Every vibe coder is standing on top of a bet Jensen Huang made years before "AI app" was a phrase anyone said out loud. Nvidia spent a decade building the shovels for a gold rush that hadn't started - and when it did, it set the price of everything the rest of us build.
The bet nobody wanted
For years, betting the company on parallel compute looked eccentric. The demand wasn't there yet. Then the models arrived, the demand went vertical, and the eccentric bet became the most valuable infrastructure on earth. The pattern is worth memorising: be early to the layer everything else will need.
You don't have to build the shovels to profit from the gold rush. But you do have to understand who owns them - because they set your costs.
Why compute is the new oil
Oil didn't just power cars; it priced entire economies. Compute now does the same for software. The cost of a token quietly decides which app ideas are viable, which margins survive, and which "just add AI" features are a business versus a bonfire. When compute gets cheaper, whole categories of product suddenly become buildable overnight.
What it means for your stack
For a solo founder, three things follow directly:
Your costs are somebody else's roadmap
The price of the model you build on is set upstream. Design so a 10x swing in token cost - up or down - doesn't kill you. Cache aggressively; don't call a frontier model where a small one will do.
Cheap compute is a tailwind you can ride
Every price drop unlocks a product that was too expensive last quarter. The founders who win watch the cost curve like a surfer watches a wave.
Want the actual tools, not just the theory?
See the 6-app stack that runs a one-person software company - and where the equity-free AI credits are.
Read the builder's story →How to build compute-smart
Start on equity-free credits (the big clouds and Nvidia's own program hand them out). Route the easy 80% of calls to cheap, fast models and reserve the expensive model for the 20% that actually needs it. And price your product on value, not on your token bill - so a cost spike is a margin dip, not an existential event.
What's next
The direction of travel is clear: compute keeps getting cheaper per unit of intelligence, and each step down opens the door to a new wave of one-person products. Jensen's bet made the shovels. The vibe economy is what happens when everyone can finally afford to dig.
Read the newsletter that top founders & VCs keep in their inbox.