The firms holding billion-dollar stakes in both horses say the race has no finish line.

The Summary

The Signal

The smart money is hedging, and that tells you everything. T. Rowe Price partner Emma Norchet disclosed multi-billion-dollar stakes in both Anthropic and OpenAI, a positioning that would be absurd if this were a zero-sum game. It's not. The frontier AI market is starting to look less like search engine wars and more like cloud infrastructure, where AWS, Azure, and GCP all print money serving different customers with different needs.

Anthropic is making the clearest strategic shift. They're moving beyond their early win in coding assistants toward becoming what Norchet calls "an AI layer through which businesses can execute entire workflows." This is the difference between selling a hammer and selling a construction company. If they pull it off, Anthropic becomes infrastructure that other businesses build on top of, not just a tool that individual developers use.

"Anthropic is entering act two of its growth story, moving from coding tool to workflow execution platform."

Meanwhile, the consumer picture is weird. A16z partner Olivia Moore reports that OpenAI and Anthropic dominate consumer AI spending, but the adoption numbers are brutal. Fewer than 5% of consumers pay for even one AI product. The revenue comes from a small cohort of power users dropping up to $1,000 per month. That's not a mass market. That's a cult with credit cards.

This is where the enterprise and consumer stories diverge completely. On the enterprise side, you've got T. Rowe Price betting billions that multiple frontier models can coexist because different companies will optimize for different things: speed, accuracy, cost, compliance, vertical-specific training. On the consumer side, you've got Moore arguing the next wave won't come from better productivity tools. It'll come from social, entertainment, health, and what she calls "multiplayer experiences."

Current consumer AI adoption:

  • Under 5% of consumers pay for any AI product
  • Revenue concentrated in power users spending $200-1,000/month
  • Productivity use cases dominate current paid adoption

Translation: ChatGPT as a better search engine maxes out at maybe 20 million paying users globally. ChatGPT as the interface layer for a new category of social or entertainment experience could be 10x that. The companies that figure out how to make AI native to new experiences, rather than bolted onto old workflows, will own the consumer side of this market. OpenAI and Anthropic have the distribution to try. Neither has cracked the code yet.

The Implication

If you're building in AI, the frontier model layer is now a question of which infrastructure you pick, not whether one will kill the others. Anthropic's pivot to workflow execution means they're building for agents and automation, not just chat. That's the Web4 play. If you're waiting for a consumer AI IPO, Norchet says public market investors will have appetite when these companies choose to go. Translation: they're IPO-ready when they want to be, not when they need to be.

For the rest of us, watch where the next billion in consumer spend goes. It won't be people paying $20/month for slightly better email drafts. It'll be the first AI-native social network, game, or health app that people can't imagine living without.

Sources

Bloomberg Tech