UBS: Data-center boom undermines ’AI bubble’ — 2026 market growth raised to 20–25%
UBS updated its deep-dive on the global data-center equipment market, saying capacity expansion shows no signs of cooling. Evidence Lab data: ~25 GW under construction and ~105 GW operational. UBS now expects market growth (power, cooling, IT equipment) of 20–25% in 2026 after a 25–30% rise in 2025, with 2027–2029 easing to 15–20% and 10–15% annual growth in 2028–2030. Low vacancy rates across regions (North America 1.8%, Europe 3.6%, APAC 5.8%) and strong cloud capex underpin the outlook. Liquid cooling stands out: UBS forecasts cooling segment CAGR ~20% to 2030, with liquid-cooling growing ~45% — driven by higher AI chip power density. AI data centers raise per-MW facility costs ~20% and IT-equipment costs 3–4x versus traditional centers, reducing tenant price sensitivity and benefiting upstream suppliers. UBS estimates GenAI-related annual recurring revenue (ARR) ~US$17bn, signaling early monetization and countering “AI bubble” narratives. Constraints include power and grid bottlenecks (notably in Europe) and delivery risks; UBS views these as value-supporting rather than cycle-ending. Technological shifts (rack power density to 100kW+, move toward 800V DC distribution) will create winners and losers among MV/LV equipment vendors. Key takeaways for traders: sustained high capex by hyperscalers through at least 2027, accelerating adoption of liquid cooling and AI-specific infrastructure, and rising demand for chips, IT hardware, power and cooling suppliers — supportive for hardware, infrastructure and select cloud-equipment stocks.
Bullish
UBS’s report signals sustained, multi-year capital expenditure from hyperscalers and broad-based demand across power, cooling and IT equipment — a structural growth story rather than a short-lived hype. Key bullish points: large pipeline (25 GW under construction), very low vacancy rates, strong 2025–2026 growth forecasts (25–30% then 20–25%), and early GenAI monetization (ARR ≈ US$17bn). Liquid cooling and AI-specific hardware command higher per-MW value, raising vendor pricing power and margins. Historical parallels: past infrastructure-driven rallies (e.g., cloud capex cycles, mobile-data rollouts) lifted hardware, chipmakers and specialist suppliers for multi-year periods, with outperformance concentrated among vendors with proprietary tech or full product stacks. Short-term impact: likely positive sentiment for related equities (chips, server OEMs, cooling/power vendors, selected cloud suppliers) and potential re-rating as revenue visibility improves. Expect episodic volatility around supply-chain or grid constraint headlines. Long-term impact: durable demand and faster refresh cycles for AI servers could support sustained revenue growth for infrastructure suppliers, favoring companies exposed to liquid cooling, high-voltage DC power distribution and high-density rack solutions. Risks that could temper bullishness: worsening macro, capex cuts by cloud players, or prolonged grid/connectivity constraints that delay capacity bring-up. Overall, net effect for crypto market participants is indirect but positive for infrastructure token-linked plays and stocks tied to cloud/compute — increased demand for compute capacity supports on-chain service providers and cloud-dependent crypto projects.