Microsoft has priced a $25 billion bond offering, its first return to the US corporate bond market in more than nine years. The deal, priced on October 29, 2026, is one of the clearest signs yet of how expensive the AI race has become — and how willing investors still are to fund it.
The offering is large by any measure, but the details matter. Microsoft has not said the whole $25 billion will go to AI, and the market reaction says as much about confidence in Microsoft as it does about AI itself.
The deal in numbers
The sale consists of $25 billion in senior unsecured notes, split into 10 tranches. Two floating-rate tranches are $1 billion each, due in 2028 and 2029. The eight fixed-rate tranches carry coupons from 4.85% to 6.35%, with maturities stretching from the early 2030s all the way to 2066. The largest fixed-rate tranches are $4 billion due in 2036 at 5.55% and $4 billion due in 2056 at 6.25%.
Microsoft expects the offering to close on November 4, 2026, subject to customary conditions. Joint book-running managers include JPMorgan, UBS and Wells Fargo, according to press reports. The notes are rated Aaa by Moody's and AAA by S&P, the top tier — a rating only a handful of companies hold.
Investor demand was striking. Reports put orders at more than $100 billion, over four times the size of the offering. That oversubscription is a big reason Microsoft could borrow at relatively favourable rates even for debt that will not be repaid until 2066.
Why now?
The timing follows Microsoft's fiscal first-quarter 2027 earnings, reported on October 28. Microsoft reported revenue of $92.28 billion for the quarter, up 19% from a year earlier, and said it spent more than $45.2 billion on capital expenditures including finance leases in the quarter alone.
On the earnings call, chief financial officer Amy Hood signalled that Microsoft would seek additional funding sources to complement its cash generation when market conditions were attractive. Press reports put Microsoft's calendar-2026 capital expenditure target at around $170 billion, covering data centres, AI chips and computing capacity for customers including OpenAI and Anthropic.
In other words, Microsoft is not borrowing because cash is short. It is borrowing because building AI infrastructure at this scale costs more than even a company of its size wants to fund from cash flow alone — and because, with a AAA rating, it can borrow cheaply.
What the money is — and isn't — for
This is the part most headlines skip. Microsoft's own documents say the net proceeds will be used for "general corporate purposes," which may include working capital, capital expenditures, repurchases of capital stock, acquisitions, and repayment of existing debt.
That means the $25 billion is not earmarked solely for AI data centres. Capital spending is one possible use among several. Press coverage has tied the deal to the AI buildout, and that link is reasonable given Microsoft's spending plans, but it is not what Microsoft itself promised in the offering documents.
Readers should keep that distinction in mind. A bond sale "to fund AI" and a bond sale that "may fund AI spending, among other things" are different claims, and only the second is what Microsoft actually said.
What it signals for the AI race
Even with that caveat, the deal fits a wider pattern. Big Tech companies are raising enormous sums to pay for AI infrastructure, and investors are, for now, still willing to lend.
Meta's reported $95 billion AI funding blitz, covered in our earlier story, shows the same pressure from a different angle: the companies leading AI feel they cannot afford to pause building. At the same time, Anthropic's record-breaking IPO filing shows that even the model makers themselves are turning to public markets for capital at unprecedented scale.
Microsoft's move also sits alongside a more human story inside the same company. Just days earlier, Microsoft offered buyouts to 60,000 employees a reminder that the AI spending race is reshaping workforces even as it reshapes balance sheets. Growth in infrastructure spending and pressure on headcount can happen at the same time.
And Microsoft is not acting in isolation on AI policy either. It was one of the companies that recently agreed on one training opt-out for publishers, a shared signal that lets websites say no to AI training in one place instead of company by company.
What to watch next
Three things will tell us whether this bet pays off. First, the November 4 close itself — a routine step, but the final confirmation that the deal completed as priced. Second, Microsoft's next capital spending update: does spending stay near the reported $170 billion target, or does it climb again? Third, whether other AAA-rated tech companies follow with their own large bond sales, or whether Microsoft's deal marks a peak in investor appetite.
For now, the market verdict is clear: investors still believe Microsoft can carry more debt to build AI infrastructure, and they were willing to lend four times what Microsoft asked for. Whether that confidence survives the next earnings cycle is the real test.
Do you think borrowing $25 billion is a smart bet on AI's future, or a warning that the buildout is getting too expensive? Share your take in the comments below.


Comments
Post a Comment