OpenAI buys up companies, Anthropic withholds a model — the AI divide between insiders and everyone else is widening

The gap between AI insiders and everyone else is rapidly turning into a chasm. This is visible in spending, in distrust, and even in vocabulary: people at the center of the AI industry are already speaking in a language incomprehensible to most. OpenAI has switched into aggressive acquisition mode.

The company consistently absorbs assets far beyond its original domain: financial applications, entertainment content, media projects. The acquisitions look disparate—but they trace a logic: building an ecosystem where OpenAI's AI infrastructure becomes the invisible foundation of several industries simultaneously. The company is constructing not a product, but an infrastructure layer.

In parallel, a demonstrative episode unfolds with rebranding: a famous shoe manufacturer announced a relaunch of itself as an AI infrastructure company. This is not an isolated quirk—it's a symptom. The word "AI" in a pitch deck or press release today works the same way "internet" did in the late 1990s or "blockchain" in 2017.

Any company that adds this label instantly gets revalued—regardless of actual content. Anthropic, meanwhile, took an atypical step: the company publicly acknowledged developing a model it considers too powerful for open release. This is rare in an industry where competition for public demos has become the norm.

What exactly makes the model "too powerful"—is not disclosed. But the fact of such a statement itself signals a shift: inside laboratories, there is growing recognition that certain capabilities require a different governance mode, not just a scheduled release. Against all this, what analysts are beginning to call the AI Anxiety Gap grows—the gap in anxiety and understanding between those working with AI from within and the broader audience.

Insiders operate with new concepts like "tokenmaximizing" (extracting maximum value from a model's context window), discussing architectural tradeoffs and safety thresholds. Outside—a growing mix of admiration, skepticism, and a sense that the train has already left. This gap is neither accidental nor temporary.

It is structural. The pace of change in the AI industry outpaces the speed at which society manages to form a common language for discussing what's happening. OpenAI's acquisitions, Anthropic's hidden models, and rebranding from shoe companies into AI players—all are symptoms of one process: the AI industry is already living in the next decade, while most people are still making sense of the current one.