Enterprise AI vendor contracts don't stick, study finds

A new Madrona Venture Capital survey of 150 enterprise IT professionals found that 77% now re-evaluate their AI vendors every six months or on a rolling basis, even as 74% plan to expand AI budgets over the next year. Budgets are growing, but the vendor relationship behind them isn't sticking.

That's a different pattern than traditional SaaS, where a signed annual contract used to buy a startup a year of predictable revenue. For an AI-focused product, a healthy ARR number no longer means the buyer is locked in past the next procurement cycle. Enterprises are treating AI tools more like commodities to swap out than platforms to standardize on, which turns a signed deal into a rolling audition instead of a settled relationship.

Pricing is part of why. A separate Andreessen Horowitz survey of 50 technical AI buyers found more than half now prefer pricing tied to the work actually produced, not token or seat counts. Usage-based pricing made it easy to start small; it also makes it easy to walk away small, since there's no seat count or annual minimum tying the buyer to the vendor.

Selling an AI-powered product to enterprise customers, the practical takeaway is that closing the deal isn't the finish line. The vendors surviving the six-month re-evaluation cycle are the ones who can point to a specific, current result the customer is getting, not just a signed contract from two quarters ago. Build the proof of ongoing value into the product and the renewal conversation now, before the next procurement review makes the case for you.