The AI Three-Body Mess: Nobody Wins, But Somebody Will Capture All The Value

(SeaPRwire) –

By: Oliver Hawthorne

The AI economy is a three-body problem. You have the frontier labs: OpenAI, Anthropic, the big spenders. You have the open-weight models, mostly from China and a few US players. Then you have the application companies that stitch everything together for the end user. Each body pulls on the others. A small nudge from one swings the entire trajectory. Nobody gets to dictate the final orbit.

Right now, the system is unstable. The frontier labs have seen explosive demand. Anthropic is growing fast. But that growth comes with scrutiny. Alex Karp from Palantir called it “tokenmaxxing” in July. Companies are burning tokens with no real productivity gain. Spending on AI is now 0.5 to 1 percent of all US white-collar salaries. At that scale, people start asking questions. Meanwhile, competition at the frontier is getting real. Meta’s Muse Spark 1.1, xAI’s Grok 4.5, they are all fielding capable models. The gap is closing.

Then there are the open models. Zhipu’s GLM 5.2 and Moonshot’s Kimi K3 are now within striking distance of the frontier on several benchmarks. They cost a fraction. That creates momentum. US open models like Thinking Machines’ Inkling and Nvidia’s Nemotron 3 are not quite at the frontier yet, but they offer a domestic alternative. The application companies see this. They are ramping up efforts to build on open-weight models. They want lower costs and more control.

The real question is not whether AI pays off. It will. The question is who captures the value. The frontier labs are trying to go deeper into the product stack. They want to widen their moats and sustain high margins. The application companies are going deeper into the model stack. They want to build moats of their own. This convergence is rational. Software companies enjoy 70%+ gross margins. But something has to give.

I think the discomfort with frontier pricing will ease. Competition will push prices down. The returns on AI spend will start to show. Much of the anxiety is a timing mismatch. Adoption is running ahead of utility. For cars and cell phones, mass adoption followed price declines. For AI, adoption happened faster. The payoff will come. Faster growth for some, cost savings for others, higher productivity across the economy.

The shift toward a multi-model world will continue. Competition drives it. Real differentiation will help. US open-weight models will become genuine alternatives to the Chinese ones. They will have clearer business models. That makes it easier for customers to take long-term bets. Eventually, the distinction between open and closed will blur. The frontier labs will support model personalization for specific needs.

Three-body systems are chaotic. The equilibrium is unsettled. The noise around open vs. closed, China panic, and hand-wringing over returns looks temporary. The genuinely interesting question is not whether AI pays off. It is who captures the value: the labs at the frontier, the open models nipping at their heels, or the applications that own the customer.

Author bio: Oliver Hawthorne, a Principal Correspondent permanently stationed at an international technology review, covering the intersection of commercial strategy and infrastructure with a sharp, skeptical eye.