Is There Enough Data Center Capacity for AI?

Goldman Sachs Research projects AI will account for roughly 30% of the data-center market within two years and forecasts overall data-center power consumption will rise about 175% from 2023 levels by 2030 (up from a prior 165% estimate). In its base case the industry builds sufficient capacity and occupancy peaks near 93% next year, but downside scenarios — driven by a slowing economy or weaker AI monetization — could push occupancy down to roughly 80% by decade-end, undermining operators’ returns on capital. The note signals that capacity additions, power constraints and the ability to monetize AI workloads will be key drivers of margins, earnings and valuations across data-center and related infrastructure stocks.

Analysis

Goldman Sachs Research projects AI will account for roughly 30% of the overall data‑center market within two years and forecasts aggregate data‑center power consumption will rise 175% from 2023 levels by 2030, up from a prior 165% estimate. The report frames that surge as a material demand shift that will take share from conventional and cloud workloads and drive significant infrastructure needs.

In Goldman’s base case the industry builds sufficient capacity and occupancy peaks near 93% next year, supporting stronger margins and returns for operators; under an "AI Downside" scenario occupancy falls to about 80% by decade‑end if the economy slows or AI monetization weakens. Goldman explicitly links occupancy to margins and earnings, signaling that supply sufficiency and pricing power are primary drivers of valuation outcomes for data‑center players.

Key execution risks are inadequate capacity additions, failure to monetize AI workloads and power constraints given the projected jump in consumption; these will determine whether elevated demand translates into sustainable earnings. The note and accompanying signals register a mildly positive but cautious tone with a moderate market‑impact score, implying investors should monitor capex announcements, occupancy trends and utility/power commitments closely.

Key Decisions for Investors

• Consider overweighting data‑center operators and critical infrastructure suppliers that demonstrate disciplined capacity expansion and clear pricing power given the base case of ~93% occupancy and a projected 175% rise in power demand by 2030
• Monitor leading indicators—published occupancy rates, announced capex plans, evidence of AI workload monetization and utility/power capacity commitments—and be prepared to trim or hedge positions if occupancy trends move toward the ~80% "AI Downside" scenario
• Maintain defensive sizing or hedges (short‑duration exposure, options protection) to guard against execution risk from slower monetization or power constraints, reflecting the report's cautious tone and moderate market‑impact signal