I completely agree that crypto, art in web3 etc has generally followed these general macro insights… really nice take magicforest 🙇
The ones with foresight = forward sight on the macro by zooming out are the ones that stay and build what comes next
On Ai… 🤭 the latest take from Jensen Huang captures it perfectly on the macro level and has been how it’s played out across all past emergent frontiers dot com era etc
AI is a five-layer cake.
1. Energy (the foundation)
The literal power grid. Turbines, solar farms, nuclear plants; whatever keeps the lights on. No electricity, nothing else moves. This is the dumb, physical constraint no algorithm can outsmart.
2. Chips (the engine)
The silicon that actually computes. GPUs, custom ASICs, the hardware that turns electrons into intelligence.
3. Cloud / Infrastructure (the plumbing)
Data centers, networks, cooling systems, the vast, capital-intensive backbone that keeps everything running 24/7.
4. Models (the brain)
The LLMs, frontier models, the flashy intelligence layer everyone obsesses over.
5. Applications (the penthouse)
Real-world use cases: AI that diagnoses disease, optimizes factories, trades markets, designs drugs, runs enterprises. This is where AI actually meets customers, solves problems, and prints money.
Huang’s insight is ice-cold and pure: The entire industry is crammed onto floor 4.
Every pitch deck, every valuation model, every founder narrative right now is laser-focused on “which model wins”, OpenAI vs. xAI vs. Anthropic vs. whoever drops the next frontier model. Trillions are flowing into that single layer while the real economic value lives one floor above and the unbreakable physical limits live three floors below.
The part that stings: the founders building these proposed models are showing zero foresight.
They’re treating the model layer as if it’s the entire skyscraper. They raise obscene capital, chase benchmarks, and market themselves as the future- while owning nothing at the base (someone else’s power and chips) and capturing almost nothing at the top (the actual applications that generate revenue and loyalty). They’re voluntarily positioning themselves as the squeezed middle: high-cost, high-competition, low-moat plumbing for someone else’s economy.
It’s the classic gold-rush mistake: spending billions to sell the picks and shovels while the real fortunes are made by the people who own the mines and the towns built on top of them. The model founders aren’t building the future, they’re becoming commoditized infrastructure in someone else’s cake.
Huang didn’t just describe the stack.
He exposed the delusion.
The smartest play isn’t winning the model war.
It’s owning the floors where the war actually ends: the energy that powers it and the applications that pay for it.
Everything else is just expensive noise in the middle. 🍰