Nvidia sells $25 billion of bonds for AI spending

Nvidia sold $25 billion of bonds to fund AI spending and refinance debt.

Nvidia tapped the investment-grade market with a $25 billion bond sale on Monday, and the order book swelled to about $85 billion. That is more than three times the size of the deal, which started near $20 billion before investors piled in.

Why Nvidia borrowed now

This deal is a sign of how aggressively big tech is funding the AI buildout. Nvidia is spending on its own priorities, but it is also backing the companies that will buy chips, build models, and expand the infrastructure behind those systems.

In the past year, Nvidia has moved well beyond selling GPUs. It took a $5 billion stake in Intel, said it would invest as much as $10 billion in Anthropic, and committed $30 billion to OpenAI’s huge funding round. That mix of investing and lending is a good way to keep the AI flywheel spinning.

The bond sale also helps Nvidia keep its balance sheet flexible. Proceeds will refinance existing debt and support other corporate uses, while the company keeps funding strategic bets without leaning too hard on cash reserves.

• The company sold notes in seven parts, with maturities from 2 years to 30 years.
• The longest-dated piece priced at 0.65 percentage point over Treasuries after tightening by 0.25 percentage point.
• Nvidia’s first bond sale since 2021 came after months of investor demand for AI-linked credit.
• The company is expected to generate more than $200 billion in free cash flow in the fiscal year ending Jan. 31, according to Bloomberg analyst estimates.

Investors wanted exposure to the AI trade

What made this deal unusual is how easily it was absorbed. Investment-grade buyers are chasing high-quality issuers with AI exposure, and Nvidia has both the brand and the cash flow to pull that off.

CreditSights analyst Andy Li put it bluntly: “I’m not surprised they would do a drive-by.” He added, “Nvidia has a dominant market and financial position, so they don’t have to market themselves hard to investors.”

That quote gets to the core of the deal. Nvidia did not need a big roadshow, a charm offensive, or a long explanation of why it matters. The company is already one of the main suppliers in the AI buildout, so bond buyers know exactly what they are buying into.

“I’m not surprised they would do a drive-by,” CreditSights analyst Andy Li said. “Nvidia has a dominant market and financial position, so they don’t have to market themselves hard to investors.”

The offering also landed in a friendly credit market. A recent agreement between the US and Iran helped push bond prices higher and risk measures lower, while US high-grade bond funds have seen inflows for 13 straight months, according to LSEG Lipper data.

That matters because cheap borrowing is a gift to a company planning years of capital-heavy AI spending. Nvidia can lock in long-dated money now, when buyers are eager and spreads are still tight.

How Nvidia compares with other AI borrowers

Nvidia is not alone. Alphabet and Amazon have also flooded debt markets with large offerings as they pour money into data centers and the power systems needed to run them.

The difference is that Nvidia is not primarily building the data centers itself. It sells the chips that power them, which means it gets exposure to the boom without taking on the same construction risk that comes with financing physical infrastructure.

• Nvidia: $25 billion bond sale, first since 2021, with orders around $85 billion.
• Alphabet and Amazon: also issuing huge amounts of debt to fund AI infrastructure buildouts.
• Market backdrop: $36 billion of investment-grade issuance on Monday across eight companies.
• Credit profile: Nvidia keeps an AA rating, which helps keep borrowing costs relatively low.

That AA rating is doing a lot of work here. Bloomberg Intelligence analyst Robert Schiffman said the long-dated sale could lower Nvidia’s average cost of capital and help fund strategic AI partnerships without weakening its credit profile.

For investors, the trade is simple: buy debt from one of the most profitable companies in AI, collect a tight spread, and avoid the messier risk of building physical capacity from scratch. For Nvidia, it is a cheap way to raise cash while keeping options open.

What this says about the AI money cycle

The bigger story is that AI is now financing itself in layers. Nvidia sells the picks and shovels, earns huge cash flow, invests in customers and partners, then raises debt to keep the machine moving. That is a very different model from the early days of tech, when companies often waited for profits before making giant bets.

This also explains why the debt market is paying such close attention to AI names. When a company with Nvidia’s balance sheet decides to borrow, it signals confidence in its own cash generation and confidence that AI spending will keep expanding.

One more detail matters: the company did this without the standard investor calls banks usually hold before investment-grade sales. That tells you Nvidia knew demand would be strong enough without the usual courtship.

My read is simple. If Nvidia keeps pairing huge cash flow with strategic investments and cheap long-term debt, other chipmakers will copy the playbook. The real question is whether AI demand keeps growing fast enough to justify all of this borrowing, or whether the market eventually starts asking for proof instead of projections.