Picks-and-Shovels Investing
Back the shared infrastructure serving every contender in a growing market
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 99%
The picks-and-shovels model shifts attention from uncertain application winners to the infrastructure used by most participants in a rapidly expanding market. An investor first identifies the indispensable inputs required by competing companies, then looks for suppliers whose revenue grows as the entire ecosystem experiments and scales. Strong candidates combine broad customer exposure with differentiated technology, switching costs, capacity advantages, or an ecosystem that alternatives cannot easily reproduce. NVIDIA illustrated the pattern when AI startups with very different products consistently built on its platform. The framework is not permission to ignore price: investors must assess whether expected market growth and supplier capture already appear in the valuation. They must also test whether large customers can design substitutes or whether a new architecture can displace the incumbent.
Origin
Extracted from Acquired's investing lesson about owning NVIDIA rather than trying to select individual early deep-learning startups.
Core principles
- 01Infrastructure can win even when individual application winners are unknowable
- 02Repeated dependency across competitors is evidence of platform power
- 03The best supplier serves many participants in the expanding market
- 04Adoption, pricing power, and substitution risk matter more than a catchy analogy
- 05Valuation still determines whether a strong company is a good investment
How to run it
- 1
Define the gold rush
Identify a market experiencing rapid experimentation, investment, and expected demand. Separate durable structural growth from a temporary speculative surge.
Pro tip Look for growth in real workloads and customer spending, not only financing activity.
Watch out A fashionable category without economic demand may not sustain infrastructure sales.
- 2
Map indispensable inputs
List the hardware, software, data, distribution, and services required by nearly every participant. Identify where supply is concentrated.
Pro tip Study the technical stacks of many competitors rather than relying on one company.
Watch out A commonly used input is not attractive if it is easily commoditized.
- 3
Find the shared supplier
Determine which provider repeatedly appears across otherwise unrelated applications. Verify that its product is important to performance or economics.
Pro tip Customer dependency is stronger evidence than management's total-addressable-market claims.
Watch out High visibility can reflect temporary convention rather than durable advantage.
- 4
Audit durability
Assess switching costs, economies of scale, ecosystem depth, pricing power, and replacement threats. Examine whether customers are motivated and able to build internally.
Pro tip Pay special attention to integrated software and accumulated developer knowledge.
Watch out High margins can finance innovation but also motivate powerful customers to defect.
- 5
Price the expectations
Compare the current valuation with realistic scenarios for market growth and supplier capture. Require an adequate return from today's price.
Pro tip Model both continued dominance and a normalization of growth.
Watch out A great picks-and-shovels company can still be a poor investment when expectations are excessive.
In the wild
Early AI startups pursued many uncertain applications, but Andreessen observed that each one effectively built on NVIDIA's platform. Buying the common infrastructure provider would have reduced the need to predict which individual application companies would survive.
→ Investors could have captured broad exposure to the AI wave through its dominant computing platform.
An investor studying a wave of battery startups finds that most depend on one specialized testing-equipment supplier. The investor examines its order growth, service revenue, switching costs, capacity, and valuation rather than selecting a single battery chemistry winner.
→ The investment gains diversified exposure to sector growth while retaining supplier-specific diligence.
Common mistakes
Ignoring valuation
The model identifies a strategically attractive position, not an automatically attractive stock. Excessive expectations can erase future returns.
Mistaking usage for power
A widely used commodity supplier may capture little value. Confirm differentiation, switching costs, or another durable advantage.
Overlooking customer integration
Large buyers may build the input internally to escape high margins or gain specialization. Evaluate their incentives and capabilities.
Is it for you?
Best for
It is best for markets where many competitors depend on a small number of differentiated infrastructure providers.
Not ideal for
It is not ideal when inputs are commoditized, easily replaced, or already priced for implausible growth.
From the transcript
“But like, man, you really, really want to invest in whoever is selling the picks and the shovels in a gold rush.”
“I don't know if any one of those startups, any given one is gonna succeed, but I'm pretty sure NVIDIA is gonna succeed back then.”
“And every single one effectively comes in building on Nvidia's platforms.”
From the episode
Nvidia Part II: The Machine Learning Company (2006-2022)