Blackstone President Gray: Betting on AI Infrastructure Still Makes Sense Even as Capital Gets More Expensive

Blackstone President Jon Gray, who runs one of the largest investors in data centers, told Bloomberg that he remains an optimist on the AI buildout even though capital is getting more expensive. He gave the interview to Yvonne Man on 13 October 2026 on the sidelines of the Bloomberg New Economy Forum in New Delhi. Gray's main argument: in the telecom boom, supply was built ahead of demand and with enormous leverage, while the current cycle, in his words, "looks and feels very different." AI infrastructure, he sums up, is "the right place to be."

What exactly Gray said

Bloomberg published only a short excerpt, with no figures on capex, interest rates or yields. It suggests that Gray acknowledges the pressure on the cost of capital but does not revise his overall view of the industry. Rising rates usually hit infrastructure projects hardest: they require large upfront investment and pay back over years. Gray, however, argues that the foundation of this cycle is different.

Why he brings up telecom

The comparison is not accidental. In the late 1990s and early 2000s, carriers laid fiber for traffic that never materialized in the volumes expected. A significant share of the projects was financed with debt. When demand fell short, companies buried in loans could not service their obligations. Bankruptcies from that period included WorldCom and Global Crossing.

Gray singles out two features of that episode: capacity was built before demand, and the buildout rested on heavy leverage. His thesis is that at least one of these factors, and by the logic of his phrasing both, looks different today. This is the assessment of a market participant, not verified statistics. The share of capacity already pre-contracted and the debt structure of individual projects could confirm it, but neither appears in the excerpt.

Blackstone has a stake in the outcome

Gray is not speaking from the sidelines. Blackstone has steadily expanded its presence in data centers:

• the purchase of operator QTS in 2021 for about $10 billion including debt;
• participation in the 2024 deal for Australian operator AirTrunk, valued at about 24 billion Australian dollars;
• plans announced in 2025 to invest $25 billion in projects in Pennsylvania, including AI infrastructure and energy.

His words are therefore best read as the position of a major asset holder with an interest in keeping market confidence intact. But such an investor also has access to data on utilization and contracts that outside analysts lack.

Key takeaways

• Source: Jon Gray, president of Blackstone, in an interview with Bloomberg.
• Venue: Bloomberg New Economy Forum, New Delhi, 13 October 2026.
• Position: optimism on the AI buildout despite the rising cost of capital.
• Argument: in telecom, supply outran demand and rested on heavy leverage; now it is different.
• Missing from the piece: figures on capital investment, interest rates, utilization and debt load.

What to watch next

For investors and operators, the question comes down to measurable indicators. First: what share of new capacity is already contracted before it comes online. Second: how many projects are financed with debt, and on what refinancing terms. Third: the availability of power and grid connections, which often constrains timelines more than money does.

Startups and corporate customers should keep in mind that expensive capital eventually shows up in the price of renting compute. If Gray's assessment is right, the market will absorb the higher costs without widespread disruption. If it is wrong, the first to suffer will be projects with high leverage and no long-term tenants, which are exactly the ones he points to as the weak link of the previous cycle.