The web is becoming a sprawling hive mind of AI agents, and Vercel wants to build a cloud to host them (www.businessinsider.com)

Vercel CEO Guillermo Rauch announced the company is building an "AI cloud" to host a coming wave of autonomous AI agents—software that plans, reasons and acts over minutes, hours or days rather than delivering single-page responses. Vercel is creating new frameworks and developer tools to treat the agent as the cloud’s primary primitive, and has been training agents internally (its COO says a “lead agent” replaced a 10-person sales development team). The company raised $300 million at a $9.3 billion valuation to fund this push, signaling a bet that web apps will shift from static pages to always-running, AI-native services that consume tokens, persist state, and require continuous compute and orchestration. For the AI/ML community this matters because agents change the infrastructure and engineering requirements: long-lived runtimes, state management, secure backend integrations, inter-agent communication, billing models for sustained token/compute use, and new testing and safety regimes. Vercel’s effort joins broader industry moves—OpenAI’s Atlas browser, Google’s Agent2Agent protocol, Anthropic’s Model Context Protocol, and even Cloudflare’s stablecoin plan—pointing to emerging standards for agent coordination, access to databases and workflows, and transaction rails. Practitioners should expect demand for agent orchestration platforms, persistent memory systems, secure connector patterns, and evaluation tools for long-horizon behavior and safety.