Anthropic’s Claude Cowork spooks markets over SaaS disruption

Claude Cowork and similar automation products highlight why investors in traditional SaaS and professional services are suddenly nervous: these tools don’t just assist workers, they increasingly perform end‑to‑end workflows that looked like defensible recurring‑revenue businesses. If enterprises can point a model at their own document stores and code repositories and have it manage projects, summarize research, draft contracts or perform first‑pass analysis, they need fewer external subscriptions for those functions.

In the race to AGI, this is a preview of how value may migrate. Frontier labs that successfully productize agentic capabilities will not only sell APIs; they’ll directly cannibalize parts of the software stack that sat above them. That creates a feedback loop: more revenue and usage for the most capable models, more data on complex workflows, and more capital to invest in pushing capabilities further.

For incumbents, the choice is stark: either deeply integrate frontier models into their offerings and accept margin compression, or risk being re‑intermediated by AI platforms that sell directly to their customers. The market’s reaction to Claude Cowork—whether overdone or not—shows that investors are starting to price in that possibility.