Stripe CEO Patrick Collison recently shared a fascinating observation on X: “We analyzed the AI sector on Stripe, and found that not only are AI-native businesses being built in large numbers, but that they're actually growing meaningfully faster than the fastest-growing SaaS antecedents.”
Stripe pulled data on its 100 fastest-growing AI customers and compared their revenue growth to 100 of the most promising SaaS companies from 2018, the year widely considered the beginning of the SaaS era. Some of Stripe’s AI customers include the hottest names in AI such as Anthropic, OpenAI, Mistral, GitHub and Midjourney.
Comparing the momentum of companies in the early years of each era, Stripe’s data revealed that AI startups took a median of 11 months to reach $1M of annualized revenue following first customer sale compared to nearly 15 months for the SaaS cohort.
And AI startups scaled from $1M to $30M at a rate five times faster than SaaS companies.
Those are exciting numbers — but we have to remember that these are very different businesses. “Unlike past generations of software companies, AI companies pay substantial compute costs out of the gates, so are under pressure to build monetization faster,” Stripe’s Emily Sands pointed out in a recent Financial Times article.
Comparing AI and SaaS business models
In addition to the higher initial compute costs, there are other differences between early-era AI and SaaS business models that have influenced the growth rates.
Revenue structure. AI companies often charge based on usage (per task, per API call, etc.), where SaaS companies traditionally charged a subscription fee (per license, per seat, etc.). However, this is beginning to change with AI technology being built into SaaS applications today — more SaaS companies (Salesforce, for one) are now moving in the direction of charging per task.
Technical focus. AI companies are heavily focused on data acquisition, data quality, model training and algorithm refinement to improve the immediate usefulness of their products. SaaS companies, on the other hand, were more focused on ongoing user experience optimization in the early category era, because customer retention was a key revenue driver.
Integration and embedding. AI products are more easily embedded into customers’ existing technology solutions. SaaS products are typically more self-contained, so while they may integrate into customers’ existing tech stack, embedding requires extensive customization.
These differences impacted the choices AI startup founders made around where they spent investment dollars — but now they are also causing investors to question the long-term profitability of AI products. Simply put, expenses are higher for AI companies … and that adds up over time. It also creates a barrier to entry.
That said, AI products can improve more rapidly than SaaS through continuous learning, which is keeping AI appeal and demand high across the globe.
AI vs Saas: Not apples to apples
Though Stripe’s data attempts to make an apples-to-apples comparison of business model performance in the early years of the AI and SaaS eras, it’s ultimately not a fair fight. The businesses themselves differ greatly — and AI has the advantage of its era beginning post-COVID, when working fully digitally and remotely had become the norm.
