The token economy: how AI changed the way enterprises buy software.

Enterprise software is now bought in tokens and credits as often as it is bought in seats. In 2026 the largest software companies began charging for AI by the unit of work it performs, and that change has rewritten how buying teams evaluate, approve and renew software. If your picture of your buyer’s journey was formed before this year, it describes a market that no longer exists.

What is the token economy in enterprise software?

The token economy is a pricing model in which software companies charge for the AI work their products perform. A token is a small piece of text, roughly three quarters of a word, that an AI model reads or writes. Every AI request uses tokens, and every token costs the vendor money in computing power.

Seat pricing worked because each user cost the vendor about the same amount. AI broke that model. One employee who runs an AI agent for an afternoon can use more computing power than a hundred employees who ask the assistant a few quick questions. Vendors could no longer charge a flat fee per person without losing money on heavy users.

Most vendors responded by adding a usage meter to the subscription. Some bill raw tokens at published rates. Most convert tokens into a branded currency called credits, which buyers prepay in bundles and spend down as their teams use AI features. Either way, the price of the software now depends on how it is used.

Which major software companies introduced token or credit pricing in 2026?

At least 28 software companies introduced or expanded token or credit pricing for AI between January and November 2026. The table lists each change with the date it was announced or took effect, newest first.

What must buying teams consider when software is priced in tokens?

Buying teams now have to judge how much a product will cost to use, in addition to what it costs to license. These seven questions now shape enterprise software decisions.

• Buyers must forecast usage before they can compare prices. A quote for credits means little until the team knows how much AI work its people will run, so many buyers now run paid pilots on real workloads to measure consumption.
• Buyers must learn what one credit buys. Each vendor sets its own credit rates by feature and by AI model, so the useful comparison is the cost to complete one task, such as resolving a support ticket or drafting a report.
• Buyers must plan for running out and for leftovers. Teams need to know whether AI features stop, keep running at an overage rate or top up automatically when credits run out. They also need to know whether unused credits carry over. Notion AI credits reset each month and do not roll over, for example.
• Buyers must decide who can choose the AI model. Premium models use credits faster than standard ones, so a setting that lets any employee pick the most powerful model can multiply the bill.
• Buyers must set spending controls before rollout. Admin tools for caps, alerts and budgets by team have become part of the evaluation, because a single heavy user can drain a shared pool.
• Buyers must bring finance in earlier. A variable bill changes the approval process. Finance and IT operations teams now help model costs during evaluation and track spend after purchase.
• Buyers must reread renewal terms. Vendors are applying new pricing at renewal, and some require customers to buy a minimum number of credits each month. A contract that renews on old terms this year may renew on very different terms next year.

Why is your buyer journey research probably out of date?

Most B2B software companies built their personas and journey maps from buyer interviews held a year or two ago. As many companies changed their pricing structure, their buyers changed their buying process. If you haven’t spoken to your target market this year, your messaging is aimed at a buyer who no longer exists.

The gaps tend to show up in a few places. Your map may miss finance as an early stakeholder. Your sales team may lead with seat price when the buyer is asking about cost per task. Your trial may be too short for a buyer who needs a month of usage data before procurement will approve a credit commitment.

Ask your buyers these 10 questions about how they evaluate and purchase software in the token economy:

• Who on your team is involved in evaluating software that charges for AI usage, and at what stage does each person get involved?
• How do you estimate the amount of credits or tokens your team will use before you sign a contract?
• What would you need to see during a trial or pilot to feel confident about a usage-based price?
• How do you compare the cost of two products when each vendor uses its own credit system?
• Which pricing model do you prefer for AI features? (Some models include a fixed allowance, prepaid credits and pay-as-you-go)
• What does your finance team need before it will approve a software purchase with a variable monthly cost?
• What spending controls must a product offer before you will roll it out across your organization?
• Who in your organization decides which AI models employees are allowed to use?
• What would make you walk away from a vendor because of how it charges for AI?
• How do you plan for a contract renewal when you are unsure what your AI usage will cost next year?

Emporia Research runs buyer journey studies with verified B2B decision makers, so you can answer these questions with data from the people who sign the contracts. Talk to our team to get started.

What do B2B marketers ask about token pricing?

What is token-based pricing?

Token-based pricing charges customers for the amount of text an AI model reads and writes on their behalf. A token is roughly three quarters of a word.

What is the difference between tokens and credits?

Tokens measure raw AI usage. Credits are a vendor’s own currency that converts tokens and other AI actions into a single unit that customers prepay and spend down.

Are seat licenses going away?

Most vendors in 2026 kept a per-seat fee and added a usage meter on top. Buyers now pay for access and for AI usage as separate line items.

How should B2B marketers respond to the token economy?

B2B marketers should interview their buyers again to learn who now joins the evaluation, how buyers forecast usage and what finance needs to approve a purchase.