Reflecting on the past year, I thought AI pricing would be somewhat settled by now. But the rate of pricing changes is only increasing.

The general trend is moving away from charging for access to software (seats) and toward a model of charging for the work delivered by a combination of software and AI agents (usage, outcomes). This shift was initially adopted by AI-native startups. Now it’s coming to SaaS companies who see their legacy business models as under threat.

With your help, I've spent the past two months investigating the latest in SaaS and AI monetization, collecting data from more than 230 software companies.

In today’s newsletter, I’ll unpack the six biggest takeaways that stood out to me:

Premium subscribers can read the full 37 page 2026 State of B2B Monetization report and join a subscriber-only AMA on May 19th (scroll to the bottom). Not a paid subscriber? Upgrade here.

Before we dive into the data, I wanted to share more information about who participated in the research. The 2026 State of B2B Monetization survey was run from April to May 2026. There was a fairly healthy mix of survey participants by company size, average annual contract value (ACV), and product types:

About two-in-five participants (37%) said they have a hybrid pricing model, which combines two or more pricing models like a per-seat subscription with AI consumption on top. This is the most popular pricing model.

Early-stage businesses with <$5M ARR gravitated most strongly to flat fees (37% adoption). This makes sense considering they have less usage history to draw from and are likely focused on acquisition at this stage. Large businesses with >$150M ARR continue to hold onto their legacy per-seat pricing (29% adoption).

These headline numbers look quite similar to last year’s report. Digging into the specific survey participants; however, the shift continues to be away from flat-fee or seat-based pricing and toward hybrid pricing. 25% of respondents said they were hybrid 12 months ago. That number has jumped to 37%.

My two cents: Tech investors used to hate variable pricing. They’re pushing for it in the age of AI. When asked which pricing model would be preferred by investors, only 10% said flat-fee subscriptions and 5% said seat-based pricing. Investors were far more keen on usage-based (24%), outcome-based (26%), or hybrid pricing (35%).

Today’s biggest monetization challenge: not enough expansion revenue. This was the most frequently cited pricing challenge overall, and was a particularly acute challenge among anyone with flat-fee subscriptions or seat-based pricing.

What can be done to fix this:

Seat-based companies are also worried their pricing model is under threat and not future-proof. Usage-based and outcome-based companies lament about how difficult it is to forecast and predict revenue. Companies with hybrid pricing are generally the happiest, yet struggle to explain their pricing to customers.

My two cents: Remember there’s no such thing as a perfect pricing model. The best you can do is to choose the pricing model that allows you to tell the unique story of what you do, who you’re for, and why you’re better than the alternative. And have a plan to manage the inevitable downsides.

Three-in-four software companies changed either pricing or packaging within the last year. What’s unusual: the largest companies are now making the most pricing changes.

It’s nearly impossible to find a >$50M ARR company that’s not rethinking both pricing and packaging for AI. In fact, the likes of Salesforce, Anthropic, HubSpot, OpenAI, Clay, Figma, Canva, SAP, and GitHub have made big pricing changes just in the first few months of 2026.

Be prepared for even more pricing changes. There’s an impending wave of new AI credits or token models. AI credit models already grew 126% year-on-year in 2025.

AI credit adoption today is at 29%. A whopping 33% of companies say they plan to introduce AI credits within 6-12 months. Among companies with >$50M ARR, roughly 1-in-2 (!) said they plan to introduced AI credits this year.

Satya Nadella summed up the sentiment well in Microsoft’s recent earnings call, saying seats are now “just entitlement to some consumption” meant to give customers a sense of budget predictability. Microsoft is pushing hard to a hybrid seats plus consumption model, and is launching AI credits for GitHub Copilot starting June 1.

My two cents: AI credits are great for vendors. They can become a nightmare for customers, especially once teams have to manage different credit models across dozens of their vendors. For now I see credits as a lifeline — just not the AI pricing endgame.

When companies look to monetize their AI capabilities, their first question is usually: what’s the target gross margin?

Internal costs and margins were the single most important factor when pricing AI capabilities (selected by 54% of respondents). Secondary factors were pricing relative to the market or competition (selected by 36%) and the human productivity gains unlocked by AI (selected by 30%).

The median target AI margin is about 50%. Only 12% aim for SaaS-like gross margins of 80%+ for their AI products. A similar number aims for 20% or lower gross margins (PostHog is a notable example).

My two cents: When products look like a commodity, companies don’t have much pricing power. Credits or tokens are the ultimate commodity pricing metric for AI.

If AI is going to deliver its full potential — and justify some truly astounding valuations — there will need to be a significant increase in technology spending. This probably needs to involve tapping into net-new pockets like headcount or services spending.

There’s limited evidence of this today. 70% of respondents said AI comes out of customers’ technology or software budgets. This figure is even higher among SaaS companies introducing AI features (75%).

That said, there is some evidence of AI-native companies growing the overall pie. AI-native companies are much more likely to tap into services budgets (35%) or, on occasion, headcount budgets (15%). This coincides with the explosion of AI-native services companies, which Sequoia thinks will create the next $1T company.

My two cents: Services companies sell work (quite literally). Imagine positioning as a replacement to Accenture, only to charge based on token consumption. Outcome-based pricing feels natural for AI-native services — although this may still get presented as a fixed-fee price for delivering a specific scope of work.

As AI pricing gets more complicated, some are giving up on the idea of one-size-fits-all pricing. 29% let customers choose between multiple pricing models, up from 21% last year.

Interestingly, offering multiple models is most common among those with usage-based or outcome-based pricing. I suspect the play is to use disruptive pricing as a part of their sales pitch: customers only pay when the product works.

My two cents: Outcome-based pricing is a great message to the market. But it might be a better message than buying model: many enterprises still crave predictability. There are four factors to successfully outcome-based pricing, which I call the CAMP framework: consistency of outcomes, attribution, measurability, and predictability.