AI Data Center Cost Per GW Jumps 40%
tengyanAI · x · 2026-07-14
This comment cites Morgan Stanley's view on AI infrastructure capex and adds a key breakdown: per GW cost surges by about 40% from GB200 to Vera Rubin, driven primarily not by the chips themselves, but by outside-the-rack power and powered shell deployment costs.
The cited content also notes:
- Morgan Stanley raised its 2028 total capex forecast for hyperscalers to $1.4 trillion, up from the previous estimate of around $1.2 trillion for 2027.
- Total hyperscaler compute capacity is expected to grow from about 30 GW in 2025 to 120 GW by 2028.
- The cost increase is tied to memory and "outside the rack" powered shell development.
Related event: Morgan Stanley Raises Hyperscaler Capex Forecast to Nearly $1.5T(3 posts)→
More from Infra
- OpenRouter agents now out-consume humans as AI usage arrives in three waves — AccBalanced · 2026-09-11
- Nvidia Is Now Core to Every Major Robotaxi Stack at Commercial Scale — pdamodaran · 2026-09-11
- 12 KV Cache Reduction Techniques Every AI Engineer Should Understand, Explained — blaizedsouza · 2026-09-11
- The shadow GPU capacity market is formalizing, with Meta selling excess compute to outside buyers — DavidLinthicum · 2026-09-11
- Engram's random reads don't suit SSDs; CPU-memory over NVLink could serve all 72 GPUs — bookwormengr · 2026-09-11
- 80% of the DIY LLM inference hype posters have already quit — it's brutally hard systems work — abhijithneil · 2026-09-11