NVIDIA 10-Q: $13.4B Unrealized Gains Inflate Profits, True P/E May Reach 60x
Taylor Wilson
Nvidia's latest quarter reported $58 billion in net income, but $13.4 billion came from unrealized equity gains. Strip those out and the real P/E jumps from the widely quoted 45–50× to nearly 60× — leaving almost no margin of safety for a cyclical chipmaker with heavy customer concentration.
Where did the $13.4 billion come from — and why strip it out?
Nvidia's 10-Q shows ~$58 billion in net income. Of that, $13.4 billion is unrealized gains on publicly traded equity holdings, including Intel shares.
This means → the money never turned into cash and wasn't earned by selling chips. If those stock prices drop, the gain evaporates.
After stripping it out, normalized net income falls to ~$44–45 billion, pushing the P/E from the headline 45–50× up to nearly 60×.
In plain terms = reported "profit" overstates actual earning power by almost a quarter. The price you pay is far steeper than it looks.
Paying $17 billion for a rival — is the moat built or rented?
The 10-Q discloses a ~$17 billion deal with Groq: $13 billion in cash plus $4 billion in committed payments, generating $14.4 billion in goodwill on the balance sheet.
Groq — a direct competitor making AI inference chips — was acquired at a premium. This means → Nvidia needed to buy technology from a rival to keep its own architecture competitive.
$14.4 billion in goodwill is substantial against Nvidia's ~$50 billion in total equity. Any impairment would hit book value hard.
This reflects a moat that is at least partly "rented, not built." The analyst cites Intel's x86 history as a warning: rented moats eventually erode under competition.
Top three customers account for 54% of revenue — what if one cuts orders?
The 10-Q discloses that Nvidia's top three customers contributed 21%, 17%, and 16% of revenue — a combined 54%.
These are almost certainly hyperscale cloud providers (AWS, Azure, Google Cloud), all of which are actively developing custom chips to reduce Nvidia dependence.
This means → if just one of them cuts orders significantly, a 15–20% single-year revenue decline is not a tail risk — it is a plausible scenario.
At Nvidia's current 30× price-to-sales ratio, a 20% revenue contraction would trigger severe multiple compression. The downside risk is highly asymmetric.
Three metrics to watch — when does the stock become worth buying?
Metric 1: Non-cancellable supplier commitments — currently $119 billion. If this figure stalls or declines in coming quarters, internal demand visibility is weakening.
Metric 2: Customer concentration — if the top-three share drops from 54% to below 40%, the revenue-diversification narrative is materializing.
Metric 3: Groq goodwill — no impairment over the next four to six quarters would signal the acquisition is creating value. An impairment would be a dual red flag for both technology strategy and balance-sheet quality.
In plain terms = Nvidia is a great company, but the current stock price already prices in everything going right. How these three numbers evolve will determine whether the story delivers — or reverses.
Content is for reference only, not financial advice.