priority: 4/5): Hartwig argues that smaller modular 25 MW units and factory-style production are better suited to today’s need for rapid, phased power deployment than giant combined-cycle plants. Community, air quality, and water stewardship (Priority: 4/5): He emphasizes community engagement, lower criteria pollutant emissions, and water-positive operation as key differentiators for projects near data centers and wildfire-prone regions. Economics and market timing (Priority: 4/5): Arbor’s strategy is to beat conventional natural-gas combined cycle on cost by the early 2030s, leveraging new manufacturing and simulation advances that make startup-scale hardware development feasible.

Key Arguments: Hartwig’s aerospace background is directly relevant because the same engineering techniques used in rockets—3D printing, turbo machinery, and oxycombustion—can be repurposed for next-generation power generation. The current market is unusually open because demand for firm power is rising much faster than the supply chain for conventional turbines can respond. Traditional turbine OEMs are optimized for large, monolithic plants and lack incentives to rapidly reinvent manufacturing, leaving room for a startup to enter with a different architecture. Modular turbines are better aligned with data center needs because they can be deployed in phases, provide redundancy, and avoid multi-year wait times. Supercritical CO2 as a working fluid enables a much smaller machine than traditional gas or steam turbines, while still supporting flexible fuels and carbon capture. Biomass deployment can create a carbon-negative power pathway and turn waste biomass—such as forestry residue—into useful energy while helping reduce wildfire risk. Community acceptance depends on long-term impacts, especially air pollution and water use, so projects must be designed with local input from the start. Arbor believes it can compete on cost without relying on a green premium, making zero-emission and water-positive power economically viable rather than niche.

Data Points: Age at SpaceX role: 21 years old - Hartwig said he was leading rocket engine manufacturing for Crew Dragon right out of school. Crew Dragon mission: First privately developed spacecraft to bring astronauts to orbit and back to the space station - Described the vehicle whose Draco engines he helped produce. Turbine unit size: 25 megawatts - Arbor’s modular turbine product is described as a 25 MW unit. Traditional turbine backlog: Seven-plus years - Hartwig said GE, Siemens, and Mitsubishi have turbine backlogs stretching beyond seven years. Conventional power plant build time: About 5 years - He cited a five-year process even when a turbine is available for a combined cycle plant. Power plant scale: 500 MW to 1 GW - Traditional plants were optimized around very large combined-cycle installations. Redundancy model: N plus two / N plus three - Smaller turbines allow backup capacity so maintenance doesn’t interrupt service. Turbine size comparison: 5 to 10 times larger - Hartwig said similarly sized gas or steam turbines would be far larger than a CO2 turbine. Water production rate: Roughly 1 million gallons per MW per year - Byproduct of oxycombustion in Arbor’s system. Gigawatt-scale water output: 1 billion gallons per year - He extrapolated water production at a gigawatt-scale site. Potential California biomass power: 5 gigawatts - Hartwig said forestry waste could support this amount of carbon-negative power in California. Cost target timeline: Early 2030s - He said Arbor expects to be cheaper than natural-gas combined cycle by then.

Pivotal Quotes: "We like to think of it as as very similar. It's gonna be like rockets for Earth." — Brad Hartwig: Summarizing Arbor’s use of aerospace engineering methods for terrestrial power generation. "What we're seeing right now is there's this massive backlog of turbines from GE, Siemens, Smitsubishi, stretching out seven plus years." — Brad Hartwig: Explaining why the turbine market has an opening for a new entrant. "Our North Star is that we will be cheaper than a natural gas combined cycle power plant." — Brad Hartwig: Defining Arbor’s economic goal and how it intends to compete in the market.

Implications: The episode suggests AI power demand, supply-chain bottlenecks, and local environmental constraints are creating a window for modular, factory-built clean firm power. If Arbor succeeds, it could reshape how data centers and other large loads get electricity.