Big Tech's AI Spending Set to Outpace Cash Flow by 2027

The five biggest tech players are approaching a moment where the AI boom starts costing more than it brings in. A Reuters analysis, based on LSEG consensus forecasts, shows that Microsoft, Alphabet, Amazon, Meta and Oracle will together spend more on AI infrastructure by 2027 than they generate from their own operations.

What the Reuters Analysis Found

The analysis's authors, Patturaja Murugaboopathy and Gaurav Dogra, compared the projected operating cash flows and capital expenditures of the five companies most heavily involved in building AI infrastructure. The conclusion is simple: spending is growing significantly faster than the cash these companies generate from their core businesses.

Capex, spending on data centers, chips, servers and power infrastructure, has been growing at a pace Big Tech had never seen before over the past two years. These are inherently long-term investments that only pay off after years, yet they are increasingly eating into the cash that companies used to direct toward dividends, buybacks or acquisitions.

Which Player Is Most Exposed

Oracle faces the toughest situation. Over four years, the company has increased the share of capital expenditures in its operating cash flow from 47 to 174 percent, and its stock has lost 36 percent of its value this year. Oracle plans to raise $45-50 billion through debt and share issuance to fund further expansion of computing capacity for AI-related cloud contracts.

Microsoft generated $35.8 billion in operating cash flow last fiscal quarter against $37.5 billion in capital expenditures, while its AI revenue reached an annualized run rate of $37 billion. Amazon posted a 30 percent increase in operating cash flow to $148.5 billion on a trailing 12-month basis, but the company's free cash flow shrank to nearly zero, even as AWS grows at 28 percent annually.

Financing Instead of Their Own Cash

The key shift concerns how this buildout is being financed. For years, hyperscalers funded investments mainly from their own profits. Now they are increasingly turning to debt and new share issuances, and some are cutting back on buybacks to free up cash for data centers. Polish market analysts, commenting on the trend in Stockwatch and Brandsit, describe this as a shift from a self-funded model to a credit-funded one, in which the AI stock rally increasingly rests on debt issuance.

AI is pushing companies toward a hybrid model where software and cloud increasingly depend on enormous physical infrastructure - analyst quoted in the Reuters analysis

If financial benefits aren't evident within two to three years, investors will question whether the investment cycle has gone too far - market strategist quoted in the Reuters analysis

What It Means for Investors and Poland

For the Warsaw Stock Exchange, where bank stocks and AI-linked companies have been hitting records this year, this is a warning sign about the scale of valuation risk on Wall Street. If revenue from AI products doesn't start realistically covering rising capital expenditures in the coming years, pressure on the valuations of the biggest tech companies could spill over into global indexes and funds where Poles hold their retirement savings.

Analysts themselves aren't saying this is the end of the AI rally, but rather that the market is entering a phase where investors will start scrutinizing more closely the relationship between spending and actual revenue from AI-based products. According to the cited estimates, the point where the capital expenditure curve grows faster than the operating cash curve falls around the third quarter of 2026.

The next quarterly earnings reports from Microsoft, Alphabet, Amazon, Meta and Oracle in the coming months will show whether the growth rate of cloud and AI services revenue is starting to catch up with the growth rate of capital spending, or whether the gap between these two curves keeps widening.