What Wall Street Sees in the Data Center Boom (www.nytimes.com)

Wall Street is watching a surging, high-stakes buildout of data centers to power generative AI: trillions of dollars in demand is reshaping markets and corporate strategy. Recent market moves — Nvidia missing data-center equipment sales expectations, and Oracle’s stock jumping after OpenAI agreed to buy $300 billion in compute — highlight how sensitive equities are to infrastructure signals. McKinsey projects U.S. AI-driven data-center demand could triple by 2030, implying nearly $7 trillion in investment; major players and alliances (OpenAI, Oracle, SoftBank, Meta, Alphabet) are committing hundreds of billions through 2029. Investors view data-center activity as an early indicator of either a sustainable AI cycle or emerging headwinds, with some warning of a speculative boom. The technical and financial implications are concrete: more than $9 billion of CMBS/ABS issuance funded data-center projects in early 2025, Meta issued $26 billion in bonds for expansion, and Microsoft’s scaled-back $1 billion play underscores lease and overcommitment risk (it faces roughly $175 billion in lease obligations). Efficiency improvements — e.g., DeepSeek’s lower-compute models — could curb long-term demand, while energy and water pressures are tangible (a 100 MW data center can use ~2 million liters of water per day; Phoenix expects a 500% rise in data-center power capacity; Virginia plans 40 GW of new capacity). The result: huge growth opportunities for AI compute, but with pronounced credit, environmental, and policy risks that will determine whether the infrastructure rush is profitable or overbuilt.