Invest in Yourself, Not AI
FF2k3 min read·Just now--
The model isn’t the moat. You are
Everybody wants to know which AI stock to buy.
That’s usually a sign the easy money already left the building, bought a house in Wyoming, and stopped answering texts.
The people actually making money with AI are not sitting around guessing which ticker gets blessed by CNBC next week. They’re using the tools. Building faster. Writing better. Automating boring work. Replacing committee meetings with output.
Radical stuff, apparently.
The Part No One Wants to Hear
The American AI story is simple: frontier models are magic, magic costs billions, and if you buy the right slice of the magic factory, you win.
Nice bedtime story.
Meanwhile, China keeps walking into the room with near-frontier open models and dropping them on the table like, “Here. Have fun.”
Qwen is already into the 3.x generation, with open-weight models that keep pushing hard on coding, math, multilingual work, and agent use. GLM is up to 5.1. DeepSeek keeps shipping cheap, powerful models that make closed-model pricing look increasingly awkward.
This is the part investors should probably sit with for a minute:
If the model is free, cheap, or close enough, then the model is not where all the value lives.
The value shifts.
It moves to compute. Distribution. Workflow. Power. Data centers. Customer access. And, for normal people, actual skill.
That last one is inconvenient because you cannot buy it with one click in a brokerage app.
The Compute Game Is Not Your Game
Anthropic recently announced multiple gigawatts of Google/Broadcom TPU capacity coming online starting in 2027 to serve Claude demand.
Gigawatts.
That is not “we bought some servers.” That is “we need industrial civilization to answer customer support tickets.”
Anthropic is running Claude across AWS Trainium, Google TPUs, and Nvidia GPUs. Translation: even the serious players are scrambling for enough compute to keep the machine fed.
Claude revenue has exploded. Demand is real. The product is useful. None of that means every AI-adjacent stock at a stupid multiple is automatically a bargain.
Google can win because Google owns compute, chips, cloud, distribution, and half the internet’s attention span.
Amazon can win because AWS is where modern companies go to turn money into latency.
Microsoft can win because it already rents software to everyone with a pulse and a procurement department.
You?
You are not winning a data center arms race from your Fidelity account while eating Wheat Thins.
The Better Trade
The better trade is boring, obvious, and therefore mostly ignored:
Use AI.
Learn how to think with it. Learn how to write with it. Learn how to code with it. Learn how to research, build, sell, analyze, summarize, automate, and ship with it.
The models are getting cheaper. The interfaces are getting easier. The excuses are getting worse.
A mediocre person with AI is still mediocre, just faster. But a capable person with AI becomes a problem.
That is the opportunity.
Not “AI will replace everyone.”
Not “AI stocks only go up.”
Not “prompt engineer your way to a yacht by Tuesday.”
The real opportunity is leverage. Personal leverage. The kind that compounds because you become harder to ignore.
The Call
Stop trying to buy the wave.
Ride it.
Invest in skill — knowing how to use the tools without becoming dependent on them.
Invest in taste — knowing when the output is good, bad, or LinkedIn-flavored garbage.
Invest in judgment — knowing when AI helps and when it confidently hands you a bag of nonsense.
Invest in output — ship things, build things, publish things, improve things.
The models will keep getting cheaper.
The people who know how to use them will keep getting more valuable.
That’s where your money should go.
Into Yourself.
— FF2K
