Vercel’s CEO Guillermo Rauch emphasized a growing trend where companies abandon exclusive partnerships with one AI lab, instead integrating multiple providers to handle different elements of their AI operations. This shift reflects a deeper understanding of the AI stack’s modular nature, where components such as the model, data platform, sandbox, and gateway are treated as interchangeable parts.
Rauch highlighted how businesses now view AI tools as “plug and play,” allowing them to mix offerings from OpenAI, Anthropic, Gemini, and emerging Chinese models like DeepSeek and Z.ai’s GLM-5.2. Gemini, in particular, is gaining traction for its cost-effective performance at scale, illustrating how price and efficiency have become as critical as sheer capability.
This evolution follows a year marked by widespread experimentation and prototyping of AI agents. Companies have moved beyond initial enthusiasm to confront practical challenges of deploying AI in production environments. Rauch noted that the industry is transitioning from a phase focused on innovation to one centered on operational realities and value delivery.
Vercel itself, known for its cloud platform that aids developers in hosting and launching applications, has observed this industry-wide recalibration. The growing complexity and expense of AI have led firms to rethink blanket approaches that squander resources. Instead, they seek more nuanced strategies to optimize costs and efficiency.
Similar approaches have emerged elsewhere. Coinbase CEO Brian Armstrong has publicly experimented with cheaper, non-U.S. AI models such as GLM-5.2 and Kimi AI’s K2.7, routing tasks to the most suitable model rather than defaulting to high-cost frontier systems for every request. This tactical use of multiple AI providers echoes how enterprises previously adopted multi-cloud strategies to avoid vendor lock-in and control expenses.
This multi-lab partnership model reflects a recognition that no single AI provider can deliver all capabilities equally well or cost-effectively. Instead, organizations build layered AI environments tailored to specific needs, maximizing both innovation and budget management.