OpenAI in talks with Citi, JPMorgan for potential IPO underwriting - Bloomberg

OpenAI is expected to confidentially file for an IPO within weeks, with the offering potentially coming later this year and additional banks, including Citigroup and JPMorgan, in talks to join Goldman Sachs and Morgan Stanley. The listing would follow SpaceX's expected $75 billion IPO and could precede a jumbo Anthropic offering later this year or early next. The story is positive for AI capital markets activity, but the bank lineup is still not finalized.

Analysis

The more important signal is not the IPO itself, but the implied reopening of elite fee pools for U.S. investment banks. If the underwriting slate expands beyond the first two incumbents, the marginal economics favor firms with the strongest distribution and balance-sheet adjacency; that should keep capital markets revenue expectations supported into year-end, even if the actual listing slips. For C and JPM, the optionality is asymmetric: they can win economics without needing to dominate the mandate, while the larger strategic benefit is relationship capture with a category-defining AI issuer and its ecosystem.

Second-order, a high-profile OpenAI process validates private-market mark-to-market optimism across late-stage AI, but it also creates a near-term competition for investor dollars. A jumbo AI IPO can crowd the tape and siphon demand from adjacent growth names, especially unprofitable software and infrastructure plays that will be judged against a newly public benchmark. If the deal prices well, it could compress the private-market discount for other frontier-AI names; if it trades poorly, it becomes a valuation anchor that bleeds into the sector for 1-2 quarters.

The contrarian risk is that the market is extrapolating a clean window for issuance while ignoring execution fragility: large AI companies are still policy-sensitive, supply-constrained, and narrative-driven. Any sign of softer AI monetization, higher inference costs, or regulatory scrutiny between filing and pricing would hit appetite quickly, making the revenue uplift for banks more back-end weighted than the headlines suggest. Near-term, this is more about sentiment and league-table optics than immediate P&L, with the real catalyst arriving at filing and then again at pricing.

Key Decisions for Investors

• Long JPM / C into the expected filing window (next 2-8 weeks): best risk/reward is on incremental ECM and advisory fee optionality; size modestly because the benefit is diffuse but the headline flow should support multiple expansion if the process stays on track.
• Relative value: long JPM, short regional-bank ETF (KRE) for 1-3 months to isolate capital-markets optionality from balance-sheet-sensitive lenders; thesis is that deal-making optimism should accrue to diversified money-center banks first.
• Buy GS/MS pullbacks on any weak tape tied to rate or growth volatility, not on IPO timing: if OpenAI broadens the bookrunner set, the market may over-discount them despite still being core distribution beneficiaries.
• Short a basket of unprofitable AI software names against a long-public-AI-IPO hedge if the offering window opens: a successful debut could raise the bar for private comps and create 1-2 quarter relative underperformance in weaker-quality growth.
• Optionality trade: buy 3-6 month out-of-the-money calls on JPM rather than stock if you want exposure to capital-markets rerating without taking full banking beta; payoff improves if the IPO calendar accelerates, but downside is limited to premium.