Software pricing has cleared its decade-long ceiling, and the new reference class is labor value

Software pricing has cleared a ceiling it held for more than a decade. Martin Casado, general partner at Andreessen Horowitz, puts the shift plainly: “People want to try new apps, but unlike the 99 cents days, they’re willing to pay 200 a month.” That is not a marginal drift in price sensitivity. It describes a structural change in what consumers and professionals believe software is worth.

Casado’s framing extends beyond the current $200 tier. He asks what the $2,000-per-month version of a given product looks like, and answers with a phrase worth sitting with: “the Birkin bag of software.” His claim is that a luxury tier is not theoretical. Willingness to pay for it, he says, is already observable. The jump from a $0.99 app-store ceiling to a $2,000 luxury ceiling is large enough that the framing deserves scrutiny, but Casado is not making it in a vacuum. He also notes, with no apparent irony, that a browsing app he built to help manage his own social-media timeline carries a $250-per-user onboarding cost. The repricing that favors buyers is simultaneously straining unit economics on the builder side.

Harry Stebbings grounds the repricing in professional context rather than consumer behavior. Legal AI subscriptions, he observes, are running $10,000 to $12,000 per lawyer per year, against salaries of $200,000 or more. His arithmetic is explicit: that is roughly five percent of compensation. The ratio matters because it suggests buyers are no longer benchmarking software cost against prior software price points. They are benchmarking it against labor value. A tool priced at five percent of salary, if it delivers meaningful productivity, prices itself the way headcount prices itself.

If $20 was the historic ceiling, what's the $200 a month skew of your product? And in fact, what's the $2,000 a month skew? Like what's the Birkin bag of software? I think we're going to have this luxury software. We're already seeing willingness to pay for it. Martin Casado

The macro backdrop Eddie Lazar supplies is relevant context, though it requires a distinction. Lazar’s claim is that AI reached $100 billion in revenue in four years, while Software as a Service, or SaaS, took 15 years to reach the same milestone. He also argues that the total addressable market for AI can be more than ten times larger than traditional SaaS. These are big numbers stated without granular sourcing in the verbatim, so they function as directional framing rather than precise benchmarks. But the directional argument is coherent: if AI is priced against labor and task value rather than against prior software pricing, the addressable market is categorically larger than the one SaaS competed for.

What ties these observations together is a change in the reference class buyers use. The old mental model treated software as a product category with its own price norms, norms set partly by the app-store era and partly by the logic of marginal-cost-zero distribution. The new mental model, at least for AI tools with clear productivity applications, treats software cost as a fraction of the labor value it displaces or augments. That shift, if it holds, has implications that run beyond headline subscription prices.

The picture is not uniformly clean. Casado’s $250 onboarding cost on a consumer app points to a tension: the same AI cycle that has raised willingness to pay has also raised the cost of delivering the product. Whether the margin math works at scale remains an open question, and the luxury-tier thesis is forward-looking enough that no current evidence confirms it fully. What the evidence does support is that the old ceiling held for a long time, broke recently, and the new price points are finding buyers. How far the repricing travels from here, and whether the unit economics on the supply side close, is the part worth watching.