An annotated project analysis: NVIDIA's FY2026 filing

This case shows how to turn a public company filing into a bounded research note. It uses NVIDIA's fiscal year ended January 25, 2026, and separates reported figures, transparent calculations, interpretation and unanswered questions. It is a historical learning example prepared for Orbitrum Academy, not a live company rating, an expert endorsement or a recommendation to buy or sell a security.

Define the scope and freeze the source period

Our question is: "What does this filing establish about annual revenue growth, gross margin and direct-customer concentration, and what remains unanswered?"

The primary document is NVIDIA's Form 10-K filed on February 25, 2026. This case does not incorporate later quarterly reports, share prices or market events. The public project profile may contain newer information, but that does not silently update the evidence in this exercise. NVIDIA FY2026 Form 10-K

The SEC explains that a 10-K is prepared by the company, not written or guaranteed by the SEC. Management discussion and financial statements have different roles within the report. An official filing is an important source, not a government endorsement of the business. SEC, How to Read a 10-K

Read the reported observations before making a judgment

The following figures are taken from the filing. Monetary values in the table are in millions of US dollars. Fiscal years refer to NVIDIA's fiscal periods, not calendar years.

The customer disclosures above concern direct customers. The filing excerpt does not name those two customers, and this case does not guess their identities or treat them as a measure of all end-customer demand. NVIDIA FY2026 Form 10-K, pp.37 and 78

Reproduce the calculations

The calculations below use the reported numbers and are made for this Academy example; they are not additional figures quoted from the report.

Revenue change: (215,938 - 130,497) / 130,497 x 100 = approximately 65.5%.

Gross-margin change: 71.1% - 75.0% = -3.9 percentage points.

Combined share of the two disclosed direct customers: 22% + 14% = 36%.

Keep the units and the distinction between percent growth and percentage-point change. The rounded customer shares should not be used to manufacture falsely precise dollar amounts. A reproduced arithmetic calculation is useful, but it is not independent verification of the underlying records.

Separate interpretation from fact

We can now write three bounded interpretations. They are Academy reading examples, not conclusions attributed to any Orbitrum expert.

First: reported revenue increased substantially over the comparison period. This is a historical growth observation. It is not a forecast of the next period.

Second: gross margin was lower while revenue was higher. Therefore, "revenue grew" and "gross margin declined" can both be true. Neither observation alone explains every change in profitability or establishes the cause of the margin movement.

Third: the two disclosed direct customers together account for a meaningful share of revenue in this period. The calculation identifies a concentration question. It does not establish which customer will change purchases or how end demand will develop.

The value of this format is that a reader can see exactly where interpretation begins. Avoid converting these observations into an unexplained label such as "safe," "overvalued" or "best investment."

Turn the open questions into a research plan

The following questions are proposed follow-up work. Their answers are not supplied or inferred by the three metrics above.

Locate additional relevant passages and records before answering these questions. When a source is unavailable, retain the question rather than completing the gap with a confident narrative.

A compact research memo

Here is the structure of the resulting note:

Question: assess three historical financial characteristics in the specified filing.

Evidence: dated issuer disclosures, with units and page references preserved.

Calculations: reproducible arithmetic, explicitly separated from quoted results.

Conclusion: growth, margin and concentration describe different aspects of the business and should be read together without turning them into a complete investment verdict.

Limitations: no later-period update, direct access to accounting records, forward model, valuation or personal suitability analysis.

Next step: select the unanswered question that matters for the actual research objective and obtain evidence specific to it.

This memo is an Academy example, not an entry in NVIDIA's expert-evaluation history on Orbitrum.

Compare this with an expert's assessment

Open NVIDIA's public Orbitrum profile to find the information and research currently available there. Look for an author's scope, publication date, reasoning and sources. Do not expect a later assessment to use only the historical information in this case.

A difference between this exercise and an expert's conclusion may be a difference of question or period. Use the critical-reading worksheet before comparing scores or summarizing disagreement.

To explore the surrounding field, start from Orbitrum's sector directory. This guide does not equate every AI or technology company with a semiconductor manufacturer.

Questions about the NVIDIA case

Is this an investment recommendation or an Orbitrum score?

No. It is a source-based learning exercise. It supplies no target price, buy/sell opinion, portfolio allocation or expert rating.

Why use fiscal-year labels instead of just saying 2026?

The case concerns the period ended January 25, 2026. Using the exact fiscal period avoids implying that the figures cover the entire calendar year or the latest available quarter.

Can I use the same method for a private startup?

You can reuse the separation of source, calculation, interpretation and unanswered question. Do not assume a private company supplies a public-company filing or that its metrics have the same definitions. See the traction evidence guide.