Pimco Warns AI Data Center Debt Is Splitting Fast

A quiet corner of global credit markets is suddenly carrying a very loud AI trade.

Money is rushing into data centers, the giant buildings that keep artificial intelligence running. But Pimco is warning investors that this debt market has already split into clear winners and weaker bets.

For Indian investors, this matters more than it first appears. Global credit stress can move foreign flows, tech stocks, and risk appetite back home.

Data center debt gets crowded

David Forgash, Pimco’s leveraged finance chief, said investors need caution in high-yield debt linked to data centers.

High-yield debt is a polite market term for riskier borrowing. Investors also call it junk debt. Companies pay higher interest because lenders see more danger.

Forgash said this part of the market barely existed a year ago. It now makes up about 4 percent of the high-yield market. He expects it could reach 10 percent within two years.

That is a very fast jump. In credit markets, speed often matters as much as size. When too much money chases one idea, weak deals also find buyers.

The logic behind the rush is simple. AI needs huge computing power. That means more servers, more power, more cooling, and more buildings.

The stress is already visible

Forgash said around 75 percent of this data center debt trades at spreads near 6 percent. The remaining portion trades at 10 percent or more.

A spread is the extra return investors demand over safer government debt. So, a wider spread means lenders want more compensation for risk.

That gap tells us something useful. The market is not treating all data center borrowers equally anymore.

Some borrowers look safe because they have strong tenants, long contracts, and visible cash flows. Others already face doubts over whether future revenue will match today’s excitement.

Bloomberg Intelligence estimates US junk debt tied to data centers has reached about $40 billion in face value. Most of it came in the past year.

Globally, hyperscalers have borrowed more than $250 billion for AI infrastructure. Hyperscalers are the large cloud and tech companies that run massive computing networks.

For ordinary savers, this may sound far away. But the chain is familiar. Cheap money creates a hot theme. Credit expands quickly. Then investors start separating quality from story.

What Pimco wants to avoid

Forgash said Pimco prefers deals backed by investment-grade companies. These are stronger firms with better credit ratings.

He also pointed to firm offtake agreements. In simple terms, these are contracts where a customer agrees to buy capacity from a data center.

Pimco wants contracts that customers cannot easily cancel. It also wants debt repaid before those contracts expire.

That sounds boring, but boring is often valuable in credit. A lender wants cash flows first, exciting narratives later.

The key issue is residual value. If a data center contract ends, the building should still be worth something useful.

That question will become sharper as AI hardware changes fast. A facility built for today’s chips may need expensive upgrades tomorrow.

This is where retail investors often miss the catch. They see “AI infrastructure” and assume all assets benefit equally. Credit investors ask a colder question. Who gets paid, and when?

Software debt looks different

Forgash also flagged another AI pressure point, software companies.

AI can disrupt software and the services attached to it. That could hurt firms built around older ways of selling software support.

Still, he said the risk looks more contained in high-yield bonds. Software firms represent about 15 percent of leveraged loans, but only 3 percent of high-yield bonds.

Leveraged loans are also risky corporate loans, often issued by indebted companies. They usually sit with institutional investors and funds.

Pimco has kept low exposure to software in actively managed loan products, Forgash said. He added that those products returned more than 6 percent through selection among roughly 1,200 loan issuers.

For Indian investors, the lesson is not to avoid technology. The lesson is to respect balance sheets.

A listed company may tell a strong AI story. But debt markets often reveal nervousness before equity markets admit it.

Why India should watch this

India’s own data center boom has real strength behind it. Cheap mobile data, digital payments, cloud adoption, and AI demand all support growth.

But capital-intensive businesses need patience. Data centers require land, power, cooling equipment, fibre links, and long payback periods.

That means interest rates matter. If global borrowing costs stay high, weaker projects feel pressure first.

This is why Forgash’s warning deserves attention in Mumbai and Bengaluru too. It is not only a Wall Street credit story.

When global investors grow cautious, they reduce risk across markets. That can hit tech valuations, startup funding, and foreign flows into emerging markets.

A young professional with a mutual fund portfolio may not own US junk bonds directly. Still, global risk-off days can pull down Indian equities.

A founder planning cloud-heavy AI tools may also feel the effect. Higher financing costs eventually show up in cloud pricing, funding terms, or slower infrastructure rollout.

Forgash said Pimco still sees chances in homebuilders and building suppliers. These sectors have sold off, even though they may gain from eventual rate cuts.

He also said the firm remains overweight energy through year-end. That is notable because AI data centers consume huge power.

The hidden debate is becoming clear. Investors once asked who would win from AI. Now they ask who can fund it without breaking their balance sheet.

That shift matters. The AI story has moved from demos and headlines into debt schedules and electricity bills.

For Indian readers, the sensible takeaway is simple. AI may remain a powerful theme, but not every AI-linked asset deserves blind faith. The next phase will reward companies with contracts, cash flow, and discipline. The rest may discover that even the hottest technology cannot make expensive debt disappear.