Engineers in Silicon Valley have been raving about Anthropic’s AI coding device, Claude Code, for months. However not too long ago, the thrill feels as if it’s reached a fever pitch.
Earlier this week, I sat down with Boris Cherny, head of Claude Code, to attempt to perceive how the corporate is assembly this second.
“We constructed the best potential factor,” stated Cherny. “The craziest factor was studying three months in the past that half of the gross sales crew at Anthropic makes use of Claude Code each week.”
AI-powered coding has developed rapidly. From 2021 to 2024, most instruments functioned as little greater than autocomplete, suggesting a couple of strains of code as builders typed. By early 2025, startups like Cursor and Windsurf started rolling out early “agentic” coding merchandise, which let builders describe a characteristic in plain language and depart the remainder as much as an AI agent.
Claude Code launched round this time too. Cherny acknowledges that early variations of Claude Code typically stumbled, making errors or getting caught in expensive loops. Cherny says Anthropic constructed Claude Code for the place AI capabilities had been headed, reasonably than the place they had been at launch.
That guess was prescient. A number of builders declare AI coding merchandise reached an inflection point in latest months, notably across the launch of Anthropic’s newest AI mannequin, Claude Opus 4.5.
Kian Katanforoosh, an adjunct lecturer on AI at Stanford and the CEO of the startup Workera, says his firm not too long ago converted to Claude Code after testing a number of AI coding instruments internally. In the end, he says, Claude Code labored higher for his senior engineers than instruments from Cursor and Windsurf.
“The one mannequin I can level to the place I noticed a step-function enchancment in coding skills not too long ago has been Claude Opus 4.5,” says Katanforoosh. “It doesn’t even really feel prefer it’s coding like a human, you kind of really feel prefer it has discovered a greater means.”
Final yr, the enterprise of AI coding brokers took off. In November, Anthropic introduced that Claude Code had reached $1 billion in annualized recurring income, lower than a yr after its debut.
By the tip of 2025, Claude Code’s ARR had grown by at the very least one other $100 million, in accordance with an individual acquainted with the corporate’s financials. On the time the product accounted for roughly 12 p.c of Anthropic’s whole ARR, which stood round $9 billion. Whereas nonetheless smaller than Anthropic’s enterprise enterprise—which provides AI methods to complete companies—coding is among the firm’s fastest-growing segments.
Anthropic has additionally advised buyers it goals to be cash-flow constructive by 2028 and that Claude Code might play an vital function in its income progress. The corporate declined to touch upon its funds.
Whereas Anthropic feels dominant in AI coding, the thrill round Claude Opus 4.5 seems to be lifting a number of firms. Cursor, which lets customers code utilizing fashions from Anthropic and different AI labs, additionally stated its coding device reached $1 billion in ARR in November. In December, the corporate posted notably robust month-over-month income progress, in accordance with an individual near the corporate. OpenAI, Google, and xAI are additionally racing to assert a bigger share of the AI coding market, growing agentic merchandise of their very own powered by in-house AI fashions.
Now, Anthropic is making an attempt to make use of Claude Code’s momentum to create agents for non-coding sectors. Earlier this month, the corporate launched Cowork, an AI agent that may handle recordsdata on a consumer’s laptop and work together with software program—with out requiring any interplay with a coding terminal.
This interview has been edited and condensed for readability.
WIRED: There’s been pleasure round Claude Code for months. Why is it taking off now?
