Why AI Data Centers Lower Energy Costs and Save Water
Crusoe CEO Chase Lochmiller debunks myths on AI data center water use and explains why massive compute builds can reduce local electricity rates.
Crusoe CEO: Why Everyone Gets GPU Depreciation & AI Energy Costs Wrong
With Harry Stebbings, Chase Lochmiller · Sunday, October 4, 2026
Chase Lochmiller, Co-founder and CEO of Crusoe, discusses the infrastructure, energy, and hardware economics powering the AI revolution following their $3.9 billion Series F. He details why traditional assumptions around data center water usage and local energy price hikes are incorrect, how Crusoe vertically integrates electrical manufacturing to bypass critical supply chain bottlenecks, and why GPU depreciation cycles will outlast market expectations.
Key takeaways
- Crusoe CEO Chase Lochmiller disputes the popular claim that modern AI data centers drain local reservoirs, pointing to closed-loop liquid cooling designs. Read more →
- When Crusoe planned its Abilene site to build over 200 megawatts of compute capacity, they committed to delivering in one year while competing bids required two and a half years. Read more →
- Crusoe closed a $3.9 billion Series F by modeling its compute business on vertically integrated oil super majors like Exxon and Chevron. Read more →
- Standard venture capital models write off high-end GPUs over three years, expecting rapid obsolescence to destroy their economic value. Read more →
- GPUs represent the single most expensive physical asset inside an AI data center; keeping compute engines idle destroys unit economics. Read more →
- Industry estimates show roughly 50% of announced AI data centers will never come online due to severe execution roadblocks. Read more →
6 articles from this episode
Crusoe Cut Power Center Lead Times from 100 Weeks to 28
Chase Lochmiller reveals how Crusoe builds data centers in 12 months by manufacturing power distribution centers in-house.
Why Crusoe Built a $3.9B AI Super Major Model
Chase Lochmiller explains why Crusoe mirrors Exxon's vertical model across data centers, GPUs, and inference tokens to hedge AI compute margins.
Why GPU Depreciation Math Is Wrong
Crusoe CEO Chase Lochmiller explains why older GPUs keep earning revenue long after standard three-year venture capital write-offs predict they expire.
Why Memory Bottlenecks Drive AI Inference Margins
Crusoe CEO Chase Lochmiller explains how KV cache management across HBM, DRAM, and NVMe dictates GPU utilization and inference unit economics.
Why 50% of Planned AI Data Centers Will Never Open
Crusoe CEO Chase Lochmiller reveals why half of all planned AI data centers will fail to break ground due to power, labor, and land hurdles.
More 20VC with Harry Stebbings episodes
- AMD Buys Fei-Fei Li's World Labs for $8.2B | Meta Poaches MongoDB's CEO | Bessemer Raises $5.75BOct 4 · 6 articles
- The Untold Story of Higgsfield | Burning $4M a Month on AI Models | CEO, Alex MashrabovOct 4 · 5 articles
- The Ads Business Model Will Die & Lessons from Working with Elon at Twitter | Parag AgrawalSep 27 · 5 articles
- Jensen Huang Declares AGI Has Arrived | Tesla Launches Cybercabs | Index Pulls Out of TownSep 13 · 7 articles