AI is a theme, not a portfolio · Plain Money Notes

Artificial intelligence can change the economy without making every company that says “AI” a good investment.

That distinction is the whole job.

An investor can be right about the technology and still lose money by buying the wrong business, paying too much, concentrating too heavily, or trusting a product whose connection to AI begins and ends in the marketing department.

So do not start with a ticker. Start with an exposure map.

The short version: A broad-market fund may already give you substantial exposure to companies building and using AI. A dedicated AI stock or ETF is an optional tilt, not a replacement for diversification. Before buying, identify which part of the AI economy you want, measure overlap with what you own, examine the price and business economics, and write down what would prove your thesis wrong.

Map the AI economy before shopping for it

“AI company” is not one business model. The economic chain has at least four layers, and a company can occupy more than one.

Layer	What it provides	Questions worth asking
Infrastructure	Chips, networking, data centers, power, cooling, and other physical capacity	Does demand justify the capital being spent? Who has pricing power, and what happens if capacity outruns demand?
Platforms and models	Cloud computing, foundation models, developer tools, and data services	Is usage producing durable revenue? How expensive is it to train and serve the product?
Applications	Software built for a particular consumer or business task	Does the product solve a problem customers will keep paying to solve? Can a larger platform copy it?
Adopters	Companies using AI to improve products, decisions, or productivity	Is AI producing measurable economics, or does it merely improve the presentation deck?

This map prevents two common mistakes.

First, it keeps you from assuming that the most visible chatbot captures all the value. Suppliers, distributors, customers, and ordinary-looking adopters may benefit too.

Second, it exposes hidden repetition. A chipmaker held directly, inside a technology fund, inside an AI-themed ETF, and again inside a broad-market fund is still the same company. Four account entries do not create four independent outcomes.

Choose the vehicle after you choose the job

There is no single “AI investment.” Each vehicle solves a different problem and leaves a different risk behind.

Vehicle	What it can do	What it does not solve
Broad-market fund	Owns many businesses, including some AI builders and adopters	Offers no precise control over the size or layer of AI exposure
Sector or industry fund	Targets a part of the supply chain, such as semiconductors or software	Can be concentrated in one industry and one business cycle
Thematic AI ETF	Packages several companies selected under an AI-related methodology	The label may hide high fees, loose definitions, concentrated holdings, and overlap with funds you already own
Individual stock	Lets you underwrite a specific company, valuation, and business thesis	Leaves company, management, competition, execution, and valuation risk on your shoulders

A narrow ETF is not automatically diversified just because it owns more than one stock. The SEC’s asset-allocation guide warns that a narrowly focused mutual fund or ETF may not provide immediate diversification. The guide also recommends checking top holdings when you own more than one fund.

That makes a broad-market holding a reasonable place to begin the investigation. Look through the fund you already own. You may discover that your portfolio already participates in AI through infrastructure companies, major platforms, software providers, and adopters. Buying a thematic fund may increase those weights rather than add a genuinely new source of return.

Run the overlap arithmetic

Here is a deliberately simplified example.

Suppose an investor has $90,000 in a broad-market fund and is considering $10,000 in an AI ETF. If 30% of the AI ETF consists of companies already prominent in the broad fund, then $3,000 of the new purchase increases exposure to familiar holdings. The remaining $7,000 may add different names, but “different” does not necessarily mean diversified: those companies may still depend on the same customers, spending cycle, financing conditions, or technology trend.

This is not an argument against the purchase. It is an argument for describing it honestly.

Do not say, “I added diversification,” when the transaction really means, “I increased my AI-linked concentration by $10,000.” A concentrated tilt can be a deliberate decision. It should not be an accidental one.

Give an AI tilt a written job

A theme should have a defined role in the portfolio. “It is the future” is a prediction, not a position-sizing rule.

Before buying, finish these sentences:

I want this holding because: ___
It gives me exposure to this part of the AI economy: ___
I already own these overlapping companies or industries: ___
The most I intend this theme to represent is: ___
I will review it when: ___
My thesis would be wrong if: ___

There is no universal correct percentage for an AI tilt. The amount depends on the rest of the portfolio, the goal, the time horizon, the ability to tolerate losses, and the investor’s willingness to follow the underlying businesses. Money needed soon should not be volunteered for a volatile technology thesis simply because the story is exciting.

One more useful question: What will this replace?

If the answer is “nothing,” you may be expanding risk without noticing. If the answer is a broad diversified holding, be explicit about the diversification you are giving up. If the answer is another speculative position, compare the two risks rather than treating the new idea as free space.

Read past the letters A and I

For an AI ETF, inspect:

The index or selection methodology. How does the fund decide that a company qualifies? Is AI central to revenue, one product among many, or merely mentioned in public materials?
The current holdings and weights. How much sits in the ten largest positions? Which holdings repeat across your other funds?
The rebalancing rules. How often can the portfolio change, and what can cause a company to enter or leave?
The complete cost. The expense ratio reduces returns, but it is not the only cost. The SEC’s fund-fee bulletin notes that transaction and ongoing fees can materially affect a portfolio. ETF investors may also face brokerage costs, trading spreads, and prices above or below net asset value.
The fund’s history and assets. A new or small fund may have a limited record and can close if it fails to attract enough assets. A long record still does not guarantee future results.

For an individual company, move from the press release to the filings. Use the SEC’s EDGAR search to find the latest annual and quarterly reports, then ask:

What product does the company sell, and who pays for it?
How much of the claimed AI opportunity appears in revenue today?
What does it cost to build and deliver the product?
Is cash flow improving, or is enthusiasm outrunning the economics?
How dependent is the company on a small number of suppliers or customers?
What competition, regulation, security, copyright, and capital-spending risks does management disclose?
What expectations already appear to be embedded in the share price?

A wonderful business can be a poor investment at an extravagant price. A cheap-looking stock can be a poor investment if the business is deteriorating. “AI” cancels neither sentence.

Separate AI research from AI authority

An AI assistant can help generate questions, summarize a document, or organize a comparison. It cannot assume responsibility for the facts, understand your complete financial situation, or make an uncertain forecast become reliable.

The SEC, FINRA, and state securities regulators warn investors not to rely solely on AI-generated information. AI output can be inaccurate, incomplete, outdated, misleading, or fabricated. Verify important claims against company filings, fund documents, regulator records, and other primary sources.

The same alert identifies familiar fraud dressed in newer language: unregistered platforms, claims of guaranteed winners, systems that supposedly cannot lose, urgency, and requests to send money through suspicious channels. AI can also make impersonation more convincing. A realistic video, voice, or message from a chief executive, celebrity, adviser, or family member is not proof that the person created it.

Be especially skeptical when a pitch includes:

guaranteed returns, little risk, or a “secret” AI trading system;
pressure to act before you can verify the seller or investment;
a platform or professional you cannot confirm through regulator records;
a celebrity or executive endorsement that appears only in a clip or direct message;
no clear business model, filings, methodology, or independently verifiable results; or
instructions to pay by cryptocurrency, wire, gift card, or an unfamiliar app.

“AI-powered” also deserves evidence. In March 2024, the SEC charged two investment advisers with making false and misleading statements about their use of AI; the firms settled and agreed to pay $400,000 in total civil penalties. A fashionable label is not due diligence.

Use a one-page AI exposure x-ray

Before making the trade, put the proposed holding beside the portfolio you already have.

Field	What to write down
Intended layer	Infrastructure, platform/model, application, adopter, or a defined mix
Vehicle and job	Broad core, sector tilt, thematic tilt, or individual-company thesis
Current exposure	AI-linked companies and industries already held across every account
Concentration	Top-ten weight, largest company weight, and major industry dependencies
Overlap	Repeated holdings and repeated economic drivers
Methodology	How the fund defines AI, or how the company makes money from it
Cost and price	Fund expenses and trading costs, or the valuation assumptions for a stock
Failure condition	The observable facts that would invalidate the thesis
Maintenance rule	Review date, target size, and rebalancing or exit rule

If you cannot complete the page, you have found the next piece of research. You have not found a reason to hurry.

The bottom line

AI may prove to be one of the defining technologies of this era. That still does not tell you which company will capture the value, which fund owns it at a sensible weight, or what price will produce an acceptable return.

Let a diversified portfolio do the heavy lifting. Treat an AI holding as a deliberate tilt with a named job, measured overlap, understood costs, and a written failure condition.

The future can be enormous. Your position does not have to be.

This article provides general educational information, not individualized investment, tax, or financial advice. Investments can lose value, thematic funds can be volatile, and fund holdings, strategies, fees, and company circumstances can change. Verify current information and consider a qualified professional who can evaluate your complete situation before acting.