Altman’s Wealth, OpenAI’s IPO Risks | Generated by AI

Home 2026.06

Question: Give me more detail on the WSJ/The Information exposé about Sam Altman’s personal investments and OpenAI’s IPO financial risks.

Answer:

This is a two-part story that broke simultaneously on June 24, 2026 — and both parts are explosive.

Part 1: The $665 Billion Off-Balance-Sheet Bomb

OpenAI’s books show zero debt and just $46M of quarterly capital spending. The catch: around $665B of commitments sitting just off the balance sheet — primarily long-term deals to rent the data centers and chips its models run on.

OpenAI spent about $34B last year and burned through $3.7B in the first three months of 2026 alone. Zero debt means little when $665B of commitments sit just out of frame.

The accounting trick: OpenAI primarily rents compute rather than buying it, so those payments flow through the P&L as operating expenses, not as debt on the balance sheet. It looks lean. It isn’t.

OpenAI is on the hook for roughly $600B in compute commitments over the next five years, with some infrastructure deals stretching to $1.4 trillion over eight years. An HSBC analyst estimate pegs the capital gap at $207B through 2030 — meaning OpenAI needs to raise that much just to honor existing commitments. The IPO is essentially a forced funding event, not a victory lap.

The Q1 2026 financials are brutal: the company is losing $1.22 for every $1 of revenue.

About 72% of OpenAI’s cost of revenue — the compute needed to run AI models — flows to related parties like Microsoft.

Microsoft receives 20% of OpenAI’s revenue as part of a revenue-share agreement, on top of the Azure compute spend.

The circular money flows are stunning. OpenAI burns Microsoft’s cash on Azure compute. That Azure consumption shows up as revenue inside Microsoft’s AI business line. Then Microsoft’s equity stake in OpenAI gets marked up to reflect the latest funding valuation, which flows to “other income.” Investor = supplier = customer, all at once.

Part 3: Altman’s Personal Wealth Machine

The core paradox: Altman holds 0% equity in OpenAI, yet controls a portfolio of $2B+ in companies that do business with OpenAI.

Cerebras

Altman started investing in Cerebras back in 2017. Court documents showed he held 89,373 shares. The stake was valued at $3.2M at end of 2025. Then: OpenAI signed a multi-year agreement worth more than $10B with Cerebras that included a $1B loan and warrants. Then Cerebras IPO’d. Cerebras shares soared 68% on their Nasdaq debut, pushing Altman’s stake to ~$27.8M.

Cerebras is handing OpenAI warrants worth up to 10% of the company — around $5B at IPO midpoint — roughly half the gross profit it stands to make on the deal. Same playbook OpenAI ran with AMD.

Helion

Altman confirmed he owns roughly one-third of Helion, a fusion startup, valued at approximately $1.65B as of late 2025. He personally invested $375M in a $500M round in 2021, then proposed OpenAI participate in a major Helion funding round. He says he recused himself from “the decision to proceed and the final approval of terms.”

The Chinese article mentions Thrive Capital (a major OpenAI investor) poured fresh capital into Helion in June 2026, pushing its valuation to $15.5B — making Altman’s personal stake worth $4.1B+. A fusion company that has never generated a single commercial kilowatt-hour.

The Pattern

Critics argue that when OpenAI signs partnerships or spends money with vendors linked to Altman’s investments, he benefits personally even if he takes no salary or equity from OpenAI.

Part 4: Congressional Blowback at IPO Timing

State AGs from several Republican-led states wrote that Altman “has a history of self-dealing and serious conflicts of interest” and that his “personal financial interests have only limited alignment with OpenAI’s financial performance” — because he holds no direct equity.

When Altman was briefly fired in 2023, the board had fretted over his personal investments. A newly formed board pledged to fix the problem, but it never went away. As OpenAI hurtles toward an IPO valued at around $850B, the question persists — how to determine whether decisions serve the company’s interest or Altman’s.

TL;DR: The Structural Play

Altman’s compensation design is almost elegant in its audacity:

This is the governance time bomb heading into the roadshow.

References:

Back Donate