ChatGPT is no longer OpenAI's most important product. Here's why.

In the evolving landscape of artificial intelligence, the focus is shifting from simply improving model performance to fostering customer loyalty. Samuel Colvin, CEO of the AI startup Pydantic, emphasizes that the priorities for companies like OpenAI and Anthropic are changing as they prepare for potential public offerings. "A year ago, their main concern was revenue, but now profit margins have taken center stage," Colvin noted in a recent discussion. The challenge lies in the fact that continuously enhancing model quality is an expensive endeavor, with leading labs investing billions in training advanced models that competitors can quickly replicate. This approach does not guarantee sustainable profits. Colvin argues that both OpenAI and Anthropic are increasingly focusing on developing products that create a sense of dependency among their users, independent of model quality. Tools like Claude Code and Codex are emerging as more profitable ventures than traditional chatbots, allowing developers to utilize large volumes of tokens for complex projects, which drives higher usage and revenue. These advanced coding tools not only enhance software development speed but also contribute to codebases that can exceed what human developers can efficiently manage. This dependency could explain the drive behind OpenAI and Anthropic's efforts to transform Claude Code and Codex into comprehensive AI-powered platforms rather than merely standalone coding solutions. For instance, Anthropic is advancing with its Cowork initiative, while OpenAI plans to integrate Codex with ChatGPT. However, there is a counter-movement among customers. Walmart, for example, has created its own coding assistant called Code Puppy to minimize reliance on any single AI provider. This system can interchange between models from various companies, including OpenAI and Anthropic, offering Walmart greater control over its coding processes and costs. This dynamic creates a tension in the AI market. On one side, OpenAI and Anthropic aim to build high-margin, sticky products to keep customers engaged within their ecosystems. On the other side, enterprise clients are seeking flexibility and cost efficiency. The companies that will thrive in the next phase of AI development will likely be those that can adeptly manage this conflict.