Morgan Stanley: $700B may reach Nvidia by 2027, but US data centers face 32GW power gap
rohanpaul_ai · x · 2026-10-20
Morgan Stanley rates Nvidia AI's best risk/reward while flagging concentration and power risks.
Bull case
- Hyperscalers and AI cloud providers are expected to spend $1.58T in 2027, up 60%, with nearly $700B reaching Nvidia
- At 15x fiscal 2028 earnings the stock is cheap; analysts underestimated the past 4 compute cycles by 38% on average
Risks
- 91.7% of 2027 revenue is exposed to infrastructure demand
- Two buyers—SpaceX and AWS—account for 54% of 2027 hyperscaler revenue and 72% of growth
- MS's own power model projects US data centers falling 32GW short of the 97GW needed through 2028
More from Venture
- Daily AI Brief: Unannounced GPT-6.1 Terra spotted in Microsoft code, Anthropic intent-reading model due this month — testingcatalog · 2026-10-20
- NVIDIA invests in 5 UK startups and commits $250M compute for agentic safety research — dr_alphalyrae · 2026-10-20
- VC hot take: if they route you to their EA or Calendly, they're not investing — vaibhavbetter · 2026-10-20
- Pokemon card bought as 'AGI diversification' is now up 20x — gabriel1 · 2026-10-20
- Google ready to invest up to $40B in Anthropic, pending conditions — emmanuelvivier · 2026-10-20
- BofA: AI capex running 50% above forecasts, hyperscaler cost of capital jumps 4% to 10% — rohanpaul_ai · 2026-10-20