Forced AI bundling: the 2026 renewal tax
The 2026 renewal quote arrived with a mandatory AI tier you did not ask for, priced 20 to 37 percent higher. Here is the tax, and the two ways off it.
The 2026 renewal quote arrived with a line item you did not ask for. An "AI-enhanced" tier, mandatory, priced 20 to 37 percent above what you paid last year, for a copilot half your team has not opened. This is not a rogue account executive. It is the single largest driver of enterprise software cost increases this year, and it has a name: forced AI bundling.
Software prices are rising about 8.4 percent on average in 2026, but the average hides the mechanism. The increase is concentrated in vendors who moved their AI features from an optional add-on into the base tier, so the only way to renew is to buy the AI whether you use it or not. Here is the shape of the tax, and the two ways off it.
What the tax actually looks like
The clearest example is ServiceNow. Its Now Assist AI add-on carries a 50 to 60 percent uplift on base-tier pricing. An ITSM Enterprise customer paying $160 to $200 per fulfiller a month is quoted $240 to $320 after the migration, for the same fulfillers doing the same work. Broader platform deployments are seeing annual contract value climb 10 to 15 percent at renewal on platform expansion and Now Assist together.
The vendors are not identical. Salesforce raised enterprise prices about 6 percent and sells its agents on separate consumption pricing. ServiceNow chose to bundle the AI into the platform tier instead. Either way, the AI uplift on 2026 renewal quotes lands in the 20 to 37 percent range, and it is the leading edge of a structural repricing across the sector, not a one-year blip.
Why the play works
Forced bundling works for the same reason every renewal increase works: switching is expensive and the vendor knows it. The AI narrative just gives the increase a story. "You are getting AI now" reframes a price hike as an upgrade, and the buyer who has not budgeted a competitive evaluation accepts the bundled proposal because saying no feels like falling behind.
The number that should stop that reflex: organizations that treat the AI add-on as a discretionary purchase requiring its own evaluation land pricing 20 to 35 percent below those who accept the bundle. The AI tier is not free, and it is not mandatory in the way the quote implies. It is a line you are allowed to interrogate.
The three questions to run at every renewal
Procurement leaders who are holding the line in 2026 run the same filter on every renewal quote. It works on the AI tier specifically.
- Is this vendor priced from a time when their unit costs were 3 to 5 times higher? Most enterprise SaaS was priced years ago. The underlying model cost inside the new AI tier has fallen sharply while the sticker went up. That gap is wrapper margin.
- What percentage of the platform do we actually use? The honest answer for a big-platform deployment is usually 15 to 25 percent. You are being asked to pay an AI premium on the 75 to 85 percent you do not touch.
- If we cancelled, what is the rebuild math? This is the question that changed. For the slice you use, on direct billing, the rebuild is often cheaper than a single year of the bundled increase.
Move one: renegotiate the AI back out
The fastest response is to un-bundle. Treat the AI tier as a separate line, ask for the base renewal without it, and put the AI capability out to a real competitive look. That alone recovers the 20 to 35 percent gap between bundle-takers and bundle-questioners. Downgrade unused tiers, rationalise seats, and trade a multi-year commit with escalators for a one-year term while you decide.
This is the right first move for most vendors. It is also a stopgap. It resets the price for a cycle; it does not change the fact that the platform is priced to extract more every year. If the tool is genuinely load-bearing and differentiated, renegotiate and keep it. If it is a per-seat wrapper on a commodity workflow, go to move two.
Move two: rebuild the slice you use
We never rebuild a whole platform. We rebuild the 15 to 25 percent of it your team actually uses, on your own cloud and your own model contract, so the AI you were about to rent runs on direct billing at a fraction of the resale price. The wrapper margin goes to zero.
Take the ServiceNow shop above, 200 fulfillers at the post-uplift $280 a month. That is $672K a year, up from about $432K before Now Assist, a $240K annual increase for the same work. The slice most of those fulfillers touch, ticket intake, routing, knowledge lookup, and status, is a well-understood build. A fixed-price rebuild of that slice, with the AI running on your direct model contract, typically lands its payback inside 6 to 12 months against a number like that, and every year after is flat while the platform quote keeps climbing.
When not to rebuild
Scope honesty matters here, because not every renewal is a rebuild case. Keep and renegotiate when the platform is deeply integrated across many teams, when the workflow genuinely spans the whole product rather than a slice, or when the switching risk outweighs a single year of increase. Security tooling, foundational systems of record, and genuinely differentiated capabilities usually belong in the keep column. If that is your situation, we will tell you so and help you scope the renegotiation instead of a build.
The point is not that every AI tier is a scam. It is that the AI tier is a decision, not a default. Run the three questions, un-bundle by reflex, and rebuild the slice only where the math is obvious. The renewal that arrives assuming you will pay the premium is counting on you not doing the arithmetic.
Read more: /rebuild/ · SaaS renewals in H2 2026 · Stack rationalisation brief · SaaS stack audit