J.P. Morgan's OpenAI loan is strange

Article’s Math and Financial Framing Critiqued

• Multiple commenters say the expected value (EV) examples are mis-specified: the “$900 EV” example mixes “above cost” and “total return” framing, and the bankruptcy case unrealistically assumes 0% recovery for secured debt.
• People note the piece confuses equity risk with debt risk, misuses bond spreads, and ignores recovery rates (several mention ~40% is a common baseline in credit).
• The assumed 90% bankruptcy probability is seen as unjustified; treating OpenAI as a random early-stage startup is called “silly.”

Nature of the JPM Facility

• Many emphasize this is a revolving credit facility, not a simple term loan; it may never be fully drawn and often serves as short-term liquidity and signaling.
• Revolvers are usually senior, heavily covenanted, and about relationship-building: banks use them as break-even or loss leaders to win future IPO, M&A, and bond mandates.
• Several argue the core “upside” for JPM is not 5% interest but the chance to lead a huge IPO or future transactions and collect massive fees.

Collateral and IP Value

• Debate over whether the loan is primarily secured by OpenAI’s IP and hardware versus its going-concern prospects.
• Some claim OpenAI’s IP would be worth little in an insolvency scenario if competition or open source surpasses it; others argue its models, brand, user base, and leases/datacenters would still be highly valuable, especially to large tech buyers.
• Microsoft is widely seen as an implicit backstop, though commenters note its rights are time-limited and it could walk away in a true collapse.

OpenAI Revenues, Losses, and Profitability

• The article is criticized for using outdated Reuters revenue/loss figures; newer reporting cited in the thread suggests much higher current revenue and large but lower relative burn.
• There is sharp disagreement over profitability: some insist there’s “no evidence” OpenAI is profitable and that capex/R&D spending far exceeds revenue; others argue inference is probably profitable and losses reflect aggressive investment, not an unworkable model.
• A more detailed critic questions the viability of rumored trillion-dollar capex, noting required ARPU would vastly exceed Meta/Google levels. Supporters respond that the trillion is a strategic ceiling to scare off competitors, not a firm plan.

Risk, Systemic Concerns, and Macro Context

• Some see this as “mixing the AI bubble with the financial system,” but others argue AI is far more broadly useful than crypto and therefore a safer basis for credit expansion.
• A few raise “China risk” and the possibility that geopolitical moves (similar to the TikTok case) could disrupt the long-loss-then-IPO playbook for AI firms.

Overall View of the Loan

• Many commenters conclude the facility is neither strange nor especially risky for JPM given: senior secured structure, likely nonzero recovery in default, OpenAI’s scale and growth, and the massive optionality on future advisory business.
• The consensus in the thread is that the article substantially misunderstands both modern venture lending practice and large-bank relationship strategy.