Most companies signing AI deals don’t know the pricing model yet.

In 2026 we’re seeing enterprise deployments of not just end-user AI chatbot assistants, but AI inserted at various entry points in workflows across three main sectors: Helpdesk Support, Enterprise Platform / Indirect Assistants, and AppSec. These are emerging as the first and most productive use of the new technology; however, the industry at large has not settled on price model, and the landscape remains highly competitive in most places, while nearly monopolized in key niches.

Green Cabbage has been tracking deals across major names such as OpenAI, Microsoft, and Google’s Gemini, in addition to emerging specialized solutions such as Arize AI, CursorAI, SkanAI, and Sierra AI, and we’re seeing AI augment platforms of pre-established suppliers such as Pegasystems, Semgrep, and Q2.

While we’re likely going to see enterprise-wide adoption of basic assistants from Microsoft or Google, there is a complex emerging market for AI assistants or augmented workflows to be inserted at various levels of the workflow — not just for end users. Suppliers are trying to find an appropriate cost model and typically run into two issues: (1) users do not consume interactions or requests at a consistent rate and (2) consumption may not fully encompass the work provided by a solution.

While AI-augmented uplifts can retain a standard SaaS pricing model, sometimes with consumption of “AI Units,” we saw suppliers in 2025 shift from per-user pricing to include consumption pricing. This arose as data showed that chatbot users tend to consist of about 80% average users, 10% above-average users, and 10% “power users” which can account for a third to a half of total consumption.

The cost of these solutions is closely tied to utilization, and the major suppliers have started to introduce consumption unit bundles and caps on regular user-based pricing.

So far it appears to only act as a hedge against customers who consume dramatically on fewer users, and we have not seen it limit any of our customers. We anticipate for the price models to retain per-user pricing in small deals (<$250K), while consumption pricing becomes more relevant in larger enterprise deployments.

Price models are important to track because they dictate what type of price protections are appropriate in any given deal, and consumption models can require many weeks, if not months, of planning to sign confidently.

This is critical as we are anticipating price hikes on the horizon AI. It’s largely understood that right now deals are mostly “entry level” pricing with low margins, and we expect these costs to rise significantly in coming years. In order to protect against these price hikes, and ensure that appropriate price models and controls are being applied, Green Cabbage reviews our customer’s proposals and provides actionable insights to materially improve the quality of your company’s contracts.