Payments and Software Stocks Drop as Analyst Warns AI Could Disrupt Revenue Models
Shares of major payments processors and software companies fell sharply Thursday after Citigroup analyst Andrew Schmidt published a research note warning that artificial intelligence could fundamentally reshape how these businesses capture revenue—or whether they capture it at all.
The sell-off hit companies across the financial technology stack. Payment processors that charge per-transaction fees and software vendors built on usage-based pricing models saw the steepest declines, as Schmidt's analysis suggested AI agents could bypass traditional monetization points entirely.
For CFOs in the sector, the note crystallizes a question many have been privately gaming out: if AI handles more transactions autonomously, do the old tollbooths still work? Schmidt's argument is that they might not. The concern isn't that AI will make these companies obsolete—it's that AI might make their pricing models obsolete, which from a finance perspective is roughly the same problem.
The thesis goes like this: today's payments infrastructure charges fees when a human initiates a transaction through a specific interface—a checkout page, a point-of-sale terminal, an API call. But if an AI agent is orchestrating thousands of micro-transactions on behalf of a user, the economics change. Does the payments company charge per transaction (making AI prohibitively expensive)? Per user (leaving massive revenue on the table)? Per AI agent (good luck defining that in a contract)?
Software faces a parallel problem. Usage-based pricing works when "usage" is a proxy for value delivered. But if an AI is hammering your API 10,000 times to accomplish what a human user would do in one session, is that 10,000x the value? Or is it just how AI happens to work?
Schmidt's note didn't provide specific revenue impact estimates or timelines, but the market's reaction suggests investors are taking the structural risk seriously. The sell-off was broad-based rather than company-specific, indicating concern about the business model rather than individual execution.
The irony, of course, is that many of these same companies have spent the past year telling investors that AI will expand their addressable markets. And it might! But Schmidt's analysis suggests the expansion might come with a catch: the revenue per transaction could compress faster than transaction volume grows.
For finance leaders, this is the kind of disruption that doesn't show up in quarterly guidance until it's already happening. The warning signs won't be in top-line revenue—that might even grow as AI adoption accelerates. The canary in the coal mine will be in the unit economics: revenue per transaction, revenue per API call, customer acquisition cost relative to lifetime value.
The question CFOs will be asking their product and strategy teams: if AI agents become the primary interface for our service, what exactly are we charging for? And can we get there before the market figures out we haven't answered that question yet?
Responses (0 )