Pricing a SaaS model was once a straightforward task, with standardised gross margins and tiered plans. But the rise of AI has done away with this playbook — and as the industry constantly develops and iterates, founders are having to adapt quickly to maintain competitive pricing structures without breaking their billing and finance operations.
Calculating usage in real time, converting and invoicing across multiple currencies and ensuring revenue is recognised correctly for both internal reporting and compliance are now core operational hurdles.
“The AI industry has largely moved away from pure subscription or pure pay-as-you-go models,” says Kamael Sugrim, head of startup and investor partnerships across EMEA at payments and billing company, Stripe, which offers the tech 78% of the Forbes AI 50 build on. “Most AI companies are now adopting hybrid pricing models, combining flat subscriptions with usage- or outcome-based components.”
So what is the new rulebook of pricing a novel technology like AI, as you scale?
Moving targets
AI founders have their work cut out when it comes to pricing, says Manal Belaouane, principal at Munich-based investment firm, HV Capital.
“AI’s unit economics are moving targets,” she says. “Inference costs shift with model choice, context length, latency/quality settings and vendor changes, so gross margin on the same workflow can vary week to week. Buyers, meanwhile, want predictable budgets, but pure usage pricing can’t ensure this predictability.”
The focus has shifted to reflecting the tangible value delivered to customers — such as time saved, improved quality or specific business outcomes — rather than simply passing through compute costs.
A major finding in Stripe’s recent Indexing the AI economy report was that innovative business models and monetisation strategies are emerging quickly within the AI industry. This means that startups have to stay flexible to keep up with a changing landscape — by partnering with a billing platform early, such as Stripes Startups Programme, switching up their pricing, or both.
Part of this has seen founders shifting away from pricing based on cost management and towards a more outcome-based approach to numbers, says Sugrim.
“The focus has shifted to reflecting the tangible value delivered to customers — such as time saved, improved quality or specific business outcomes — rather than simply passing through compute costs,” she says. “This evolution allows pricing to directly reflect the actual work performed or the measurable business results achieved, which customers inherently perceive as higher value.”
For example, rather than charging based on usage, the team behind software company Intercom’s new AI-powered customer service model collaborated with Stripe to develop innovative pricing based on successful outcomes. Users pay $0.99 per resolution — meaning they are only charged when a customer confirms the AI answer resolved their issue, or doesn’t ask for more help after the last AI answer.
“We realised, 'hey, it might be nice if we don't have to build all of this ourselves' and we can lean on the Stripe Billing platform for some of it," says David Lynch, Intercom’s VP of engineering.
London-based integration platform StackOne is another example.
“We anchored our price to customer outcomes, not just features,” says CEO and founder Romain Sestier. “For us that meant time-to-integration across a wide breadth of tools, the ability to win or retain revenue and the engineering months we save versus building and maintaining dozens of connectors in-house.
“We started with a simple structure: a core platform fee plus usage tied to connected accounts and integration domains. We avoided per-event billing so sales and finance leaders could forecast costs, and we found this to be more aligned with how our customers price their AI Agents,” he adds.
Scaling up
Once the initial pricing plans have been set, founders then need to consider the evolution of these numbers as they — and their clients — scale.
“When scaling, pricing models should be designed to encourage productive usage and align with how customers naturally grow with the product. This involves balancing predictability, ease of adoption and scalability, ensuring pricing scales alongside increased usage and customer value,” says Sugrim.
This is where a hybrid model of pricing can come in handy: ‘The hybrid pricing model offers inherent flexibility to adapt as usage grows. Founders can adjust their pricing elements by introducing tiered volume discounts, modular add-ons for new features or committed usage discounts for high-volume customers, all within a predictable framework,” she adds.
Founders should seek a provider that offers out-of-the-box support for common hybrid models.
And more billing platforms, like Stripe, are accommodating these flexible plans.
“Stripe helps AI companies set up and evolve hybrid pricing models, which are prevalent in the AI application layer. Founders should seek a provider that offers out-of-the-box support for common hybrid models and highly flexible components to build and configure various pricing permutations without extensive custom code,” says Sugrim.
For StackOne, scaling meant rethinking the structure of the pricing model they had begun with.
“As we grew into more enterprise deals, we moved from one-size pricing to a modular model: core platform, domain packs [and] volume of connected accounts. We also introduced committed-use tiers and expansion ramps so customers can plan their spend as they grow their customer base,” says Sestier.
These changes weren’t just to reflect the growth of StackOne, but also served to accommodate the growth of their customers.
“The shift was triggered by security and procurement needs and multi-product buyers wanting clear economics as they scale,” Sestier says.
Clear communication
The importance of communicating pricing plans — and any changes as the tech develops — also shouldn't be overlooked.
“Founders should approach pricing with a structured, value-first five-step framework. This begins with defining clear value metrics from the customer's perspective, then selecting charge metrics that align with both value and cost,” says Sugrim.
Founders should approach pricing with a structured, value-first five-step framework.
For Sestier, this clarity also proved crucial when speaking to VCs.
“Investors cared that our model was value-linked, could support software-grade margins without pro-services bloat, and had clear expansion levers [like] more domains, connected accounts and enterprise add-ons,” he says.
AI companies require unprecedented business model flexibility and a billing system that scales. For early-stage, venture-backed founders, Stripe Startups is where to start. Enrol in the programme and receive access to credits on Stripe fees, expert insights and a focused community of other founders building on Stripe. Apply to Stripe Startups here.
