Last week’s earnings confirmed what Wall Street has suspected all year: Big Tech isn’t easing up on artificial intelligence investments.
Amazon boosted its full-year capital expenditure forecast and said spending will rise again next year. Alphabet raised its capex guidance for the third time this year, signaling another “significant increase” in 2026. Microsoft’s CFO Amy Hood said the company’s AI investments will “grow even faster” in fiscal 2026 than in 2025.
The reason? Demand for computing power continues to outpace supply. Both Microsoft and Meta admitted that their internal AI teams need more infrastructure capacity to meet skyrocketing workloads. Citi analysts estimate that global cloud data center spending will jump 24% in 2026, benefiting chipmakers like Nvidia (NVDA), Broadcom (AVGO), and AMD (AMD).
Meta’s AI Costs Rattle Investors
While AI investment has fueled optimism across Wall Street, Meta’s (META) aggressive spending has become a cautionary tale. The company’s shares fell nearly 12% last week after raising its capex outlook and warning that expenses will surge again next year.
Meta’s total costs rose 32% year-over-year, compared to 12% in the prior quarter. Its heavy AI hiring—often with multimillion-dollar pay packages—has strained margins. Reports of layoffs inside its AI division suggest internal pressure to rein in spending. JPMorgan analysts noted that Meta’s costs are “outsized relative to Google and Amazon,” both of which have direct monetization paths through their cloud businesses
Google’s AI Search Turns Skepticism Into Strength
Alphabet (GOOGL), once seen as lagging in the AI race, surprised markets with stronger search growth—thanks to AI. Its new “AI Overviews” and “AI Mode” features helped boost query volumes rather than cannibalize them, as many feared.
Search revenue rose from 10% in Q2 to 15% in Q3, while paid clicks increased 7%. Analysts at Bank of America said the results “dispel fears that chatbots would replace search,” showing instead that AI is expanding the information economy.
Microsoft and Amazon Keep Wall Street Calm
Executives at Microsoft and Amazon reassured analysts that demand for AI services remains broad-based. Microsoft’s record $392 billion backlog—up 51% year-over-year—covers “customers of all sizes,” CFO Amy Hood said. CEO Satya Nadella emphasized that “the wider enterprise adoption cycle is just starting.”
Behind the AI Boom: Rising Debt and Risk
Yet not everything about the AI revolution is bullish. Debt financing is quietly becoming a key fuel source. Tech firms issued $75 billion in AI-related investment-grade debt in just two months this fall, led by Meta, Oracle, and Alphabet. Oracle alone has raised over $25 billion this year, as it finances $1 billion-per-year data centers to power OpenAI’s infrastructure.
Meta secured $27 billion in off-balance-sheet financing through private lenders like Blue Owl Capital—mirroring the financial engineering seen before the 2008 crisis. Meanwhile, CoreWeave and TeraWulf issued “junk” bonds to fund AI data centers, drawing scrutiny from the Bank of England and Morgan Stanley, which warned of potential “overbuild risk.”
Analyst Gil Luria of D.A. Davidson compared the situation to a “credit-fueled AI bubble,” noting that while Microsoft’s balance sheet remains solid, smaller players are “taking on debt to fund demand that isn’t there yet.”
From Wall Street’s perspective, the question isn’t whether AI will change the world—but whether it will pay off before the bill comes due.
If AI adoption delivers the promised productivity boom, the debt will look visionary. If not, as Bain & Co. warned, the sector could fall $800 billion short of the revenue needed to sustain its infrastructure spending by 2030.
For now, investors seem willing to believe in the AI future—just as they did during the dot-com era. The difference this time? The stakes are far larger, and the balance sheets far heavier.
Disclosure: This article does not represent investment advice. The content and materials featured on this page are for educational purposes only.
