The scene starts on a weekday evening in New York. A commuter leaves Union Square, hops on the subway, and pulls up a phone app that can write code for him. He types a simple request: “Show me the data I just uploaded, put it in a database, and make it searchable on a web page. ” While the train rattles through tunnels with no signal, the request is still being processed. When the train reaches the Manhattan Bridge and regains connectivity, he sees the finished product on his screen: a website, a music‑app feature, or even a small game. By the time he reaches Brooklyn, the whole project is usually done.

This workflow is known as “vibe coding, ” a term coined by AI researcher Andrej Karpathy. It isn’t traditional programming; it’s about telling a chatbot what you want and letting the AI work out the details. Popular tools for this include Claude Code from Anthropic, OpenAI’s Codex, and Google Gemini. Claude Code alone brought in a billion dollars for its company in just six months. In November the tool’s performance improved dramatically, and since then it has helped many people finish side projects that had been dormant for years. The result is a steady stream of new software, often completed in about half an hour of human effort and an hour for the AI.

The improvement is surprising. Earlier AI coding assistants were slow and awkward, but now they can run for a full hour and produce functional websites or apps that may not be perfect, yet are credible enough to use. This shift has sparked intense discussion among tech enthusiasts and professionals alike.

The broader question is how this affects the software industry, which has long been a stable middle‑class occupation built on skill and craftsmanship. If millions or even billions of people can create software with a few prompts, what does that mean for the future of coding jobs? Some see it as an unprecedented opportunity, while others worry about a decline in demand for human developers.

Market reactions have been swift. Major software companies—such as Monday. com, Salesforce, and Adobe—saw their stocks drop sharply, and the Nasdaq 100 lost half a trillion dollars in two days. Legal‑tech firms also suffered after Anthropic released tools that automate legal tasks, and financial services companies faced similar pressure. Investors fear that legacy software may become obsolete when AI can code anything quickly and cheaply.

These concerns feel premature to many, but markets often react before the full picture emerges. The idea that AI could replace human developers is unsettling, especially when someone watches a large language model solve complex problems—like migrating data from an old system to a new one—in seconds. The shift feels seismic.

From a practical standpoint, the cost savings are clear. A former software‑services CEO estimated that redoing his own website would have cost $25, 000 if done by a human. A friend’s request to clean and visualize a massive dataset could have cost $350, 000 in traditional consulting. AI can now perform these tasks for a fraction of the price.

The rapid rise of AI coding tools is reshaping how software is built and who can build it. Whether this leads to widespread job displacement or new opportunities remains to be seen, but the conversation is already underway.

AI Coding Turns Work Into Play

Actions