Amazon Just Borrowed $31 Billion in 48 Hours, and It Tells You the AI Boom Has Changed

A data center corridor representing AI infrastructure debt

Amazon generates tens of billions of dollars in cash every quarter. It is one of the richest companies that has ever existed. And this week it borrowed $31.5 billion in 48 hours to keep funding its AI buildout.

Stop and think about that. The question is not whether Amazon can afford AI. It is why a company this flush suddenly needs to borrow on this scale. The answer reveals something structural about where the entire AI boom has quietly arrived.

Here is what happened, and why it is bigger than one loan.

$31.5 billion in two days

Amazon secured a $17.5 billion senior unsecured term loan, disclosed in an SEC filing, with Citigroup as lead arranger and a syndicate that includes JPMorgan, Bank of America, HSBC and Wells Fargo, TechCrunch reported. The loan is priced just above SOFR and can be drawn through September.

That came just two days after Amazon sold roughly $10 billion of Canadian-dollar bonds, the largest Canadian corporate bond deal on record, according to TheNextWeb. Add it up and Amazon stacked about $31.5 billion of fresh debt in under 48 hours. And it has been issuing bonds in euros, dollars and Swiss francs since March.

The number that explains it

Why does a cash machine need to borrow? Because of one stat that reframes everything. Hyperscaler capital spending is now consuming nearly 100 percent of operating cash flow in 2026, versus a 10-year historical average of about 40 percent, per Morgan Stanley.

In plain English, AI data centers are now eating essentially all the cash these companies produce. There is nothing left over. So even a giant like Amazon, planning around $200 billion in capital spending this year and having already spent $43.2 billion in a single quarter, has to turn to debt markets to keep building. The piggy bank is not empty, it is just fully committed.

Amazon’s debt pile just jumped 50 percent

The scale of borrowing is showing up on the balance sheet. Amazon’s total short- and long-term borrowing, including leases, topped $225 billion as of the end of March, up about 50 percent from roughly $150 billion a year earlier, TheNextWeb noted.

Some of this money is also funding AI equity stakes. Amazon has committed up to $50 billion to OpenAI and holds a multibillion-dollar position in Anthropic, per Bloomberg via Yahoo Finance. So the borrowed cash is going into both the data centers and the AI labs that will fill them.

This is an industry-wide shift, not an Amazon quirk

Amazon is not alone, and that is the real story. Morgan Stanley projects nearly $570 billion in AI-related debt issuance globally in 2026, with nearly $236 billion already issued through May, four times the pace of the same period in 2025, the same report found.

AI-linked debt has now surpassed US banks to become the largest single segment of the investment-grade bond market. Let that land. The biggest slice of the safest corner of corporate debt is now AI. Alphabet announced an $85 billion raise and sold a rare 100-year bond. Meta announced its largest-ever $30 billion offering. The whole industry has shifted from paying cash to borrowing.

The question hanging over all of it

TechCrunch put the central worry bluntly: the open question is “whether the returns will ever justify it.” That is the whole game. These companies are borrowing half a trillion dollars a year to build AI capacity, and the revenue that pays it back has not fully shown up yet.

If the AI revenue arrives, this is the railroad buildout of the century, and the debt will look like genius. If it does not, the industry has just wired a credit-market problem into an AI problem, and the unwind would not be contained to tech. That is the bet being placed, $570 billion at a time.

Why This Matters

When the richest companies on earth start borrowing to fund their core strategy, the financing has become the story. AI is no longer something Big Tech pays for with spare cash. It is something they are leveraging up to chase, which raises the stakes for everyone if the math does not work.

It also means AI risk and credit risk are now tangled together. Half a trillion dollars of AI debt sitting in the investment-grade market means that if AI returns disappoint, the pain shows up in bond portfolios, pension funds and the broader economy, not just in tech stocks. The boom has roots in places most people never think about.

The NewsSparq Takeaway

Three things to hold onto.

One, even Amazon has to borrow now. A company that prints cash stacked $31.5 billion of debt in two days. That is not weakness, it is a signal that AI capex has outgrown what even the giants can self-fund.

Two, this is structural, not a one-off. $570 billion in projected AI debt for 2026, now the biggest chunk of the investment-grade market. The whole industry switched from cash to credit at once.

Three, the returns question is everything. Borrowing this much only works if the AI revenue eventually justifies it. Until it does, every one of these deals is a bet that the boring layers of the AI economy will pay off.

Amazon borrowing $31.5 billion in 48 hours is not a sign of trouble at Amazon. It is a sign of how the whole game changed. The AI buildout got so big that the richest companies in the world ran out of pocket and turned to the bond market. Now the only question that matters is whether the revenue ever catches up to the debt.

Sources: TechCrunch, TheNextWeb, Yahoo Finance, TechTimes.

By The NewsSparq Editorial Desk