You’ve probably noticed the breathless headlines every time a new AI model drops. We’re told that benchmarks and parameters are the ultimate currency of this new gold rush. But if you look past the PR spin, you’ll see a terrifying reality: the current valuations of both OpenAI and Anthropic are built on sand.
The dirty secret of the AI industry is that foundation models are rapidly becoming commodities. The real differentiation is just price, and whoever wins the price war ultimately loses the war. AI models are becoming commodities. The real differentiation is price, and whoever wins the price war loses the war.
Think about the enterprise space. You might assume companies are ditching their legacy tech for shiny new AI startups. They aren’t. Sticking with Salesforce and bolting on its new AI features is infinitely simpler than paying the massive switching costs to adopt an AI-first startup that might not even exist in three years.
So where does that leave Sam Altman and OpenAI? It leaves them in a state of desperate, underdog panic. They are burning cash on multiple fronts, trying to build a defensible moat before they are forced into an IPO. Because if they go public without a real, locked-in competitive advantage, public market investors will cut their throats. Public market investors don’t reward science experiments; they cut throats.
This is exactly why OpenAI is pivoting to hardware. It looks like they’re doing too much, but it’s a survival mechanism. They realize they cannot beat Apple in software ecosystems—Apple’s double-sided lock-in is simply too strong. But Apple has a glaring vulnerability: they don’t have their own foundation models. Apple Intelligence is a Frankenstein experience compared to a device built AI-first from the ground up.
Apple has the hardware but lacks the brain. OpenAI has the brain but no body. The race is to see who builds the missing half first.
OpenAI’s only viable long-term strategy is to build a vertically integrated hardware business. By owning the device and the model, they create the switching costs that software simply cannot. They are targeting the massive chunk of Android users who aren’t already locked into Apple’s ecosystem. It’s a massive, capital-intensive gamble, but it’s the only play that makes sense.
And what about Anthropic? Their coherent focus on enterprise might look smart in the short term, but it’s a dead end. The cost of paying Anthropic is too high to justify any real customer captivity. Soon, enterprises will realize they can get maximum leverage by training cheap open-source models on their own proprietary data. Anthropic’s sky-high revenue is a mirage built on short-term desperation, and their fall will be brutal.
If you want to track the real trajectory of AI, stop obsessing over model quality. The battle for AI supremacy won’t be won by whoever writes the best code. It will be won by whoever builds the box it runs on.
FAQ
Q: Why can't OpenAI just win by having the best AI model?
A: Because AI models are becoming commodities. When the only differentiator is price, margins collapse. Without hardware or ecosystem lock-in, there are no switching costs to keep users loyal.
Q: What's the practical implication for enterprise buyers?
A: Stop overpaying for closed-model APIs. The long-term value lies in training cheap, open-source models on your own proprietary data, rather than renting intelligence from startups that might not survive the next capital cycle.
Q: What's the contrarian take on Apple's position?
A: Apple's ecosystem lock-in is impenetrable, but their lack of a proprietary foundation model is a massive blind spot. OpenAI can disrupt them not by fighting the software war, but by offering an AI-first hardware experience Apple cannot replicate.