Vercel aims to be the AWS of the AI agent era

Vercel, a cloud infrastructure company, is experiencing significant scale in its daily operations. The company, which provides cloud infrastructure that allows developers to deploy agents without managing servers, currently sees 6 million deployments a day. Additionally, more than 1 trillion tokens flow through the company’s AI gateway daily. Following its ShipNYC conference last week, Vercel CEO Guillermo Rauch argued that the industry is reaching a critical architectural turning point. Rauch stated that the sector is currently deciding whether the model and the agent will remain coupled. This shift follows a transition from last year’s prototyping phase to a focus on running AI agents securely in production.

Rauch is positioning Vercel to serve as the infrastructure layer that allows developers to separate their AI models from the agents running them. This decoupling allows companies to swap out models based on price and performance, utilizing providers like OpenAI, Anthropic, Gemini, or models from DeepSeek and GLM-5.2. By keeping these layers separate, developers can avoid being locked into a single AI lab’s ecosystem. Rauch frames this open approach as central to the company’s long-term strategy. According to Vercel CEO Guillermo Rauch, “We’re going to be the AWS of this generation, so obviously we’re fighting for a world of open protocols.” AWS, or Amazon Web Services, is the dominant cloud provider.

As AI agents become more integrated into software development, security and data control have emerged as primary challenges. Rauch highlighted the risks associated with coding IDEs—integrated development environments used by software engineers—such as Devin or Cursor. He noted that without proper sandboxing, these tools can expose proprietary code. Rauch warned that a real risk of AI that I always think about is, when you get a coding IDE like Devin or Cursor, if you’re in the wrong setting, they may train on your entire codebase. He noted he had discussed this risk with the president of aerospace company Airbus. To mitigate this, Vercel is promoting sandboxed environments to control data access and prevent sensitive codebases from being uploaded to external clouds for model training. This focus on data control also extends to internal corporate agents, which companies are deploying to bypass traditional bottlenecks like waiting for a Q1 sales dashboard project to complete.

Why it matters

Vercel is positioning itself as the ‘AWS of this generation’ by providing infrastructure that allows developers to decouple models from agents, directly competing with AI labs that seek to keep users within their own enclaves.