Brookfield wants to build AI data centers in London’s answer to Wall Street

Brookfield Asset Management says AI infrastructure is its top theme, highlighting a multi-gigawatt global data-center portfolio and plans to add AI data centers in London’s Canary Wharf. CEO Connor Teskey links the opportunity to surging energy demand, digitalization, and supply-chain rewiring, arguing investment is supported by long-term contracts and productivity gains from AI. While acknowledging some “froth,” he emphasizes greater investment discipline rather than bearish sentiment on a data-center bubble.

Analysis

This is a confirmation event for the capital stack behind AI, not a near-term earnings inflection. Brookfield is signaling that the scarce asset is no longer only GPUs; it is power-ready real estate with long-duration contracts, which should keep pricing power with infrastructure capital providers and away from generic office landlords. In the next 1-3 months, the incremental winner is BAM through higher fee-related earnings visibility and fundraising optics, but the real monetization likely shows up over 6-18 months as project equity, construction financing, and asset sales close.

The second-order winners are grid equipment, substation, and power-generation names that can clear interconnection queues, while the losers are traditional London office owners without dense power capacity or conversion optionality. Canary Wharf is a useful tell: if prime office stock can be repurposed for compute, then the market should start valuing urban real estate on megawatts-per-square-foot rather than rent rolls alone. That argues for a rerating gap between power-enabled assets and obsolete offices widening over time.

Contrarianly, the market is probably overestimating how quickly Europe can convert AI enthusiasm into cash flow. Government-led adoption is slower, more political, and more sensitive to electricity pricing and permitting than hyperscaler-led U.S. builds, so the main failure mode is not lack of demand but a 12-24 month bottleneck in grid connection, cooling, and approvals. If power costs rise or connection timelines slip, the bubble narrative returns fast and would hit the highest-duration infrastructure stories first.

Key Decisions for Investors

• Long BAM over the next 3-6 months on any pullback; thesis is fee-earning AUM and project origination optionality, with downside if fund-raising commentary does not convert into signed contracts by the next earnings print.
• Pair trade: long BAM / short a basket of UK office REITs with weak power conversion optionality (e.g. BLND.L, LAND.L) over 6-12 months; goal is to capture the spread between power-enabled assets and obsolete office stock.
• Buy NVDA only on weakness, not chase strength; this is a validation of AI capex persistence rather than a fresh demand surprise. Use a 3-6 month horizon and treat any underperformance in data-center order commentary as the falsifier.
• Set a watch item on European grid and utility bottlenecks: if interconnection timelines or electricity tariffs worsen, reduce exposure to the broader AI-infra complex and rotate toward contracted cash-flow names only.