Engineers in silicon Valley has been raving about Anthropic’s artificial intelligence coding tool, Claude Code, for months. However, lately the hype seems to have peaked.
Earlier this week, I met with Boris Cherny, head of Claude Code, to try to understand how the company is dealing with this moment.
“We built the simplest thing possible,” Cherny said. “The craziest thing was when I found out three months ago that half of the sales team at Anthropic uses Claude Code every week.”
AI-based coding has evolved rapidly. Between 2021 and 2024, most tools only worked in autocomplete mode, suggesting a few lines of code as developers typed. In early 2025, startups like Cursor and Windsurf began rolling out early “agentic” coding products, which allowed developers to describe a feature in plain language and leave the rest to the AI agent.
Claude Code also launched around this time. Cherny admits that early versions of Claude Code often stumbled, making mistakes or getting stuck in pricey loops. Cherny says Anthropic built Claude Code where AI capabilities were headed, not where they were at launch.
This bet was prophetic. Several developers claim that AI coding products have plateaued inflection point in recent months, especially in connection with the launch of Anthropic’s latest artificial intelligence model, Close job 4.5.
Kian Katanforoosh, an assistant professor of artificial intelligence at Stanford University and CEO of the startup Workera, says his company recently switched to Claude Code after internally testing several AI coding tools. Ultimately, he says, Claude Code worked better for his senior engineers than Cursor and Windsurf.
“The only model I can point to where I have recently noticed an improvement in step functions in terms of coding skills is Claude Opus 4.5,” says Katanforoosh. “It doesn’t even feel like he’s programming like a human, it feels like he’s found a better way.”
The activity of AI coding agents has gained momentum in the last year. In November, Anthropic announced that less than a year after its debut, Claude Code had achieved $1 billion in annual recurring revenue.
By the end of 2025, Claude Code’s ARR has grown by at least another $100 million, according to a person familiar with the company’s finances. At the time, the product represented approximately 12 percent of Anthropic’s total ARR of approximately $9 billion. While still smaller than Anthropic’s corporate business, which provides AI systems to entire corporations, coding is one of the company’s fastest-growing segments.
Anthropic also informed investors that it aims to be cash flow positive by 2028 and that Claude Code can play an significant role in the company’s revenue growth. The company does not want to comment on its finances.
While Anthropic seems to dominate AI coding, the buzz around Claude Opus 4.5 seems to be buoying several companies. Cursor, which allows users to code using models from Anthropic and other AI labs, also reported that its coding tool hit $1 billion in ARR in November. According to a person close to the company, the company recorded particularly sturdy month-on-month revenue growth in December. OpenAI, Google and xAI are also racing to capture a larger share of the AI coding market by developing their own agent products based on internal AI models.
Currently, Anthropic is trying to build on Claude Code’s momentum to create agents for non-coding sectors. Earlier this month, the company launched Cowork, an artificial intelligence agent that can manage files on a user’s computer and interact with software – without requiring any interaction with a coding terminal.
This interview has been edited and condensed for clarity.
WIRED: There has been excitement surrounding Claude Code for months. Why is he running now?