Corporate clients are shortening software contracts because of Anthropic and other AI services

AI is already changing not only the products themselves, but also the conditions of their sale. Corporate customers are increasingly declining long-term software contracts because they fear being left with a tool that will become obsolete before the contract ends.

Why timelines are shrinking

For more than ten years, corporate software followed established rules: a vendor sold a subscription for several years ahead, received predictable revenue and high margins, while the customer received a discount in exchange for the long term. This is the model on which much of Silicon Valley's SaaS market was built. But rapid AI progress breaks the very idea of long-term lock-in.

If a company signs a three-year contract today, it is no longer confident that in 12 months that product will not look obsolete compared to new agents, models, and interfaces. This anxiety is already visible not only in negotiations but also in the public market. Following reports that Amazon's cloud division is developing an AI agent to automate sales and business development functions, a software ETF fell more than 4%, while individual stocks declined by about 9%.

Investors read the signal directly: if large companies can automate part of office work with agents, the classic "pay per user" model will no longer look as stable as it used to.

Customers want flexibility

For buyers, the main risk has become not the price itself, but the risk of making the wrong platform choice. According to Andreessen Horowitz investor Julie Yu, startups selling AI solutions to hospitals used to almost always aim for multi-year contracts. Now they increasingly offer annual or even monthly agreements because customers want to compare multiple products in parallel. Instead of one long decision, the market is shifting into a mode of constant testing: first a pilot, then a short contract, and only after that is a more serious purchase possible. This shift has several reasons:

• prices on AI products and computing change too quickly to lock them in for years ahead;
• new models and features come out every few months and can dramatically shift the price-to-quality ratio;
• many tools enter companies from the bottom up, when employees first use Claude, Lovable, or similar services themselves;
• sales cycles are shrinking from months to weeks, which means purchase decisions are increasingly made as an extension of already-started usage.

"I advise companies to avoid three-year contracts on AI products: no

one knows what their cost will be in three years."

Sales come from the bottom

One of the most striking changes in how AI is sold to business is the disappearance of a clear boundary between B2C and B2B. According to Menlo Ventures investor Derek Xiao, many corporate deals now start with personal use. Employees experiment with tools for code generation and internal applications, then bring them to their work teams.

If the experiment proves useful, the company does not go through the long path of classical procurement from scratch, but simply formalizes what has already taken root internally. For suppliers, this accelerates growth but makes customer relationships less stable. The logic of competition is also changing.

Experienced enterprise startups already see that their advantage cannot be built solely on long implementation and high cost of switching vendors. Legal tech startup Filevine, which competes with new players like Harvey and Legora, bets on agent functions on top of the existing system to speed up launch with customers and not lose to those who promise faster onboarding. For the venture market, this is an unwelcome turn: the old assumption that enterprise sales are more reliable than consumer sales is beginning to weaken.

There is higher churn risk, it is harder to maintain margins, and it is more difficult to defend the product moat.

What this means

AI is changing not only the categories of software themselves, but also the mechanics of procurement. For business, this means a shorter planning horizon and caution toward any long-term commitments. For vendors, it means the end of an era when a multi-year contract was almost automatically seen as a sign of strength. Now the winner is not the one who locks in the customer for three years better, but the one who proves value faster and is willing to live in a mode of constant revision of terms.