Anthropic is on a different trajectory than its better-known competitor. While OpenAI chases consumer dominance through ChatGPT, a brand so strong it has become synonymous with the category itself, Anthropic is quietly building the infrastructure that matters: direct file access, local computation, and deeply integrated workflows. The $10 billion funding round at a $350 billion valuation makes this clearer than ever. This is a company that multiplied revenue tenfold year-over-year, focused ruthlessly on enterprise use cases where the real money lives.
The tactical shift this week—blocking access to Claude through third-party platforms like OpenCode while launching Claude Cowork—signals a strategic choice. Rather than compete on interface, Anthropic is betting on integration. Cowork removes the friction of copy-pasting between a chat window and your actual work. A researcher analyzing datasets, a consultant building models in Excel, an academic writing in LaTeX: these aren't niche users. They're the majority of knowledge workers, and they've been locked out of AI's practical benefits because existing tools assume you work in a browser.
This is where the real market expands. Coding agents like Claude have already reshaped how developers work—they've become supervisors of an army of agents writing code, validating output, steering direction. The same leverage applies to any knowledge work that touches files: a consultant analyzing regulatory documents, a doctor reviewing patient records across years of health data. But the command-line interface has been a barrier. Cowork removes it.
Meanwhile, both OpenAI and Anthropic launched health-focused products within weeks of each other—ChatGPT Health and Claude Health—integrating wearable data, medical records, and AI reasoning into a single interface. This isn't about replacing doctors. It's about giving them (and patients) supervision tools over vast datasets no human can fully process. The model becomes a research partner that notices what humans miss, flags anomalies in historical data, and suggests which gaps need filling. That's valuable whether you're a physician optimizing diagnosis or a researcher mining years of quantified self-data for patterns.
The real tension isn't between AI and employment—it's between those who learn to work with these tools and those who don't. Demis Hassabis, DeepMind's leader, put it plainly at Davos: new graduates should become "super proficient" at AI tools. Not because jobs are disappearing overnight, but because the productivity gap between skilled and unskilled users will become the defining stratification of the next decade.