Powering AI is an architecture problem, not a power problem, says MIT Tech Review
nordicinst · x · 2026-09-10
MIT Technology Review argues AI data center outages are architecture failures, not supply failures. A July 2026 transmission fault in Ashburn, Virginia knocked over 3 GW of load offline in seconds; two years earlier a single failed surge arrester dropped 60 facilities and 1,500 MW. AI campuses swing up to 70% of load in milliseconds during training runs — a behavior the grid wasn't built for. The proposed fix: moving power protection up the voltage stack, outside the building, and into the power path with medium-voltage AI UPS, which also reshapes density, permitting timelines and backup power economics.
More from Infra
- Meta approves nonvoting Class C shares as 2026 capex guidance hits $135B–$145B — thione · 2026-10-26
- Unsloth Ships MiniMax-M3.1-Flash Quants: 2-bit Retains 91.5% Accuracy, Runs in 150GB RAM — danielhanchen · 2026-10-26
- Engineer with thousands of fabricated devices pours cold water on photonic crystal modulator yields — jwt0625 · 2026-10-26
- Anthropic S-1 reveals 48 cents of compute cost per dollar of Q2 revenue — nmasc_ · 2026-10-26
- Persistent AI joins UR network to route private AI inference with encrypted, IP-hidden traffic — 0xSammy · 2026-10-26
- Anthropic's disclosed compute commitments: $114.5B Google, $161.2B Broadcom top the list — BenBajarin · 2026-10-26