Dario Amodei’s prediction from last year of AI writing 90% of code appears to have largely come true, and has an even more dramatic prediction to make at the beginning of 2026.
Speaking about the trajectory of AI development, the Anthropic CEO outlined a vision of recursive AI improvement and its implications for software engineering. Amodei described a feedback loop where AI models capable of coding and AI research would accelerate their own development, while also revealing that some engineers at his company have already stopped writing code themselves. Most remarkably, he suggested the timeline for AI handling the entire software engineering workflow end-to-end might be closer than many expect.
“The mechanism whereby I imagined it would happen is that we would make models that were good at coding and good at AI research, and we would use that to produce the next generation of model and speed it up to create a loop that would increase the speed of model development,” Amodei explained.
The theoretical framework is already manifesting in practice at Anthropic. “We are now in terms of the models that write code—I have engineers within Anthropic who say, I don’t write any code anymore. I just let the model write the code. I edit it, I do the things around it,” he noted.
Then came the prediction that will likely reverberate through the tech industry: “I think we might be six to 12 months away from when the model is doing most, maybe all of what software engineers do end-to-end, and then it’s a question of how fast does that loop close?”
Amodei was careful to temper expectations about the speed of subsequent developments, however. “Not every part of that loop is something that can be sped up by AI, right? There’s chips, there’s manufacturer of chips, there’s training time for the model. So I think there’s a lot of uncertainty. It’s easy to see how this could take a few years.”
The implications of Amodei’s timeline are profound, suggesting a fundamental transformation of software development within the year. This isn’t merely about AI assistance or copilots as it has been so far—it’s about AI systems independently handling the full spectrum of engineering tasks. His comments align with emerging evidence from the field: Anthropic’s lead engineer recently revealed that 80% of Claude Code’s code is written by Claude Code itself, while the tool’s creator shared he didn’t open an IDE all of last month, relying entirely on Claude Code for his work. Node.js creator Ryan Dahl went further, declaring that “the era of humans writing code is over.” While Amodei acknowledges significant constraints—chip manufacturing, training infrastructure, and physical hardware production cannot be recursively improved at software speeds—his six-to-twelve-month window for comprehensive AI software engineering capabilities represents one of the most aggressive timelines proposed by a major AI lab CEO. If accurate, the tech industry may be on the cusp of its most disruptive transformation yet.