"AI Programming" Milestone: Claude Code "Reshapes" the Global Software Industry

Anthropic’s AI coding tool, Claude Code, is leading the global software development industry into a critical turning point. With its latest “agentic” capabilities, the product is not only radically changing the way code is written but also achieving explosive growth in commercialization.

According to WIRED on January 22, Boris Cherny, head of Anthropic Claude Code, revealed in an interview that, with foundational model capabilities leaping forward, software development is shifting from manual code-writing to being fully delegated to AI. This transformation is reshaping engineers’ workflows and the productivity logic of the entire industry.

Reportedly, insiders say that Claude Code’s annual recurring revenue (ARR) grew by a further $100 million by the end of 2025, after announcing in November that ARR had exceeded the $1 billion mark. Currently, Claude Code accounts for 12% of Anthropic’s total ARR of approximately $9 billion, making it one of the company’s fastest-growing business segments.

This breakthrough is mainly attributable to the launch of Anthropic’s latest AI model, Claude Opus 4.5, which many developers see as a “step-function” improvement in programming ability. Unlike earlier tools that could only complete code, Claude Code based on the new model can understand natural language instructions and autonomously complete subsequent development. This “agentic” trend has prompted startups like Cursor and Windsurf, as well as tech giants such as OpenAI, Google, and xAI, to accelerate their moves, competing for shares in this emerging market.

As AI agent capabilities mature, Anthropic is attempting to replicate this transformative model in the software industry into broader domains. The company has recently launched a new product called Cowork, aiming to extend AI agent functions from programming terminals to general office scenarios like file management and software interaction, seeking to spark a similar productivity revolution outside the coding sphere.

From Auxiliary Tool to Core Productivity

The evolution of AI coding tools has outpaced expectations. According to Wired, from 2021 to 2024, most tools could only provide autocomplete features, suggesting a few lines of code as developers typed. By early 2025, startups like Cursor and Windsurf began launching early “agentic” programming products.

Claude Code head Boris Cherny said in an interview that Anthropic bet on the future direction of AI capabilities in its product designs, rather than the actual level at the time. Cherny said:

“We built the simplest thing. The craziest thing was learning three months ago that half of Anthropic’s sales team were using Claude Code every week.”

This strategy proved visionary. Cherny revealed that the proportion of code he wrote using Claude Code rose from an initial 5% to 30% in May last year with the release of Opus 4 and Sonnet 4, and in the two months following the launch of Opus 4.5, 100% of his code was written by Claude Code.

It is reported that after testing several AI coding tools, Workera ultimately chose Claude Code. Katanforoosh stated that for the company’s senior engineers, Claude Code outperformed competitors like Cursor and Windsurf.

Agentic Model Changing Ways of Working

The core advantage of Claude Code is its agentic working mode. Cherny explained that, unlike a year ago when users mainly used the chat function, Claude Code and Cowork have truly achieved agentic operation, enabling tool usage, system file reading, and interaction with Slack and Google Sheets.

“It’s the golden age for the easily distracted,” Cherny said. The most efficient Claude Code users run multiple tasks simultaneously with Claude working on each, then check progress in turn. Cherny himself typically runs five to ten agents at once in terminals, mobile devices, and web interfaces.

This way of working is spreading internally at Anthropic. At internal review meetings before external product launch, CEO Dario Amodei once asked whether employees should be required to use the product. In reality, nearly 100% of Anthropic technical staff frequently use Claude Code, and 95% of the team’s code is written by Claude Code.

Anthropic’s enterprise clients also exhibit similar usage patterns. Cherny noted that enterprise clients’ needs for security and product interaction are very similar to Anthropic’s own, allowing the company to optimize the product via internal use.

Expansion into Non-Programming Domains

Building on Claude Code’s success, Anthropic launched Cowork this month, an AI agent product for non-programmers that can manage files on users’ computers and interact with software, without touching the code terminal.

Cherny described Cowork as “Claude Code for non-programmers.” He himself already uses Cowork for project management, such as having it check whether engineers have filled out progress reports and reminding those who haven’t via Slack.

“AI agents will be able to handle all the fiddly things in life. This has already been realized in engineering this year, and I believe it will happen in all other fields. Agents will be able to handle tasks like filling in forms, moving data in different places, sending emails, etc.”

Cherny acknowledged that such a transformation would bring disruptive impacts, requiring the industry to respond collectively. However, he believes it will make work more enjoyable, freeing people from tedious tasks.

Intensifying Competitive Landscape

Claude Code’s success is attracting more competitors to the AI coding market. Cursor announced annual recurring revenue of $1 billion in November last year. The platform allows users to code using models from Anthropic and other AI labs. According to insiders, Cursor achieved especially strong month-on-month revenue growth in December.

OpenAI, Google, and xAI are also accelerating development of agentic coding products based on their own AI models, seeking to capture bigger market shares. The core of this competition lies in the foundational AI models, and the release of Claude Opus 4.5 seems to give Anthropic a temporary edge.

For engineers experiencing this transition, Cherny offered a historical perspective. His grandfather once programmed with punch cards in the Soviet Union, after which programming evolved through machine code, C, Java, Python, and so on.

“As an industry, we’ve constantly been through transformations -- it’s an ever-increasing abstraction. I believe agents are just another point on this continuum.”

Cherny predicts that as the learning curve smooths, effective use of tools like Claude Code and Cowork will become easier and easier. Anthropic plans to achieve positive cash flow by 2028, with Claude Code, as one of the fastest-growing segments, playing a key role in this goal.

Risk Warning and DisclaimerThe market carries risks, and investments should be made cautiously. This article does not constitute personal investment advice, nor does it take into account the unique investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article suit their particular circumstances. Any investment based on this is at your own risk.