AI Picks-and-Shovels Stocks: Are Suppliers Safer in a Bubble?

“Sell the picks and shovels” is a popular response to uncertainty about which AI applications will win. It can be a useful business-model distinction. It is not a guarantee that suppliers are insulated from an investment boom turning into a spending pause.

Separate first-order and second-order demand#

First-order demand is the order a supplier receives. Second-order demand is the customer's ability to earn a return from what it buys. The supplier may collect revenue before that return has been demonstrated.

The BIS 2026 annual report discusses how competitive investment can create uncertain returns across the AI ecosystem. Our inference is that supplier analysis should include the economics of the customers funding expansion, not only current orders.

Avoid double-counting the same boom#

If several infrastructure companies ultimately depend on the same group of buyers, holding all of them may diversify product risk without diversifying the capital-spending cycle. This is a question to investigate through customer disclosures, not an assertion that every supplier has identical exposure.

For Nvidia, AMD, Micron and HPE, review the relevant filings before assigning a concentration figure or treating all reported revenue as AI revenue.

Stress-test an order pause#

Model a period with slower new orders, less favorable pricing and increased inventory. Distinguish growth deceleration from an outright revenue decline: both can disappoint a valuation built on acceleration.

Also examine whether the business can reduce costs quickly. Long supply commitments can make a spending slowdown more expensive than a simple revenue forecast suggests.

Compare price with protection#

An attractive competitive position may already be reflected in an expensive stock. A supplier with stronger cash generation can still deliver a poor return from an excessively optimistic entry price.

The picks-and-shovels argument is strongest when supported by resilient customer economics, manageable commitments and a valuation that does not require uninterrupted expansion. It is weakest when the metaphor substitutes for that work.

More in this analysis series#

Start with AI stocks versus railroad mania. Continue with AI Bubble Warning Signs: What to Monitor Without Predicting a Crash.

Analysis published October 1, 2026. Historical comparisons and hypothetical examples are not forecasts. Educational information, not investment advice.