Cloud infrastructure platform Vercel is witnessing a major shift in how enterprises deploy AI, moving away from vendor lock-in to optimize performance and costs. According to insights shared by CEO Guillermo Rauch after the ShipNYC conference, the tech world is entering an era of putting agents into production rather than just prototyping. This transition is driving intense competition between infrastructure platforms like Vercel and major AI labs.
Diễn biến chi tiết
According to Rauch, last year was about prototyping, while today’s focus is on bringing applications into production. Vercel currently powers about 6 million deployments a day, half of them triggered by coding agents. Additionally, more than 1 trillion tokens flow through the company’s AI gateway daily. The Vercel CEO pointed out two killer use cases for agents today: coding assistants and internal corporate agents that optimize operations, allowing employees to access data intuitively without bottlenecks.
Bối cảnh & Nguyên nhân
The rise of agentic systems introduces significant security risks regarding corporate data leakage. For instance, developers using AI-powered IDEs might unintentionally upload proprietary source code to the cloud for training. To address this, Vercel introduced the Eve framework to set agent instructions in natural language, paired with Vercel Sandbox to enforce strict data control. This setup allows enterprises to leverage AI intelligence while preventing sensitive data from leaving the secure sandbox.
Phân tích kỹ thuật & Công nghệ
In the current tech landscape, developers' model preferences are shifting from single-lab partnerships to diversification. Instead of relying solely on OpenAI or Anthropic, companies are now adopting a plug-and-play approach with models like Google’s Gemini for its price-to-performance ratio, alongside open models like Deepseek and GLM-5.2. This transition requires a modular architecture where the model, harness, data platform, and sandbox operate independently.
Ý kiến chuyên gia & Nhận định
CEO Guillermo Rauch believes the industry is deciding whether the model and the agent will remain coupled. He stated, "We’re fighting for a world of open protocols," where users can get modules and building blocks from different providers to build on top, much like traditional software engineering. Rauch envisions Vercel as the AWS of this generation, competing with the closed ecosystems of major AI labs.
Tác động & Tương lai
Decoupling AI models from agent applications will reshape the global software market, forcing SaaS giants to open up their locked data. For the tech community, this trend offers maximum flexibility, lower operating costs, and enhanced data security. This is a positive signal for businesses and software engineers globally, enabling them to optimize costs by combining various AI solutions rather than relying on a single provider.