Gartner: $3.5T of AI infra spend may face write-downs by 2030, 80% of neoclouds to fold
rohanpaul_ai · x · 2026-10-20
Gartner forecasts AI providers will spend nearly $1T on infrastructure in 2026 plus $6T from 2027-2030, and warns over $3.5T of that could need write-downs by end of 2030 as revenue won't cover costs.
Key points:
- Most AI-native providers run asset-light, so less than half of AI revenue reaches companies owning significant infrastructure.
- GPU-rental neoclouds are most exposed; Gartner expects 80% to cease independent operations by 2030.
- AI-optimized IaaS spending grows only from $66B (2027) to $174B (2030), too slow for most providers to recoup investment before current AI chips are replaced.
Related event: Gartner Predicts Nearly $7 Trillion in AI Infrastructure Spend by 2030(3 posts)→
More from Infra
- Semicon West & OCP recap: $300B+ wafer-equipment TAM, Tesla TeraFab orders, TSMC N5-to-N3 conversions — BenBajarin · 2026-10-20
- Qualcomm and Liquid AI push on-device AI with Liquid Context, compressing multimodal inputs on Snapdragon — JosephJacks_ · 2026-10-20
- PyTorchCon enterprise day features executive sessions with Red Hat, NVIDIA and IBM — PyTorch · 2026-10-20
- SOAP & SPlus scale pretraining across 96+ DiLoCo workers via update whitening — aaron_defazio · 2026-10-20
- Amazon CEO says AWS is out of compute until beyond 2028 — rohanpaul_ai · 2026-10-20
- Nebius launches live spot GPU price comparison across regions, with API access — kevinsxu · 2026-10-20