Crusoe CEO Chase Lochmiller on AI infrastructure bottlenecks, blue-collar workforce, and the ratepayer protection pledge

Summary

Chase Lochmiller, CEO of Crusoe, an eight-year-old vertically integrated AI infrastructure company, identifies energized data center capacity as the single most binding constraint on AI buildout today. Chips themselves are not the bottleneck; getting power to those chips is. Labor is a co-equal constraint alongside power, with electricians and construction workers representing a meaningful supply gap across the entire data center ecosystem.

Infrastructure Bottlenecks

Lochmiller frames Crusoe's stack as spanning land acquisition, power procurement, data center engineering and deployment, GPU cluster orchestration, self-healing cluster management via its Auto Clusters product, managed Kubernetes, and a managed inference product for model hosting. The software layers are becoming easier to build, but the physical layer remains structurally constrained.

On the skilled trades question, Lochmiller draws a deliberate distinction from the software world. A top-performing electrician might reduce a job from 40 hours to 30. The leverage differential versus a 10x software engineer is real, but the constraint is volume, not individual output variance.

Workforce Development

Lochmiller says he is personally focused on reskilling the labor force for the AI infrastructure build-out. His preferred models include AI-accelerated trade school curricula modeled on Alpha School's personalized learning approach, apprenticeship programs with guaranteed job placement on completion, and a reference to Lambda School's income-share coding model as an analogue for the trades. He also argues that the decline of hands-on k-12 programs like auto shop and wood shop has compounded the pipeline problem over time.

Natural Gas and Energy Independence

On Middle East conflict risk, Lochmiller argues Crusoe is largely insulated because its primary fuel exposure is to Henry Hub natural gas pricing, not globally traded oil. U.S. domestic natural gas is regionally priced; the liquefaction cost embedded in LNG exports creates enough friction that domestic spot prices did not sustain the initial spike triggered by the conflict and returned to pre-conflict levels. Oil does not carry the same friction, making U.S. producers more exposed to global arbitrage. The Trump administration's energy independence posture and private sector investment in domestic gas production are credited for the stability.

Ratepayer Protection Pledge

Lochmiller is constructive on the ratepayer protection pledge, framing it as consistent with Crusoe's founding thesis around on-site power generation, which originated with flared gas monetization. His forward-looking model is built around what he calls "across the meter" infrastructure, a bidirectional grid architecture where an AI data center co-locates with an on-site generation asset, whether gas, wind, solar, or battery storage, connects to the grid at a single interconnection point, draws from the grid when on-site generation falls short, and exports surplus power back to local ratepayers when generation exceeds load.

The commercial logic is that AI data centers become net contributors to grid supply during off-peak generation periods, putting downward pressure on retail electricity rates rather than inflating them. Lochmiller acknowledges the pledge carries no legal enforcement mechanism at this stage, but characterizes it as a directionally correct platform for large, well-capitalized AI companies to self-fund generation and transmission upgrades at a scale that public utilities have not historically managed.