{"slug": "wall-street-banks-are-being-asked-to-bankroll-trillions-in-ai-data-centers", "title": "Wall Street Banks Are Being Asked to Bankroll Trillions in AI Data Centers", "summary": "Morgan Stanley expects hyperscalers and their joint ventures to issue $250 billion to $300 billion in debt in 2026, part of a broader AI-related debt market topping $570 billion, as Bank of America puts total hyperscaler AI spending at roughly $725 billion for the year, up 77% from $410 billion in 2025. JPMorgan projects $150 billion in AI-related debt deals over the next five years and estimates annual data center securitization issuance could reach $30 billion to $40 billion in 2026 and 2027, while Morgan Stanley expects private credit firms to supply another $800 billion in data center financing over the next two years.", "body_md": "*Hyperscalers are on pace to spend roughly $725 billion on AI infrastructure in 2026, and Morgan Stanley expects banks to underwrite $250 billion to $300 billion of that from hyperscalers and their joint ventures alone.*\n\nThe chips get the headlines. The money behind them doesn't, and that's starting to change. Morgan Stanley now expects hyperscalers and the joint ventures built around them to issue $250 billion to $300 billion in debt this year, part of a broader AI-related debt market the bank sees topping $570 billion in 2026, more than double where it stood twelve months ago. Bank of America puts total hyperscaler AI spending at roughly $725 billion for the year, up 77% from the $410 billion the five biggest cloud companies deployed in 2025.\n\nThat's the number that should worry you. Even Amazon, Microsoft, Google and Meta don't have $725 billion sitting in a checking account. Free cash flow across the group covers a fraction of what they want to build. The rest has to come from somewhere, and increasingly that somewhere is a bank syndicate, a bond desk or a private credit fund.\n\nIn February, Duke Energy unveiled a $103 billion five-year capital plan, the largest ever filed by a regulated US utility, according to Utility Dive. Duke counts Amazon, Microsoft, Google and Meta among its data center customers and added 2.7 gigawatts of newly contracted data center load in the first quarter alone. Southern Company followed with its own increase, raising its five-year plan to $81 billion, up from $63 billion the year before, after locking in 10 gigawatts of contracted large-load customers and identifying a 75-gigawatt pipeline of potential data center interest.\n\nBoth utilities are effectively borrowing against demand that hasn't fully shown up yet. That's the bet. If the AI buildout slows, ratepayers and shareholders are the ones left holding infrastructure sized for a boom that stalled halfway through.\n\nThe physical math doesn't help. Data Center Knowledge reported in December that operators face a projected shortfall of more than 45 gigawatts of power, with utility interconnection queues running years long. Boston Consulting Group puts the eventual gap even higher, projecting a 50 to 80 gigawatt shortfall by 2030. The power just isn't there. Some developers are responding by generating power on-site rather than waiting on the grid, a workaround known in the industry as bring-your-own-power.\n\n## Who's Actually Writing the Checks\n\nJPMorgan projects $150 billion in AI-related debt deals over the next five years and estimates annual data center securitization issuance could reach $30 billion to $40 billion in both 2026 and 2027. The bank has already helped lead a $38 billion syndicated loan funding two data centers in Texas and Wisconsin, plus a separate $18 billion facility for a site in New Mexico. JPMorgan pegs the total funding gap at $1.4 trillion and argues every corner of the capital markets, banks, private credit, insurers, pension funds, needs to show up to close it. That's a big gap to fill.\n\nPrivate credit is moving in fast to fill what banks won't hold on their own balance sheets. That's fast money. Morgan Stanley expects private credit firms to supply another $800 billion in data center financing over the next two years. Outstanding loans to AI-related companies have gone from close to zero to more than $200 billion in just a few years, a pace that would have been unthinkable in traditional infrastructure lending a decade ago.\n\nThis pace is making bond investors nervous. Forbes reported in July that bond investors are starting to push back on AI-linked debt as issuance heads toward $570 billion, demanding wider spreads and tighter covenants on deals tied to companies whose AI revenue hasn't caught up with the capex. That's a real tell. When the people writing the checks start asking harder questions, the terms get worse for every borrower behind them.\n\nMoney can move fast. Turbines and transmission lines can't. Duke's own project timelines show gas turbine orders booked years out. Closing a trillion-dollar funding gap doesn't shrink the wait for the actual hardware. The AI boom now has enough capital lined up to keep building through 2026 and beyond. The power grid, the turbine makers, the bond market: whether they can all move at the same speed is the open question nobody underwriting these deals has fully answered.\n\n**Also read:** [Best AI Agent Platforms for Founders in 2026: A No-Nonsense Comparison](https://startupfortune.com/best-ai-agent-platforms-for-founders-in-2026-a-no-nonsense-comparison/) • [Microsoft Quietly Tests MAI-Realtime, a Voice AI That Talks While It Listens](https://startupfortune.com/microsoft-quietly-tests-mai-realtime-a-voice-ai-that-talks-while-it-listens/) • [OpenAI Fires Back at Apple Trade Secret Lawsuit With Its Own Email Evidence](https://startupfortune.com/openai-fires-back-at-apple-trade-secret-lawsuit-with-its-own-email-evidence/)", "url": "https://wpnews.pro/news/wall-street-banks-are-being-asked-to-bankroll-trillions-in-ai-data-centers", "canonical_source": "https://startupfortune.com/wall-street-banks-are-being-asked-to-bankroll-trillions-in-ai-data-centers/", "published_at": "2026-08-04 09:55:56+00:00", "updated_at": "2026-08-04 10:26:09.634537+00:00", "lang": "en", "topics": ["ai-infrastructure", "ai-policy", "ai-ethics"], "entities": ["Morgan Stanley", "Bank of America", "JPMorgan", "Duke Energy", "Southern Company", "Amazon", "Microsoft", "Google"], "alternates": {"html": "https://wpnews.pro/news/wall-street-banks-are-being-asked-to-bankroll-trillions-in-ai-data-centers", "markdown": "https://wpnews.pro/news/wall-street-banks-are-being-asked-to-bankroll-trillions-in-ai-data-centers.md", "text": "https://wpnews.pro/news/wall-street-banks-are-being-asked-to-bankroll-trillions-in-ai-data-centers.txt", "jsonld": "https://wpnews.pro/news/wall-street-banks-are-being-asked-to-bankroll-trillions-in-ai-data-centers.jsonld"}}