Anthropic and OpenAI might be building the future, but someone else may own it

Wall Street is preparing for what could be the three largest initial public offerings in history, with Anthropic and OpenAI set to follow SpaceX's lead. Despite generating significant revenue, both companies are facing doubts over their ability to generate enough value to justify their massive investments. Anthropic lost over $8 billion last year, while OpenAI expects to see nearly tenfold growth in revenue, despite projecting a substantial cash burn.

However, the economics of the AI race may not be as promising as they appear. Frontier AI labs may not capture the majority of the value generated by their breakthrough technologies. Instead, those who build complementary assets, such as products, platforms, and services, may reap the greatest benefits.

History provides examples of this phenomenon. During the early 20th century, electricity transformed numerous industries, but it was manufacturing companies like Ford that reaped the largest productivity gains from its use. Similarly, the internet's long-term value accrued to businesses that built services on top of it, such as search engines, e-commerce, and social media platforms.

Frontier AI is becoming infrastructure, not software, with leading labs investing heavily in AI chips, data centers, electricity, and cloud infrastructure. The models themselves are becoming a bedrock that others capitalize on, much like electricity or the cloud.

Moreover, AI models are becoming increasingly commoditized, making technological leadership harder to defend. The companies that control the complementary assets, such as Amazon, Alphabet, Meta, Microsoft, and Oracle, already hold a significant amount of cloud infrastructure, software, and customer relationships. This means that the value generated by AI may ultimately flow to these hyperscalers rather than the labs developing the models.

Still, a glimmer of hope exists for businesses that can leverage AI to reinvent themselves and gain a competitive advantage. Specialist companies with deep knowledge of their customers, products, and markets may have more to gain from AI than the companies building the models. The real value may lie in those that can successfully adapt to the AI-driven landscape, rather than simply inventing the technology.