The panic is loud right now. Analysts are warning about a potential AI infrastructure bubble worth trillions of dollars. Critics compare the spending to the dotcom crash. But Blackstone’s Jas Khaira just said what most people in finance won’t say out loud: the fears are overblown. And he’s pointing at something most retail investors are completely missing.
The Fear Looks Scary on Paper
Here is the setup. Major tech companies have committed to spending at a scale with no real historical comparison. According to Bloomberg, the five largest U.S. tech companies together planned to spend over $320 billion on AI infrastructure in 2025. That number has grown into 2026.
Most of this spending is financed with debt. Data centers, chips, power systems, and fiber all cost money upfront. The returns come later. When you borrow to build something that won’t generate cash flow for two or three years, traditional value investors get nervous. According to the Financial Times, bond investors started pricing in higher risk premiums on tech company debt in late 2025, signaling real concern in credit markets.
The concern isn’t crazy. History is full of examples where companies borrowed big to build for a future that took longer than expected to arrive. And AI has an extra layer of uncertainty: nobody knows exactly which use cases will generate the most revenue, or when.
Why Khaira Says the Math Actually Works
Blackstone manages over $1 trillion in assets. They are one of the biggest investors in data centers on the planet. When Khaira talks about AI infrastructure debt, he’s not guessing. He’s looking at actual lease agreements, actual power contracts, and actual revenue flowing into these facilities.
His argument is straightforward. The debt is secured by real assets. Data centers aren’t like software startups. They have physical value. They have long term contracts from hyperscalers like Microsoft Azure, Google Cloud, and AWS. According to JLL, data center vacancy rates in major U.S. markets sat below 3% in early 2026. That’s not a bubble. That’s a supply crunch.
The other part of his argument is timing. The “rich versus poor” split in this story is really about who understands the difference between short term cash burn and long term asset building. Average investors see the debt number and panic. Smart operators see the underlying contract structure and stay calm.
I’ve watched this pattern play out in real estate for decades. When someone builds a skyscraper on debt, people call it reckless. When the building fills up with 30 year leases, the same people call it visionary. AI infrastructure is following the same arc, just faster.
According to McKinsey, AI adoption in enterprise settings is projected to add $4.4 trillion in annual economic value globally. Even if that timeline slips, the demand for infrastructure that powers AI is not going away. The companies building data centers are building for a demand curve that only goes up.
The key difference between the dotcom crash and now: the dotcom boom was mostly software and websites with no physical assets and no paying customers. AI infrastructure is physical. It has customers. Those customers have multiyear commitments. That’s a fundamentally different risk profile.
If you’re looking at your own debt situation while watching these headlines, it’s worth knowing your options. A tool like SuperMoney loan comparison can help you see what rates different lenders are offering, especially if you’re thinking about refinancing or managing your own borrowing in this environment.
What This Means For You
Here is where most people get this wrong. They see a headline about AI debt fears and they either panic sell their tech holdings or ignore it completely. Both reactions are lazy.
Here is what I would do instead.
First, separate the infrastructure layer from the application layer. The fear about debt is mostly about companies building the physical hardware: chips, data centers, power. The application layer, companies actually selling AI software and services, has a different risk profile. Know which one you own.
Second, look at contract duration. Companies with long term contracts backing their AI infrastructure spending are in a very different position than companies speculating on future demand. Blackstone’s point is that their data center investments are mostly locked up with signed leases. That’s not speculation. That’s a business.
Third, watch your own financial health during this period. Markets will stay volatile as long as the AI investment thesis is being debated. Keeping your credit strong gives you options. If you’re not monitoring your credit regularly, IdentityIQ credit monitoring is a practical way to stay on top of your score and catch problems early before they cost you real money.
Fourth, don’t confuse fear with fact. The fear is real. The headlines are real. But the underlying business case for AI infrastructure, according to Blackstone and other major institutional investors, is also real. Panic selling into a media cycle is how average investors lose money that smart money captures on the way back up.
The Bottom Line
Jas Khaira isn’t asking you to ignore risk. He’s asking you to understand the difference between noise and fundamentals. The AI buildout is the biggest infrastructure investment in a generation. Some of the debt will go bad. Some companies will overspend and underdeliver. But the core demand driving all of it, real businesses needing real compute power, isn’t going anywhere. The people calling this a bubble are the same people who called cloud computing a fad in 2010. They were wrong then. They’re probably wrong now.
Frequently Asked Questions
What did Jas Khaira say about AI debt fears?
Khaira, a senior figure at Blackstone, argued that fears surrounding AI infrastructure debt are overblown. His case rests on the fact that most of Blackstone’s AI infrastructure investments are backed by long term contracts with major cloud providers, making them more like traditional real estate debt than speculative venture bets.
Is AI infrastructure spending really creating a debt bubble?
The concern is legitimate but the comparison to a bubble may be overstated. According to JLL, data center vacancy rates in key U.S. markets were below 3% in early 2026, meaning demand is outpacing supply. A bubble typically involves overbuilding into weak demand. That’s not what the data shows right now.
How does AI infrastructure debt affect regular investors?
It depends on what you own. Companies building physical AI infrastructure with long term signed contracts carry a different risk than software startups burning cash on AI development. If you hold tech ETFs or individual tech stocks, it’s worth understanding which part of the AI value chain you’re actually exposed to.
Should I be worried about AI debt fears affecting my portfolio?
Volatility is real. But long term, the demand for AI compute capacity is not in serious dispute. The debate is about timing and capital efficiency, not whether AI infrastructure will be used. Staying focused on contract backed, asset heavy players reduces your risk exposure compared to pure play AI software speculations.
