Just a few months ago, it seemed like today’s Anthropic IPO, expected to be the biggest in history, would consummate a marriage between the eccentric world of artificial intelligence and the ruthless realm of finance.
For years AI and Wall Street had existed as profoundly different realities—one obsessed with the superhuman and utopian and speculative, the other with the immediate and material and commodifiable. But the two domains seemed to converge decisively across 2025 and the first half of 2026, especially after Claude Code made it clear that AI could really do the work of certain highly educated humans—puncturing expectations of an AI bubble and sending Wall Street’s AI valuations to new heights.
Since the summer, though, the two realities have separated once again. Anthropic’s debut may help rescue the relationship, but the wedding date has been postponed.
That’s not because of any want of ambition on the AI side. The latest talk in Silicon Valley is, if anything, more radical than ever, envisioning shorter timelines for superintelligence, exponentials going up and up, AI building new AI which will build new AI in a spiral that could make the next few years decisive for the human future. There are more and less measured versions of these expectations, but the assumption of continued takeoff is broadly shared across the industry.
In the world of finance, though, the AI launchpad suddenly looks a lot more fragile.