Shareholder note · 4 August 2026
NVIDIA, through a shareholder's eyes.
A simple owner-earnings view of an extraordinary business—and the evidence that would make me change my mind.
I own NVIDIA shares. That fact should make readers more skeptical of my conclusions, not less.
This is not an argument that NVIDIA's share price can only rise. It is an explanation of how I think about the underlying business as a shareholder: what NVIDIA earns, what belongs economically to each share, what the valuation already assumes, and what evidence would make me change my mind.
A shareholder owns a claim on per-share economics—not headlines, revenue announcements or accounting labels.
My central question is simple: after the cash required to operate and invest in the business, and after recognizing the economic cost of employee stock compensation, how much earning power belongs to each NVIDIA share?
The business today
An AI infrastructure system, not merely a chip.
NVIDIA supplies much of the computing stack required to build and operate modern AI: accelerators, networking, systems, software and the development ecosystem around them.
Q1 FY27, quarter ended 26 April 2026. NVIDIA earnings release · Form 10-Q
My analytical lens
Three earnings numbers. Not one.
Each measure answers a different question. They should never be added together.
GAAP net income
The accounting result attributable to shareholders. Essential, but it can include investment gains and other items that are not recurring operating earnings.
Free cash flow
Cash from operations after cash capital expenditure and relevant principal payments. Closer to the economic engine, but operating cash flow adds stock compensation back.
Owner earnings, my way
Free cash flow minus stock-based compensation. It is deliberately conservative and is my non-GAAP analytical measure—not a number NVIDIA reports.
What NVIDIA earned in Q1
Cash earnings were exceptional. Accounting earnings were higher still.
Q1 GAAP net income was $58.3B, while my owner-earnings measure was $46.6B. Across the trailing twelve months, my derived owner-earnings measure was approximately $112.2B.
After $19.3B of cash repurchases and $0.2B of dividends, roughly $27.1B of Q1 owner earnings remained after cash returns. Repurchases are capital allocation—not additional earnings.
Trailing-twelve-month owner earnings: $90.189B for FY26 + $46.626B for Q1 FY27 − $24.661B for Q1 FY26 = $112.154B. FY26 Form 10-K · Q1 FY27 Form 10-Q. Owner earnings and retained cash are derived from filed inputs.
Book equity
A useful audit bridge—not the destination.
NVIDIA's shareholder equity rose from $157.3B to $195.5B in Q1, an increase of $38.2B. That bridge is exact, but it is not the same as owner earnings.
Book equity also moves with investment marks, repurchases, stock-compensation accounting, dividends, issuance and other comprehensive income. I use it to audit the accounting. I use owner earnings per share to judge economic progress.
The seven-year question
How much exceptional execution is already in the price?
At the 3 August close of $206.64, NVIDIA's market capitalization was approximately $5.0T. That valuation does not require a good company. It requires enduring economic leadership.
| Scenario | Seven-year owner earnings | Present value at 10% |
|---|---|---|
| Bear | $1.50T | $1.02T |
| Base | $2.93T | $1.92T |
| Bull | $4.40T | $2.81T |
In my base case, the present value of the first seven years of owner earnings covers about 38% of today's market capitalization. The rest depends on earning power beyond FY33, a different required return, or terminal value.
These are illustrative business scenarios, not company guidance, a price target or a prediction. FY27 starts from consensus low / central / high revenue anchors; subsequent growth, margins, share count and capital-return assumptions are mine. Consensus reference
Scenario methodology and yearly assumptions
Company owner earnings equal revenue multiplied by the stated owner-earnings margin. Present value discounts each annual result at 10% at year-end. Share-count assumptions inform the per-share lens; they do not change the company-level totals above.
| Scenario | FY27 revenue | FY28–FY33 revenue growth | FY27–FY33 owner-earnings margin | Annual net share shrink |
|---|---|---|---|---|
| Bear | $358.4B | 22%, 12%, 8%, 6%, 5%, 4% | 46%, 45%, 43%, 42%, 41%, 40%, 39% | −0.2% in FY27; 0% thereafter |
| Base | $393.9B | 42.6%, 27%, 20%, 15%, 11%, 8% | 52%, 52%, 51.5%, 51%, 50.5%, 50%, 49.5% | 1%, 1.2%, 1.4%, then 1.5% |
| Bull | $415.5B | 50%, 35%, 27%, 22%, 18%, 15% | 56%, 57%, 58%, 59%, 59%, 58.5%, 58% | 1.5%, 1.8%, then 2% |
The shareholder thesis
What must remain true.
My conviction is conditional. These are the operating facts that must continue to support ownership.
AI demand must become durable utilization.
The infrastructure buildout must translate into workloads and customer returns—not simply pull future demand into the present.
NVIDIA must keep winning at the system level.
Accelerators, networking, CUDA, libraries, developer familiarity and deployment speed must remain more valuable together than rival components are separately.
Product cadence must stay intact.
Blackwell execution and the transition to Rubin need to preserve NVIDIA’s performance and time-to-market advantage.
Gross margin must keep proving pricing power.
A changing product mix can move margins quarter to quarter. Persistent deterioration without a credible reinvestment payoff would be different.
Customer concentration cannot become customer control.
A few hyperscalers fund enormous demand. Their scale also creates bargaining power and the risk of synchronized capital-spending pauses.
Buybacks must improve per-share economics.
Repurchases help continuing owners only when shares are retired at sensible prices and the effect exceeds stock-compensation dilution.
How I can be wrong
The risk is not that AI disappears. It is that the economics disappoint the valuation.
Demand digestion
A hyperscaler spending pause could expose excess capacity and slow NVIDIA’s revenue conversion.
Custom silicon
Large customers may move more workloads to internally designed accelerators when economics or control justify it.
Competition
AMD and others may narrow performance, networking or software gaps faster than my assumptions allow.
Export controls
China is already excluded from the Q2 Data Center compute outlook; broader restrictions could remove more markets.
Investment marks
Equity gains can lift GAAP income without improving recurring operating cash generation.
Valuation duration
Outstanding business execution may still produce a weak stock result if the market shortens the duration or multiple it will pay for.
NVIDIA reported $119B of manufacturing, supply and capacity commitments, separate from $30B of cloud-service commitments and $6B of other vendor commitments. Form 10-Q
My quarterly shareholder scorecard
Eight questions. Every quarter.
I prefer a repeatable scorecard to a new narrative after every price move.
A shareholder's conclusion
The business is exceptional. The price demands that it stay exceptional.
NVIDIA combines extraordinary growth, cash generation and a deeply established computing ecosystem. The evidence supports a business capable of producing exceptional owner earnings.
The share price also requires exceptional execution. Those two statements can be true at the same time.
I remain an NVIDIA shareholder because I believe the company's underlying economic earning power continues to strengthen. My view would change if owner earnings per share stopped compounding, gross margins deteriorated without a credible reinvestment payoff, customer concentration became structural bargaining power, or NVIDIA lost its product and software cadence.
Ownership is not a declaration of certainty. It is a commitment to keep doing the work.
Manish Sainani
Founder & CEO, Hussh
Primary reading