Alphabet issues $20b bonds and plans 100‑year debt to fund AI buildout

Alphabet’s decision to raise US$20 billion in one shot and line up century‑long debt explicitly to fund AI infrastructure is one of the clearest signals yet that hyperscalers view this as an arms race measured in trillions, not billions. A planned US$185 billion in 2026 capex, largely for data centers and AI compute, effectively locks in a multi‑year ramp in training and inference capacity for the Gemini line and beyond. When you add similar guidance from Amazon, Meta, Microsoft and Oracle, you get a picture of an ecosystem willing to financialize AI bets over decades via ultra‑long bonds.

For the race to AGI, this matters less as a one‑off financing event and more as a structural commitment: debt investors are now co‑funding the frontier model race, and they appear comfortable with that risk profile. That lowers the marginal cost of capital for additional model training runs and agentic platforms, even if near‑term revenues lag. It also hardens the barrier to entry for smaller labs that cannot tap global bond markets. In practice, this deal pulls more of the future compute landscape into the hands of a few US‑based giants and shrinks the room for slower, more cautious development approaches.