Hedge Fund Manager Dan Loeb: Nvidia Still Undervalued at $5 Trillion Market Cap
Alina Collins
Third Point CEO Dan Loeb says Nvidia's $5 trillion market cap is still cheap measured against future earnings — the real barrier is not fundamentals but investor psychology, where "too big" gets confused with "too expensive."
How can $5 trillion still be "undervalued"?
Loeb's core argument: Nvidia's share price looks enormous in absolute terms, but measured against earnings potential over the next several years, the stock is "absolutely" still cheap.
This means → he is not betting on a short-term bounce — he is betting that Nvidia's profit growth can keep outrunning its market-cap expansion.
In plain terms = if a company will earn far more next year than this year, it can still be inexpensive on a forward basis even after a massive rally.
Why are investors afraid to buy?
Loeb argues the root cause is psychological: investors struggle to accept that one company can be worth $5 trillion, so they automatically equate "big" with "risky."
He also flags a structural factor: many long-short hedge funds are required to maintain short positions, and Nvidia's sheer size makes it a natural short target — regardless of its fundamentals.
This means → some of the selling pressure comes not from bearish conviction but from portfolio construction rules.
Is there a historical precedent?
Loeb draws a parallel with Google and Amazon: "Google was a seemingly can't-lose short. Amazon was too."
His view: the stock will hover at a valuation level for a while, then break through — and Nvidia will follow the same path.
This reflects a recurring market pattern — when a company creates an entirely new category of demand, traditional valuation frameworks tend to underestimate its ceiling.
What underpins Nvidia's case?
Since early 2023 Nvidia's stock has risen nearly 14×; year-to-date it is up roughly 10%, making it one of Wall Street's biggest winners in the generative-AI wave.
The company supplies training and inference chips — the core hardware that lets AI models "learn" and "run" — to OpenAI, Google, Anthropic and others, cementing its central position in AI infrastructure.
Whether Loeb's thesis pays off hinges on one question: can Nvidia convert its current computing-demand advantage into sustainably expanding profits — demand is not the same as margin, and scale is not the same as a moat.
Content is for reference only, not financial advice.