Stripe President John Collison said coding models have become advanced enough to significantly accelerate the rate of AI improvement [1].
This shift suggests a feedback loop in technological development. When AI tools can write and optimize the code used to build future AI, the pace of innovation may move beyond human-led manual programming.
Collison discussed the trend during an interview with Bloomberg Television [1]. He said, "The coding models have gotten so good that they are now contributing in a significant way to the rate of improvement of AI itself" [1].
Stripe has integrated these capabilities into its own growth strategy. The company has executed 288 AI-related product launches as it builds out economic infrastructure for the sector [3].
Beyond technical development, the company has maintained a high market profile. In a tender offer discussed during a Feb. 24, 2026, interview with CNBC, Stripe was valued at $159 billion [2].
Collison's observations highlight a transition where AI is no longer just a tool for end-users, but a primary driver of its own architectural evolution. This capability allows developers to iterate faster and solve complex software engineering problems that previously required extensive manual labor.
“The coding models have gotten so good that they are now contributing in a significant way to the rate of improvement of AI itself.”
The emergence of AI-driven AI development indicates a potential transition toward recursive improvement. If coding models can autonomously optimize the frameworks that power them, the industry may see an exponential increase in capability gains, reducing the reliance on human programmers for baseline architectural updates.
