AI datacenter power is hitting a 36-month turbine bottleneck
tengyanAI · x · 2026-07-22
The binding constraint on new AI datacenter power may not be turbines alone, but the manufacturing capacity for rotor and hot-section parts.
- Gas turbine orders are running ahead of what qualified makers can produce, creating roughly 36-month delivery queues.
- Supply is inelastic and concentrated among a small number of producers who cannot ramp output within a single order cycle.
- Howmet’s single-crystal nickel-superalloy blades are a key bottleneck for finished turbine output, giving it pricing power as GE Vernova and Siemens Energy backlogs grow.
Related event: New AI Compute Bottleneck: Gas Turbine Lead Times Exceed 3 Years(2 posts)→
More from Infra
- SmolVM open-sources persistent computer infrastructure for agents that outlive chat sessions — aniketmaurya · 2026-09-11
- PyTorch Day Korea 2026 launches first offline conf, CFP closes Sept 13 — PyTorch · 2026-09-11
- Local LLM server dilemma: 4x CMP-170HX (price up 53% in 20 days) vs Mac Studio M5 Ultra — rumboll · 2026-09-11
- llama.cpp lands Flash Attention tuning for RDNA4, big prefill gains on AMD — pmttyji · 2026-09-11
- Your p99 latency benchmark may be lying: a deep dive into coordinated omission — Franc0Fernand0 · 2026-09-11
- Running MiniMax H3 on 12GB VRAM: quantization, Turbo LoRAs and attention backends compared — Possible_Mood676 · 2026-09-11