NVDA
NVIDIA
Semiconductors
★ Quality 90/100Price today
$207.54
what the market pays
Worth
$56.57
calculated cycle value
Price is 267% above its value
charlieapp.co
The global standard for artificial intelligence
NVIDIA created CUDA over 15 years ago. Today every large AI model in the world is trained on its chips. Replacing it would take years and billions of dollars.
Price vs Intrinsic Value
The line is the price month by month; the green band marks when it's undervalued (−20% off value). When the price drops in there, the discount is real, not opinion.
Where does the value come from?
5-year cycle
Money made year by year
This business has cycles — good years and bad years. That's why we don't use just the last year.
We use the average of those 5 years as the base for the math — not the good year, not the bad year. That way the value reflects what the company makes steadily.
How many times the cash flow
Semiconductors · vs 11 peers
You pay 80.9x times this business's cash flow; its sector median is 60.1x.
73% above what Charlie thinks it's worth (22x) — you're overpaying, sector or no sector.
Why Overvalued?
The model estimates an intrinsic value of $56.57 per share. Today's price of $207.54 is 73% above the calculated value — the market is paying a premium over what the model sees as fair.
The price is 267% above the calculated value. You're paying more than it's worth.
Buying here lowers your expected return and wipes out the safety margin.
The next reasonable entry zone starts at $53.74.
At what price would buying make sense?
Entry zones
🟢 Undervalued
≤$45.26
-20% off Value
🟡 Discounted
≤$53.74
-5% off Value
🔴 Today
$207.54
-73% Valor
Overvalued
The market prices it too high versus what the business makes. Patience pays off here.
“Time is the friend of the wonderful company, the enemy of the mediocre.”
— Warren Buffett
Model updated · July 2026
The business
In the AI gold rush, NVIDIA is the only store selling shovels.
NVIDIA designs the chips (GPUs) that train and run artificial intelligence. It sells those accelerators to the big clouds —Microsoft, Amazon, Google, Meta— who buy them tens of thousands at a time. Data center is already 80% of the business: in FY2026 that division grows nearly 94% to about $176B.
The moat isn't just the chips, it's CUDA: the software layer millions of developers have been programming on for fifteen years. Switching providers means rewriting everything from scratch. Add a two-generation design lead and 75% gross margins that fund the next round of R&D.
Revenue history
From $26.9B to $215.9B in 4 years — nearly 8.0x its size. Selling more and more is the base of everything else.
Where each $100 of sales goes
Revenue $215.9B · FY2026
Of every $100 in sales, $44 goes to costs and operations; $56 is left as net profit (56% margin).
Catalysts and risks
Data Center growing ~94% to $176B in FY2026.
Rubin GB300 Ultra arrives in the second quarter of 2026.
H200 reopening in China: $3-5B potential upside.
It's a cyclical business: once customers finish building, orders cool off fast.
Its biggest buyers design their own chips (Google's TPU, Amazon's Trainium).
Export restrictions to China that can slam shut overnight.
Charlie's note
“Paying 22x normalized flow for the best semiconductor business in the world, growing 65% with 75% gross margin, is no extravagance. The trap is confusing the top of the cycle with the permanent state and paying for optimism that hasn't yet been booked.”
Analysis · July 2026
So when would be a good price for NVIDIA?
By our calculation, not yet. We will email you the day it drops to $53.74 — so you do not have to keep checking.
🔔 Email meFree · no card
Charlie informs, it doesn't advise. No information should be taken as a recommendation to buy or sell.