By ByteBot
The Bank of England warned TODAY that a $5 trillion AI infrastructure spending boom—financed largely by debt—could trigger broader market unraveling. In its Financial Stability Report released December 2, 2025, the UK central bank flagged “materially stretched” valuations and estimated that roughly half of the projected spending through 2030 will come from external debt financing rather than company cash flows. This marks the first time a major central bank has explicitly connected AI investment to systemic financial risk.
While AI hype dominates headlines, financial regulators see something different: a debt-fueled bubble forming beneath the surface.
The Staggering Scale of AI Debt
The numbers are unprecedented. McKinsey estimates $5.2 trillion in capital expenditures will be needed for AI data centers by 2030, with the Bank of England projecting approximately $2.5 trillion of that total coming from debt markets. Morgan Stanley Research breaks it down further: $2.9 trillion in AI infrastructure spending between 2025 and 2028, with $1.5 trillion financed externally—including $800 billion from private credit alone.
This represents a fundamental shift from how tech infrastructure has historically been funded. Hyperscalers like Google, Microsoft, and Amazon traditionally built data centers using operating cash flows. However, the AI boom’s scale and speed demand something different: massive external borrowing. But debt amplifies both gains and losses. Consequently, if AI revenue disappoints, companies can’t simply walk away—they owe billions to creditors.
The Bank of England didn’t mince words: “If material credit losses on AI lending were to occur, directly or indirectly, this could have spillovers to broader credit conditions, including in the UK.”
Oracle: The Canary in the Coal Mine
Abstract warnings matter less than concrete stress signals. Enter Oracle, which borrowed $38 billion in October 2025 to build Texas and Wisconsin data centers primarily for OpenAI workloads. The market’s reaction? Oracle’s five-year credit default swaps—essentially insurance against the company defaulting—spiked to 1.25%, a three-year high.
That’s more than triple the spread from 2023. Credit markets are pricing real risk here.
Morgan Stanley warned that Oracle’s net adjusted debt will balloon to approximately $290 billion by fiscal year 2028, up from around $100 billion today. Analysts flagged the “inflated OpenAI-related backlog” as uncertain—Oracle’s betting $38 billion that OpenAI’s demand fully materializes. Moreover, if it doesn’t, Oracle’s in trouble. And if Oracle’s in trouble, who’s next?
Credit default swaps don’t lie. They’re how smart money prices risk before equity markets catch on. Think Bear Stearns in 2008—CDS spreads widened months before the collapse. Oracle’s spike is an early warning worth heeding.
Dot-Com Bubble Déjà Vu—With a Twist
The Bank of England drew an explicit comparison few analysts want to make: current AI stock valuations resemble the dot-com bubble peak. The cyclically adjusted price-to-earnings (CAPE) ratio sits close to 40, matching levels not seen since the tech crash 25 years ago. AI-related stocks now comprise 44% of the S&P 500’s market capitalization, with the top five companies representing 30% of the index—a 50-year high.
But here’s the critical difference: dot-com companies had no earnings. They were ideas, vaporware, PowerPoint decks. Federal Reserve Chair Jerome Powell noted as much: “These companies actually have earnings, unlike the dot-com period when these were ideas rather than companies.”
AI companies are profitable. Microsoft, Google, and Alphabet print money. Additionally, the forward P/E ratio for AI stocks sits around 26x—elevated, yes, but not the 70x multiples seen in 2000. So what’s the problem?
Debt amplification. The dot-com bubble was mostly equity-financed speculation. In contrast, the AI boom is leveraged to the hilt with $2.5 trillion in borrowed money. Equity corrections hurt investors. Debt defaults trigger contagion. That’s why the Bank of England is worried—and why you should be too.
What Developers Should Watch
The Bank of England’s warning is early, which is good for preparedness but frustrating for timing. Dot-com warnings appeared in 1998; the crash didn’t hit until March 2000. Bubbles can inflate for years after first alarms sound. Nevertheless, early doesn’t mean wrong.
Watch for these signals that trouble’s brewing:
More CDS spikes beyond Oracle. If other AI infrastructure companies see credit spreads widen, the stress is spreading.
Private credit losses. $800 billion is exposed to AI infrastructure loans. The first defaults will reveal how concentrated and interconnected these risks really are.
Funding environment shifts. If venture capitalists get cautious, AI startup hiring slows fast. Specifically, check whether those 49 US AI startups that raised $100 million-plus in 2025 can keep the party going.
Tech stock corrections. AI drove two-thirds of the S&P 500’s gains in 2025. A reversal means broad market impact, not just AI names.
Career resilience matters here. Don’t specialize solely in AI and LLMs—maintain skills in “boring” infrastructure like databases and security. Furthermore, if you’re at an AI startup burning over $1 million monthly with less than 12 months of runway, get nervous. Companies with minimal debt and positive unit economics will outlast leveraged competitors if funding dries up.
The debate breaks along predictable lines. Optimists like Jeff Bezos argue bubbles can be good: “When the dust settles and you see who the winners are, society benefits from those inventions.” Goldman Sachs remains cautiously optimistic, saying a bubble hasn’t formed yet but warning investors to diversify. Meanwhile, Chris Wood at Jefferies predicts a “massive overinvestment bust,” calling $350 billion in annual AI infrastructure spending unsustainable.
Both sides have valid points. AI is transformative technology with real value. AND the buildout is debt-fueled and potentially over-leveraged. These aren’t contradictory statements.
Key Takeaways
- The Bank of England flagged a $5 trillion AI infrastructure boom financed by $2.5 trillion in debt as a systemic financial risk
- Oracle’s credit default swaps spiked to 1.25% after borrowing $38 billion for AI data centers—an early stress signal
- AI stock valuations match dot-com bubble levels (CAPE ratio ~40), but companies have real earnings unlike 2000
- Debt amplification makes this riskier than the dot-com era, which was mostly equity-financed speculation
- Developers should diversify skills, monitor employer debt levels, and prepare for potential funding environment shifts
Central banks rarely cry wolf. When they do, smart people pay attention.
