AI Borrowing Now Moves Bonds More Than War or Elections

The five biggest AI infrastructure spenders — Microsoft, Meta, Amazon, Alphabet and Oracle — have issued $271 billion in debt this year, a sum roughly the size of New Zealand's entire economy. That figure comes from Goldman Sachs credit analyst Amanda Lynam, in a Tuesday client note, and it now ranks as a bigger driver of the U.S. bond market than the Iran war, Middle East peace hopes or the midterm elections.

How did AI displace geopolitics as the bond market's main catalyst?

For much of 2026, Treasury yields moved higher alongside oil prices, especially when fears of an escalating Iran war intensified. Yields often eased when oil fell on hopes for a peace deal. That correlation has now broken down, as AI concerns have grown, capital has become more expensive and the war has drifted into its ninth month.

Since late October, sharp drops in tech stocks have sent investors flooding into the roughly $32 trillion U.S. Treasury market for safety. That demand has kept a lid on Treasury yields. It has not, however, translated into cheaper borrowing for AI's biggest spenders.

"The competition for capital has been so fierce that everyone is paying up to borrow," Matt Brill, head of North America investment-grade credit at Invesco, said Tuesday. The borrowing has become so large, he added, that many in the fixed-income community on Tuesday cared more about the latest batch of hyperscaler bond deals than the likely impact of the midterm elections. "To me, this is fueling the economy," Brill said. "It's key to follow where the dollars are coming from."

The plain takeaway: AI financing, not geopolitics, is now the force bond investors watch first.

Why are tech giants borrowing so heavily?

Most of the tech giants leading the AI buildout — Microsoft is the exception — have already burned through their enormous piles of free cash flow. The spending is expected to be the largest U.S. infrastructure project since the building of the railroads, leaving the sector dependent on debt markets.

The cost of that dependence is rising. Nearly every dollar the hyperscalers have borrowed since the start of the year now costs more to secure, Brill said. Most of the debt sitting in investor portfolios is underwater.

"Investors are a little bit frustrated," he said. "What seems to be a good opportunity hasn't been to date." The hope is that performance improves and recently higher yields adequately compensate for the risks investors are taking — a hope, not yet a result.

The concrete deals show the pressure. Meta was looking to raise about $15 billion from the U.S. portion of a cross-currency bond deal on Tuesday, according to Bloomberg. That deal is part of a trio of hyperscaler offerings in recent days, alongside bond sales from Amazon and Alphabet. Oracle and Meta have also outlined plans in recent days to issue both stocks and bonds, underscoring the need to relieve the AI funding strain on the investment-grade corporate bond market.

On days when hyperscalers bring large deals to market, Treasurys have often sold off, pushing yields higher — evidence of the "crowding out" of capital borrowers now face.

What are the yields telling us?

The pricing gap between the AI borrowers is wide. Oracle's 10-year bonds priced this week to yield about 8%, according to the source, while Microsoft's bonds issued late last week yielded around 5.5%. Meta's 10-year bonds were expected to price closer to 6.5%. Corporate bonds price at a spread over the risk-free 10-year Treasury yield.

Will Compernolle, macroeconomic strategist at FHN Financial, described the shift in a Monday client note: "AI's role in the bond market is becoming close to how geopolitical headlines affected yields earlier this year — unscheduled, sometimes closer to rumor than fact, and the catalyst for extreme narratives that shift from one week to the next."

Is the AI trade still priced for perfection?

Oliver Shale, U.S. investment specialist at Ruffer, said the AI debate has moved from "funding ambition at any cost" to expecting or requiring returns on that investment. Earnings remain robust and committed capital spending continues, he noted. But "exceedingly high expectations" are baked into market prices, leaving only a "small margin of safety" for everything to turn out as hoped.

The market's Tuesday session captured the tension. Tech stocks closed higher after a brutal couple of weeks, while the benchmark 10-year Treasury yield rose 5.8 basis points to 4.951% — its biggest daily climb since Sept. 28, according to Dow Jones Market Data.

The question ahead: whether higher yields compensate investors for AI risk, or whether the crowded market simply keeps repricing it.