Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Building AI Factories
TL;DR
Crusoe Energy CEO Chase Lockmiller explains how AI data centers represent history's second-largest infrastructure investment, driven by the economic potential of scalable 'digital labor.' He reveals Crusoe's strategy of building massive AI factories in stranded-power locations like Abilene, Texas, to overcome the industry's critical bottleneck: energized data center capacity.
📈 The Economics of Digital Labor 3 insights
Hyperscaler CapEx exceeds Manhattan Project and space program
The five hyperscalers are collectively investing over $650 billion annually in AI infrastructure, making it the second-largest capital deployment in US history behind only the defense budget.
AI represents scalable digital labor force expansion
Unlike human labor with 20-year lead times, AI enables instant scaling of digital workers through data centers, fundamentally altering the Cobb-Douglas economic production model for GDP growth.
Data centers amalgamate every engineering discipline
These facilities represent the physical consolidation of electrical, mechanical, chemical, and computer engineering, requiring massive power and cooling systems to operate intelligence infrastructure.
⚡ The Energy-First Strategy 3 insights
Crusoe inverts traditional data center location logic
Instead of building in traditional hubs like Northern Virginia, Crusoe moves data to stranded energy assets in locations like Abilene, Texas, where renewable over-production caused negative power prices.
Abilene campus scales to 2.1 gigawatts for OpenAI and Oracle
The site hosts Project Stargate with eight buildings powered by a 1-gigawatt private substation—the largest in the US—plus a 350-megawatt natural gas plant, employing 9,000 workers daily.
Renewable tax credits created stranded power opportunities
West Texas wind and solar developers built excess capacity incentivized by production tax credits, but lacked transmission infrastructure, creating ideal conditions for power-hungry AI compute clusters.
🏗️ Operational Bottlenecks and Vertical Integration 3 insights
Bottleneck shifted from chips to energized data centers
While GPU availability has improved, the primary constraint is now finding 'power shells'—physical locations with available high-voltage infrastructure ready to energize massive chip deployments.
Vertical integration unblocks infrastructure development
Crusoe handles everything from energy development and substations to buildings and cooling, allowing rapid response as bottlenecks move between components like switchgear, chillers, and transmission.
Coherent cluster architecture enables massive training jobs
The Abilene facility interconnects all data centers on a high-performance backend network, allowing a single training workload to run simultaneously across all chips in multiple buildings.
Bottom Line
Developers must secure stranded power assets in non-traditional locations before attempting to deploy GPU clusters, as energized data center capacity—not chip availability—is now the primary constraint on AI scaling.
More from Stanford Online
View all
AI in Healthcare Series: Inside the Rise of AI in Healthcare, Open Evidence and Cyber Risks
Former U.S. Chief Data Scientist DJ Patil warns that healthcare systems are dangerously unprepared for AI-enabled cyberattacks from nation states, while simultaneously seeing rapid democratization of medical knowledge through tools like Open Evidence that are fundamentally reshaping the doctor-patient relationship.
Stanford CS153 Frontier Systems | Scale, AGI, and the Future of Everything
Sam Altman explains how AI has fundamentally altered startup economics, enabling small teams to achieve unprecedented scale, while sharing OpenAI's journey from research lab to product company and arguing that pushing systems beyond conventional scaling limits often reveals emergent properties that consensus thinking misses.
Stanford CS547 HCI Seminar | Spring 2026 | The Modern Motivators of Play
The speaker challenges the game industry's outdated assumption that players primarily seek competition, presenting 2024 data showing only 18% of gamers are motivated by competition while 50% seek stress relief and 40% want community. They introduce a framework of nine motivators divided into classic (Fun, Mastery, Competition, Immersion, Meditation, Comfort) and modern (Self-expression, Companionship, Education), arguing that successful games must layer social and creative motivators onto traditional designs to serve contemporary player needs.
Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Applications, Applied AI
Base 10 CEO Tuhin explains why AI inference is shifting from frontier models to custom post-trained models as companies scale, driven by 70-90% cost savings, latency requirements, and the strategic need to own proprietary data rather than feed it to potential competitors.