AI company research

Research AI stocks beyond the headline

AI stocks are shares in businesses that build, supply or use artificial intelligence. Selling processors, renting computing capacity and charging for software create different sources of revenue and risk. This guide organizes the theme into business questions you can investigate. It is an educational framework, with illustrative companies rather than ranked investment picks.

Map the business model before comparing stocks

An AI label does not tell you how a company earns money. Start with the product, the paying customer and the cost of delivery. The categories below are Quaryn’s research framework; companies can operate in several categories, and the boundaries can change.

Follow the economic relationship through the chain. A large infrastructure purchase can be revenue for a supplier and a cash outflow for its customer. Strong demand in one part of the chain does not automatically mean attractive returns throughout it.

Map the business model before comparing stocks
Business categoryWhat the company may sellQuestion to investigate
Computing hardwareProcessors, servers and networking equipmentHow durable are demand, margins and customer commitments?
Cloud infrastructureComputing capacity and managed servicesCan utilization and customer revenue support the investment required?
Software and applicationsSubscriptions, usage or workflow productsAre customers paying more, staying longer or adopting a new product?
AI within an existing businessAn improved product or more efficient operationIs there evidence of financial impact beyond a product announcement?

Use company examples to practice source research

NVIDIA’s annual reports provide a starting point for studying accelerated computing and its platform business. Microsoft’s 2025 annual report describes AI offerings across cloud infrastructure and software. AMD’s annual filing library supports research into its processor and data-center businesses. These examples illustrate different exposures; inclusion is neither a recommendation nor a claim that one company offers better value.

For each example, open the most recent available filing and read subsequent updates before making a current comparison. A historical annual report can explain a business model while leaving later product, financial or competitive changes unanswered.

Use company examples to practice source research
Illustrative companyResearch starting pointUseful comparison question
NVIDIA · NVDAAnnual reports: business, products and risksWhich customers and products support the reported demand?
Microsoft · MSFTAnnual report: cloud, software and infrastructureHow do monetization and infrastructure costs develop together?
AMD · AMDAnnual filings: segments, products and competitionWhich reported measures support the data-center narrative?

Source references: NVIDIA — Annual Reports and Proxies · Microsoft — 2025 Annual Report · AMD — Annual SEC Filings

Ask what an AI revenue claim actually measures

A company may discuss AI demand without publishing a separate AI revenue figure. Do not substitute an entire cloud or data-center segment for AI revenue unless the source supports that definition. Keep management’s description, your inference and an explicitly reported measure in separate columns.

Check the unit as carefully as the number. Revenue, bookings, contracted commitments, customer counts and usage are different measures. A customer announcement establishes a relationship; it does not by itself establish recognized revenue, profitability or a recurring payment stream.

  • Copy the exact name of the reported metric, then explain it in plain English.
  • Record the period and whether the comparison uses the same definition.
  • Identify whether the figure is reported, forecast or inferred.
  • Leave AI-specific revenue unavailable when the source does not isolate it.

Put growth beside the cost of delivering it

Our worksheet pairs demand with its funding requirements. Compare the revenue story with cash generation, capital expenditure, operating expenses and share count. A business can grow while spending heavily to build capacity, and the timing of those costs affects how its financial statements should be read.

Consider a hypothetical company that reports 30% revenue growth while its shares outstanding rise 10%. Those two figures do not imply 30% growth in revenue per share: 1.30 divided by 1.10 gives approximately 1.18, or 18% growth. This arithmetic example shows why consistent denominators matter; it is not a forecast for any listed company.

Valuation adds another question: how much future success does today’s price require? Write down the assumptions needed for a comparison, including margins and reinvestment. Avoid using a single growth rate as a substitute for understanding the business.

Make AI-related risks specific

A useful risk statement explains an exposure and a possible consequence. Replace a vague note such as ‘competition is a risk’ with a question about which product could face lower pricing, whether customers can switch and what evidence would reveal the change. The checklist below is a starting point to investigate in each company’s own disclosures.

  • Customer concentration: How much demand depends on a small group of buyers?
  • Supply and capacity: Which components, facilities or energy arrangements are essential?
  • Product competition: Could a different design or supplier change pricing or demand?
  • Funding: What happens if capacity spending arrives before the expected revenue?
  • Legal and operational exposure: What does the company disclose about intellectual property, security or applicable restrictions?
  • Portfolio overlap: Could several holdings depend on the same spending cycle?

Build a comparison that preserves uncertainty

Create one row per company using the fields below. Populate it from dated documents, and retain unanswered questions. An incomplete but honest worksheet is more useful than a complete-looking table built from incompatible estimates.

An AI stock and AI stock analysis mean different things: the former describes an investment theme; the latter describes a research process using AI. Neither the theme nor the tool guarantees an investment outcome. Continue to our verification guide to learn how to check generated research.

Build a comparison that preserves uncertainty
Worksheet fieldWhat to write
Business exposureThe actual product, customer and payment model
EvidenceFiling URL, section, reporting period and metric definition
EconomicsGrowth, profitability, cash requirements and share count on a consistent basis
UncertaintyWhat is estimated, undisclosed or dependent on management expectations
Next questionThe evidence that would strengthen or weaken your interpretation

Sources & corrections

Use the linked primary sources to check definitions and company disclosures. Filings and service details can change; verify the relevant period before relying on a figure.

Found an error? Send a correction with the page, the claim and a supporting source. These guides provide general education; they do not assess your financial circumstances or recommend a trade. Read our research disclosures.

Your research. Your privacy.

Optional analytics help us understand site use. Optional advertising measurement records report requests. Both stay off until you choose. Research works either way. Privacy policy · How Google uses data