Anthropic’s preliminary second-quarter revenue topped $11.5B ahead of its IPO. Let that number sit for a moment without reaching for any adjective. It is simply what happened.
The AI bubble debate has become a full-time spectator sport. One camp sees froth and echoes of 2000. Another sees a paradigm shift comparable to the internet itself. Ed Zitron was on Squawk Box at 7:50am ET talking about the bubble, and the argument proceeded exactly as expected. Nobody is wrong on the surface. Nobody is right either, because the entire conversation is about the story instead of the conditions that produce the number.
The NVIDIA question is real but narrow. You can believe the GPU buildout is overdone and still acknowledge that Anthropic just generated $11.5B in a quarter. The bubble frame collapses the distinction between infrastructure spending and revenue generation. They are related but not the same. A company can have both a legitimate business and an overvalued stock. The two conditions can coexist. That insight is not welcome on either side of the debate because it ruins the certainty of the take.
This is the same cognitive error embedded in how we think about reading, creation, and any field where the output looks like magic. A recommendation algorithm appears to discover your taste. In reality it is predicting what you will buy based on what you already highlighted 65,000 times. The algorithm has no opinion. It is a mirror held up to the work you already did. The moment you stop creating the input, the mirror shows nothing. That is not a failure of the algorithm. That is how mirrors work. Most recommendation systems are built to predict what you will purchase, not what you will love. The distinction matters. One optimizes for the sale. The other depends on the signal you provide through your actual behavior.
The same logic applies to memorable events. You wait for the breakthrough, the viral moment, the call that changes everything. But the breakthrough is usually the visible output of invisible infrastructure. You created the conditions for it by showing up consistently, by shipping when no one was watching, by building relationships before you needed them. The event did not happen to you. You created the climate, even if you did not realize you were doing it at the time. The mistake is putting the cart before the horse. You cannot harvest the fruit without first tending the soil.
This is what the soil metaphor captures. Take care that you choose a good climate, not subject to destructive storms, and a soil that is naturally strong. In tech terms: choose a market with real demand, not just narrative momentum. Build unit economics that survive a rate hike. Sign power contracts before you need the power. These are not glamorous tasks. They are the things nobody tweets about at 2am. They are also the only things that survive a market turning.
Anthropic’s number matters because it is evidence that someone created those conditions. The IPO is not a bet on AI. It is a forced disclosure of revenue that already exists. The private market allowed the story to float free from the unit economics. The public market will not. That discipline is the point. An IPO forces a discipline that private markets never did. The companies that survive are those that built real unit economics in the private market phase rather than only having growth curves.
The bubble crowd will point to the valuation and say it cannot hold. The bulls will point to the growth curve and say this time is different. Both are stories. The revenue is the condition. You can argue with a story. You cannot argue with a number that already happened.
The companies that survive the transition are the ones that understood this early. They did not confuse the narrative with the work. They built the soil while everyone else was chasing the weather.
Create the input. Choose the climate. Tolerate the boredom of infrastructure. Ignore the headline. Watch the revenue compound.
The story will take care of itself. The soil will not.
Leave a Reply