SaaS под ударом: почему ИИ заставляет инвесторов бежать из облаков

SaaS Under Pressure: Why AI Is Forcing Investors Out of the Cloud

The cloud software market right now resembles a party that the police suddenly burst into and shut down the music. For ten years in a row we've been hearing that software eats the world, and the subscription model is the perpetual profit engine. But recent reports from Salesforce, Workday, and other sector giants have sent investors fleeing to the exits in panic.

Stocks are falling tens of percent in a single day, and this is not just a temporary weakness. We are witnessing the beginning of a massive revaluation, where the greatest threat has become that very artificial intelligence everyone was so hopeful about. The context is simple: a decade of cheap money allowed companies to bloat their headcount and buy hundreds of software licenses that often sat idle.

Now the rules of the game have changed.

The core problem lies in the very DNA of modern SaaS — the per-user or so-called seat-based pricing model. This system worked perfectly as long as company growth meant hiring new people. More people — more licenses — higher vendor revenue.

But generative AI and autonomous agents shatter this logic to pieces. Where once it took twenty people with CRM licenses to process incoming sales inquiries, tomorrow one operator and a couple of smart scripts will handle it. For the client this is a blessing and efficiency, but for Salesforce it means revenue falling ten-fold.

Investors have finally realized that AI-driven automation is a zero-sum game for traditional cloud services. The better AI works, the fewer people are needed in the system, and the less money the software maker gets.

Companies like Klarna are already openly saying they're freezing hiring and abandoning chunks of corporate software in favor of their own AI solutions. This sets a dangerous precedent. Why pay millions of dollars a year for a heavyweight platform when you can build a custom agent based on GPT-4 or Claude that's perfectly tailored to your workflows? Old players are trying to defend themselves by embedding AI assistants into their interfaces, but often it looks like trying to bolt a jet engine onto a cart. Clients don't want to pay for AI as an add-on; they want AI to simply solve their problems. This leads us to the inevitable shift from paying for access to paying for results.

The shift to consumption-based or outcome-based pricing looks like a nightmare for Wall Street giants. It makes revenue unpredictable and forces companies to actually be accountable for the value of their product. You can't just sell a three-year contract and forget about the customer anymore. Now you need your algorithm to actually work and deliver profit. In this new reality, the advantage goes not to those with the longest feature list, but to those who are most deeply integrated into workflows and possess unique data for training models. Vertical AI, created for specific niches, is beginning to displace horizontal platforms that tried to be everything for everyone.

The big question now is whether the old cloud kings will manage to transform themselves. So far their attempts look unconvincing. They're caught between the need to invest billions in AI infrastructure and declining growth rates in their core revenue. Investors see this and are shifting money toward hardware and pure infrastructure, leaving applied software in the red zone. We're entering a period of major consolidation where only those who can prove their product is not just a convenient cloud spreadsheet but an indispensable intelligence will survive. The era of easy subscription money is officially over, and from here on it only gets more interesting.

Main point: The per-user payment model is dying under pressure from AI agents, and the market will have to learn how to value software anew.