How Are SaaS Companies Pricing AI?

Pricing AI functionality has unique and material financial ramifications because AI creates COGS. Invidia is not worth three trillion dollars for nothing; AI consumes significant computing resources, which a SaaS vendor pays for and then decides to absorb, pass along, or profit from.

I recently surveyed my LinkedIn colleagues (n=60) about how they priced AI. While the results are not a quantitative gold standard, they provide some directional insights.

First, only 15% of respondents said they were NOT planning to launch AI capabilities. That’s a bit shocking.  If you think simply offering AI will be a differentiator for your firm, think again. All your customers will soon be approached by their various SaaS vendors with new AI functionality. They are not going to pay more for every one of them. In addition, a new wave of AI-centric products is coming to market, competing for budget and functionality.

If the value-add you your company’s new AI functionality is immediate, obvious, and material, you can monetize it and share the value with your customers. Conversely, AI might be a free or low-cost add-on if the value-add is less obvious or material. In fact, according to the survey, of those companies offering new AI functionality, 26% were not planning to charge for it. They seem to recognize that AI is not a game-changer for their product and will roll it out to bolster core pricing and retention. That’s sensible, but it will cause a P&L hit due to the higher COGS supporting AI consumption.

Of those planning to charge for AI, 40% will charge an additional subscription fee, while 60% will charge on a consumption basis. A bias toward consumption pricing makes sense for three reasons.  First, if a customer is not using the new AI functionality due to a lack of training, inertia, or little perceived benefit, subscription pricing will drive churn or shrinkage. Second, customers may have aggressive AI users that will cost more than they are paying if everyone is paying a flat subscription fee. Third, and this is where consumption pricing becomes the default, there may be no “seats” at all to base the subscription fee on as the value is driven through APIs.

In each case, consumption pricing does a better job than subscription pricing of linking value to price.

Adding a subscription fee is the “easy button” to get an AI product to market, but usage-based pricing is uniquely suited to match the value AI delivers. Implementing usage-based pricing is hard for those who have not previously dealt with it, but as pricing and billing capabilities mature, most AI offerings will migrate to usage-based pricing.