The AI arms race is heating up, and its effects are starting to show up where investors actually care – on the Big Tech balance sheets.

According to UBS, global AI capex is now expected to hit $423 billion in 2025, and $571 billion in 2026. By 2030, UBS expects spending to reach $1.3 trillion, implying a 25% compound annual growth rate.

Those numbers aren't abstract anymore. They're turning into debt issuance, lease obligations, and quietly rising financial risk. The era of funding AI infrastructure with excess cash is over.

Cash Flow Leaders

Alphabet is in the cleanest position by far. As of its most recent earnings, the company held nearly $100 billion in cash and marketable securities against just over $20 billion in debt. Furthermore, it generated around $48 billion in operating cash flow in a single quarter.

Capex remains well below operating cash flow, even with heavy AI investment. In other words, Alphabet can fund AI growth internally without distorting its balance sheet. That flexibility matters in a world where compute demand is accelerating faster than revenue clarity.

Microsoft sits just behind Alphabet in balance-sheet strength. It has a net cash position, strong free cash flow, and diversified revenue streams across cloud, software, and enterprise services. AI capex is rising, but it's still comfortably covered by cash generation. Microsoft is spending aggressively, but not desperately.

Balance Sheet Engineers

Legally sound, yes. Economically clean? Not really. Investors treated the bonds like Meta's anyway, thanks to backstop guarantees. Days later, Meta issued another $30 billion in traditional bonds. The firm is spending big, although it doesn't want the leverage to look big.

Then, there is Oracle, whose stock lost nearly 50% since the peak of AI optimism.

The credit market is flashing warning signs on the firm as its five-year credit default swap recently surged to 139 basis points. It is one of the highest readings since 2008, and a sign that the market is highly concerned about the firm's leverage.

Oracle is betting heavily on AI infrastructure, but unlike Alphabet or Microsoft, it doesn't have the cash cushion to absorb mistakes.

Oracle is effectively making decades-long bets on demand visibility that doesn't yet exist. On top of that, the company is expected to spend roughly $300 billion over five years on GPUs and related equipment. Capex is set to more than double to $50 billion this fiscal year, roughly three-quarters of projected revenue.

This is where the AI divide becomes obvious. Companies with massive free cash flow can treat AI capex as an option. Companies relying on debt are turning it into a wager, and the market is rightfully pricing that in.

Read Next:

• Nvidia Vs. Alphabet: Which Could Be The World’s Biggest Company By End-2026?

Image: Shutterstock