Summary: Alphabet upsized a seven-part dollar bond sale to $20 billion on February 9, 2026 after an order book exceeded $100 billion, helping fund a planned $185 billion of 2026 AI investment. The filings also introduced AI-related risks for bond investors, making credit-market appetite a constraint on infrastructure expansion.
On February 9, 2026, Alphabet headed to the dollar bond market with a seven-part offering, raising money for the largest annual investment program it has ever undertaken. According to Reuters, the deal launched at roughly $15 billion, drew orders of more than $100 billion, and was ultimately upsized to $20 billion — the biggest dollar bond sale in the company’s history. Behind the raise sits a single purpose: the $185 billion Alphabet has lined up for AI investment in 2026.
The real story is not “Google’s parent borrows money.” It is that the funding model behind AI infrastructure is shifting. Once a year’s capital spending outgrows what free cash flow can absorb, the bond market steps in as a fresh referee for the AI arms race — and how fast data centers get built now depends on interest rates, credit spreads, and investor appetite.
Anatomy of the Deal: Seven Tranches, Sevenfold Oversubscription
The offering spanned seven dollar tranches covering a range of maturities. Bloomberg had Alphabet targeting about $15 billion at the outset, but demand ran well past plan: per Reuters’ figures, the order book topped $100 billion — roughly seven times oversubscribed. Alphabet lifted the deal to $20 billion, setting a fresh high-water mark for its own dollar issuance.
The maturity mix is the detail worth pausing on. CNBC and Bloomberg both flagged a 100-year component in the lineup. Century paper is rare in the corporate market and generally the preserve of only the strongest credits — and anyone willing to lock up money for a hundred years is casting a vote, in real dollars, that these AI investments will still be generating returns on a very long horizon.
Where the Money Goes: A $185 Billion Year
The proceeds plug straight into Alphabet’s 2026 capital program: $185 billion of AI investment for the year, spread across data centers, custom silicon, and cloud capacity expansion. Set against the industry backdrop, the number grows larger still — Reuters reports Alphabet and Amazon at the head of an AI-infrastructure borrowing spree, with hyperscaler infrastructure investment forecast to top $630 billion in 2026.
It is the acceleration of a trajectory we flagged in our opening-of-year outlook: AI capex moving from cash-funded to leverage-funded. The follow-through confirms the direction — in the first half of 2026 alone, Alphabet sold more than $50 billion of additional debt, including a separate offering of up to $25 billion.
New AI Risk Language in the Debt Filings
CNBC’s same-day coverage surfaced a detail that is easy to skim past: in the documents for this entry into the debt market, Alphabet disclosed new AI-related risks. For a company used to talking growth on earnings calls, the disclosure was written for bond investors rather than shareholders — a crowd focused on downside protection, cash flow coverage, and covenant terms.
The move marks the first time AI investment has been formally placed inside the credit market’s pricing framework. Equity investors can pay a premium for uncertain-but-huge upside; bond investors have to underwrite the worst case. As debt takes on a growing share of hyperscaler AI capex, how well these disclosures are written will feed directly into financing costs — and thereby set the ceiling on how fast the buildout can run.
The Wider Hyperscaler Borrowing Wave
Pull the camera back and Alphabet’s deal is one tile in a larger mosaic. Reuters names Alphabet and Amazon as the leaders of the AI-infrastructure borrowing wave, set against more than $630 billion in projected 2026 hyperscaler infrastructure investment — a scale no single company can absorb out of operating cash flow alone.
Borrowing spreads the enormous outlay over years of future revenue, freeing construction pace from the constraints of any single year’s income statement. The cost is a fixed interest burden plus exposure to the credit cycle: if AI demand grows more slowly than hoped, the debt does not shrink with it. Which is why this dry-looking bond story is worth tracking for every team that leans on cloud capacity and model APIs.
What Developers Should Take From It
Three practical consequences. First, capacity supply: the $20 billion turns into data centers and accelerator capacity coming online through 2026 and 2027, directly easing GPU scarcity. Second, pricing pressure: interest costs stacked on top of capex recovery will work their way into cloud and model API pricing — expect the era of bills that only climb to become more visible. Third, a leading indicator: order books, spreads, and risk disclosures in the bond market are a more honest gauge of AI buildout health than earnings calls; a cold reception for the next offering would be the cooling signal for the entire capex cycle.