AI’s $10 Trillion Buildout Will Reach 3.6% of GDP. Who’s Going to Pay for It?
Big Tech is funneling nearly $200 billion into AI infrastructure every single quarter, a share of the economy larger than the entire railroad expansion. But the balance sheets tell a more complicated story about who is actually footing the bill.
This post may contain links from our sponsors and affiliates, and Flywheel Publishing may receive compensation for actions taken through them.
AI infrastructure spending is expected to total roughly $10 trillion and average about 3.6% of GDP a year through the early next decade, according to a piece on Fox Business’s The Bottom Line. Amazon (NASDAQ:AMZN | AMZN Price Prediction), Alphabet (NASDAQ:GOOGL), Microsoft (NASDAQ:MSFT), Meta (NASDAQ:META), and Oracle (NYSE:ORCL) are already pouring tens of billions of dollars into data centers every quarter.
That is a larger share of the economy than the railroad expansion took, and far larger than the Interstate Highway System infrastructure buildout, prompting Madison Alworth to warn during the segment: “Our economy has never been so dependent on the buildout of one industry.”
Big Tech Is Pouring Nearly $200 Billion a Quarter Into AI
Hyperscalers are putting roughly four-fifths of their cash flow back into the business. As an analogy, this could be compared to spending income rather than running up credit card debt, a real difference from the past leveraged railroad boom and the dot-com bubble.
Much of the demand is also under contract. At Microsoft, commercial remaining performance obligations (signed business not yet delivered) rose 84% to $678B. Remaining performance obligations were $664B at Oracle. Amazon’s AWS backlog reached $496 billion.
Debt and Off-Balance-Sheet Financing Are Rising
Companies also use off-balance-sheet structures including joint ventures, leases and partner-funded vehicles. A company commits to pay for capacity while outside investors hold the asset and debt, creating a real commitment that never shows up as corporate borrowing. If a significant share of spending goes through these structures, it means that a clean balance sheet may be hiding a significant amount of borrowing.
Additionally, traditional borrowing that does show up on the balance sheet is rising too. Alphabet raised roughly $70B in combined equity and debt and stopped its buyback. Long-term debt at Meta reached $83.66B, and its quarterly free cash flow fell to $784M from $8.55B. Free cash flow was -$7.6B at Amazon over the trailing twelve months and -$5.4B at Oracle for the quarter.
The $10 Trillion AI Number Is Still Just a Forecast
The $10 trillion figure is a long-term forecast, not money that has already been committed. However, spending is already enormous, with private data center construction spending through July exceeding all other private construction combined.
Big Tech’s numbers tell the same story. In their latest quarters, capex reached $54.21 billion at Amazon, $44.92 billion at Alphabet, $35.80 billion at Microsoft, $30.12 billion at Meta, and $28.5 billion at Oracle. Combined, that’s nearly $194 billion in a single quarter.
Index Investors Already Own the Biggest AI Spenders
Most investors own all of AI’s biggest spenders if they hold index funds:
| Stock | Price (Sept. 25) | YTD | 1-Year |
|---|---|---|---|
| GOOGL | $343.92 | 10.09% | 40.29% |
| AMZN | $249.67 | 8.17% | 14.45% |
| META | $751.66 | 14.15% | 0.69% |
| MSFT | $516.17 | 7.41% | 2.64% |
| ORCL | $137.08 | -28.98% | -52.4% |
Oracle has fallen the furthest this year due to the company’s highest levels of AI financing stress. Microsoft stayed free-cash-flow positive at $19.6 billion for the quarter.
AI’s Next Bottleneck Could Be Electricity
AI computing is expected to face a power shortfall by the end of the decade. The U.S. Energy Information Administration’s Annual Energy Outlook 2026 projects national electricity use growing 0.9% to 1.6% a year through 2050.
In its high-demand case, data center servers alone could reach 818 billion kilowatt-hours in 2050. At an April House hearing, witnesses pointed to Georgia and Virginia billing data centers more so smaller customers are protected.
Key Takeaways
Amazon, Alphabet, Microsoft, Meta, and Oracle are already spending at a pace few industries have ever matched in American history.
For investors, the dividing line is cash flow. As long as operating cash flow keeps pace with capex, Big Tech can finance much of the AI buildout itself. If that changes, expect more debt, leases, joint ventures, and other outside financing to fill the gap.
Contact [email protected] for any questions or corrections.
Thomas Richmond is a financial writer and content strategist with 5+ years of experience covering stocks and financial markets. He has published over 500 articles focused on individual stock analysis, helping investors better understand business fundamentals, stock valuations, and long-term opportunities.
Thomas previously served as a Content Lead at TIKR, a stock research platform, where he helped scale the company’s blog to hundreds of articles per month and contributed to a weekly newsletter reaching more than 100,000 investors.
Outside of work, Thomas enjoys weight lifting and soccer.
