The legendary F1 driver Ayrton Senna once said “You cannot overtake fifteen cars when it’s sunny weather, but you can when it is raining.”
The sellers of AI are preparing for one of the bloodiest and most capital-intensive fights in history. Every day, new models, chips, and infrastructure companies raise billions to out-train, out-compute, and out-scale each other. Margins are collapsing even before profits arrive. The arms race has begun, and the battlefield is crowded.
They’re assembling tremendous scale. They’re burning capital. They are looking for governments to backstop their infrastructure spending and are operating like too big to fail financial institutions. Their spending is so large that it is keeping parts of the economy from slipping into recession. They’re preparing for dominance in models, energy, chips, datasets, infrastructure and platforms.
Investors are rushing in trying to pick winners. On the surface it is obvious: you want “the next big platform for AI”. But look closer: margins are thin. The competitive pressure is overwhelming. Differentiation is limited or temporary.
For every one AI seller that emerges dominant, hundreds will die and burn an insane amount of every form of capital in the process.
Now let’s flip the frame. What if the real opportunity isn’t in building the AI or selling it, but in using it? In companies that buy the AI and use it to gain an advantage?
By the end of all of this capital burning and “the only risk is under-investing in AI,” there will be a few winners left standing. But there will be thousands of winners in the application of AI.
Just as in the internet era we came to say “every company is an internet company”, so in the future I expect we will say that every successful company is some form of applied AI company.
I personally like those odds and the opportunities in applied AI, much more than hunting for the seller of AI.
Infrastructure gets built often at enormous cost, most times with limited upside. Application and usage come after. The value often flows to those who use rather than those who build.
What looks exciting is the building, but what often wins is the usage, especially when the users have a unique edge.
So, with AI instead of asking which model will win or which chip company will dominate, ask this: which companies will harness this technology, embed it, scale the advantage, and ride the cost decline of intelligence and do so with proprietary data and distribution already in place? Who will become more efficient, more productive, more differentiated because of AI and because of what they uniquely bring?
If you focus on capturing applied AI’s impact, you can sidestep the hyper-competitive hype cycle and instead focus on the buyers who will reap benefits as that infrastructure commoditises. But more than that: focus on the buyers who have data moats and distribution moats.
