When the bank handling your $2 trillion IPO also finances the entire industry you're disrupting, every forecast becomes a conflict of interest.
The Summary
- Morgan Stanley, set to underwrite Anthropic's massive IPO, insists AI spending won't slow despite market concerns, while also banking nearly every major AI infrastructure player
- Anthropic faces pressure to delay its $2T public offering over AI safety and regulatory concerns, as METR, the AI safety evaluator, battles conflict-of-interest allegations involving shared funders and staff
- The timing collision: Anthropic's political team is positioning AI safety as a midterm election issue just as their underwriter tells investors to ignore safety concerns and keep spending
The Signal
Morgan Stanley's public stance on AI spending carries unusual weight right now. They're the lead underwriter for what could be the largest tech IPO in history, and they bank the infrastructure companies building the data centers, the chip makers supplying the GPUs, and the cloud providers renting the compute. When they say spending won't slow, they're not just analyzing the market. They're describing their own revenue forecast.
The timing gets messier. Critics are urging Anthropic to delay the IPO until regulators can assess safety frameworks that don't exist yet. Meanwhile, METR, the organization that's supposed to independently evaluate AI safety risks, is defending itself against allegations that it shares funders with Anthropic, receives free API tokens from the company, and employs former Anthropic staff in key roles.
"When the safety watchdog and the company being watched share the same donors, 'independent evaluation' becomes performance art."
METR denies taking Anthropic money directly, but the viral report detailed three structural conflicts:
- Shared funding sources between METR and Anthropic
- Free compute tokens from Anthropic for testing (how do you objectively evaluate a product when you can't afford to test it without their help?)
- Staff who rotate between METR and Anthropic, blurring the line between regulator and regulated
Then there's the political play. Anthropic's Sarah Heck is making AI safety a voter concern ahead of midterms, framing the company as the responsible actor in an industry gone wild. It's smart positioning, but it sits uncomfortably next to an IPO roadshow where the underwriter needs to convince investors that AI risk is priced in and spending is infinite.
The contradiction is loud: Anthropic's advocacy team says AI safety matters enough to sway elections, while their investment bank says it matters so little that capital deployment won't pause. Both can't be right. Either safety concerns are real enough to slow deployment (bad for the IPO), or they're manageable theater (bad for the political narrative).
The Implication
Watch how Anthropic navigates the next 60 days. If they delay the IPO, it signals that regulatory risk is real and they're willing to sacrifice near-term liquidity for long-term legitimacy. If they push forward, it means Morgan Stanley's bullish forecast matters more than their own safety rhetoric.
For investors, the question isn't whether AI spending continues. It's whether the companies spending it can credibly claim they're doing it safely when the safety evaluators share their funding, their staff, and their incentives. The market will price that risk eventually. The only question is whether it happens before or after the IPO.