Yesterday, Confluent filed to go public. Looking at the numbers in their S1, there were clear undertones of a shift in infrastructure trends. All focus was on a small (15%) but growing revenue segment for Confluent Cloud, a usage-based cloud offering. Confluent Platform, the on-prem subscription solution, shared none of the spotlight.
Some of the best-in-class SaaS startups have usage-based pricing: Snowflake, Stripe, Twilio, and Datadog. Usage pricing allows costs to be more correlated with value delivered, for both the customer and the startup.
I think that usage-based pricing is just another go-to-market strategy. It's about getting to your customers and making it as easy as possible for them to try out your product. Usage lowers the barrier by letting customers integrate with a level of commitment they feel comfortable with. Put it on the spectrum of open-source (free, but you need to host it yourself) and freemium (free, but you need to integrate it yourself) of go-to-market models.
But there are some downsides. Future cash flows become less predictable and customers might not have a good idea of what a service will cost if the basis is confusing. This is an easy problem to solve with committed spend, something that happens often with larger enterprise contracts because its in the interest of customers and providers to forecast. Doing usage based early on, and then usage based + committed spend later on looks a lot more like subscription with a freemium go-to-market. Except, this model is more closely aligned with delivered value.
Finally: Why now? Cloud and autoscaling are the technical inflection points that enable usage-based pricing. It's difficult to do usage-based accounting on-prem and without autoscaling, providers would find it nearly impossible to right-size their infrastructure for varying usage.