Jon Gray on Private Credit, AI Infrastructure & an All-Weather Firm | Blackstone Q1 2026 Results

Key points

• **Q1 distributable earnings +25%, ~$70B raised**, including 'one of our best quarters ever for private credit institutional fundraising.' Headline read: institutional LPs are leaning *into* private credit, not away from it, despite the redemption-noise in retail-facing semi-liquid vehicles.
• **Portfolio-company LLM spend up 15x YoY.** Off a small base but the gradient is the point. Eight of Blackstone's ten best-performing Q1 investments were in **data centres, LNG, battery storage**. Confirms the Lemkin/Sacks/Chamath thesis from the operator-allocator side: real money is moving into the AI-infrastructure layer, not into the model-layer or the application-SaaS layer.
• **The supply-side bottleneck list, from a buyer of these assets:** 'compute is the issue — shortages of turbines for power, power plants getting data centres entitled, chips, all of this.' Identical to PTJ's macro framing and Reiner Pope's technical framing. Three independent voices in this issue identify the same physical-infrastructure constraint.
• **Sora was shut down because of compute scarcity** (per Gray) — used as evidence that demand outstrips supply. Direct rebuttal of the DeepSeek-driven 'are we overinvesting' narrative from a year ago.
• **On private credit semi-liquid critiques (Lloyd Blankfein's 'illiquid funds' jab):** Gray defends the structure. Two conditions for legitimacy: (1) disclosure quality (B-CRED prospectus has six bold liquidity-warning bullets on page one); (2) actual return premium delivered. B-CRED has a ~60% premium over leveraged loan market over 9.5 years. 300,000+ semi-liquid customers; 'we are not hearing complaints that people didn't understand the products.' Useful baseline for the credit-cycle predictions in the Issue 02 ledger.
• **Mark-defense on the loan book:** B-CRED 700 loans listed every quarter w/ purchase mark and current mark. Today's marks at ~96, *below* the leveraged loan market. 'When people are selling securities, it doesn't always reflect a lack of confidence in marks.' Notes $12.5B of redemptions in leveraged-loan and HY funds in March driven by macro nerves not credit issues.
• **Real-estate mark validation:** 'we sold $38 billion of properties since interest rates started going up at a premium to where these things were marked.' The biggest AUM-credibility statement in the call — the marks held up against actual transaction prints, not just modelled NAV.
• **'Don't bet against super-intelligence' but expect frustrations on timing.** Gray on AI productivity diffusion: 'real strides at some companies, but not showing up in numbers in a big way because it's pretty early.' Ancestry digitising records, accounting firm using AI for tax/audit, Blackstone itself using LLMs to summarise historical investment-committee memos to evaluate live deals. Direct contrast with AppLovin's 80-90% AI-coded throughput — the diffusion gradient between operator-led tech firms and traditional services firms is enormous.

Notable quotes

Themes

• Private credit institutional flows continue despite retail redemption noise
• Allocator validation of compute / power / chip supply bottleneck
• Mark integrity defended via realised transaction prices
• AI infrastructure (data centre, LNG, battery storage) as Q1 alpha source
• Productivity-diffusion gradient between operator-led and traditional firms