By Nathan Brooks

77% of enterprises reevaluate their AI vendors every six months or on a rolling basis, per a Madrona survey of 150 enterprise IT professionals. The venture firm called the cadence a sharp break from the past.

Madrona writes in its report: "This creates a 'fast in, fast out' dynamic that is fundamentally different from traditional enterprise SaaS, where multi-year contracts provided a moat of inertia. In enterprise AI, switching costs are lower and the re-evaluation cadence is relentless."

What does the spending picture look like?

Madrona's respondents are not pulling back. 74% plan to expand AI budgets in the next 12 months. The other 26% plan to hold spending flat. None reported planned cuts in the published finding.

That appetite sits inside a much larger pool. IDC projects $4.25 trillion in global technology spend during 2026, almost all driven by AI. The enterprise checkbook stays open.

Are the pilots actually working?

Not reliably. Fewer than half of enterprise AI pilots reach full production, Madrona found. The figure qualifies as progress.

Last year, MIT reported a 5% success rate on enterprise AI ROI. Going from 5% to sub-50% is a meaningful jump, although the bar stays low. Vendors with a working product still face a graveyard of stalled proofs of concept.

Why does the ARR feel shaky now?

The volatility extends past pilot stage. Enterprises that adopt AI also refuse multi-year commitments. 77% review vendors every six months or sooner, the survey found.

The implication reaches beyond vendor scorecards. Startup ARR, the annualized revenue figure that powered 2025's AI boom, no longer enjoys the protection of a long contract. A signed deal can vanish at the next budget cycle, and that prospect keeps CFOs awake.

What changed about pricing?

Token-based billing follows a SaaS-era template. Once an enterprise knows it needs email, HR software, or cloud storage, it pays per seat or per gigabyte without much fuss.

AI buyers reject that pattern. A separate Andreessen Horowitz survey of 50 technical AI buyers found more than half want fees tied to outcomes rather than to tokens consumed. The shift pushes vendors to price per report processed, ticket closed, or lead generated.

A16z partners Tugce Erten and Sarah Wang wrote that pricing "around the recognizable work" makes the product "economically valuable to both sides." The framing helps sellers prove worth at the demo stage and helps buyers defend cost internally.

What does this mean for sales teams?

Reps must now pitch value-per-output, not throughput. The deal narrative moves from "you will consume X tokens" to "we will produce Y reports for you." Pilots succeed when the unit of work is named up front.

For larger sales orgs, that means reshaping discovery calls around measurable outcomes. Smaller teams selling into one or two verticals can codify that playbook faster. Either way, six-month reviews now stand between a signed contract and the revenue it was supposed to lock in.

What's the forward read?

The 2025 AI boom was funded by enterprise trial budgets. The expectation for 2026 was that those trials would harden into multi-year deals. That contract moment keeps slipping, and the next test arrives at the next budget cycle.

A startup whose ARR rests on shaky enterprise rails has six months at most to prove durable worth, or its revenue evaporates and the cycle restarts.