Scaling sustained scarcities

I keep returning to three things people will continue to need: energy, health, and safety. The companies providing them will change. The human reasons for wanting them are unlikely to disappear.

That persistence is what I mean by a sustained scarcity. Demand for a useful outcome keeps expanding while reliable, affordable supply remains difficult to scale. It is a place to begin investigating, not a conclusion that any company operating there deserves investment.

There is an ethical trap in the phrase. If the business depends on keeping people short of something essential, its growth may preserve the very constraint it claims to solve. My interest is in scaling the capacity to meet the need. I want to manufacture better future service—and the credible cash flows that can finance it.

Abundance moves the constraint¶

When one input gets cheaper, the next bottleneck becomes more important. Cheap intelligence can increase the usefulness of electricity without creating another grid connection. Better diagnostic reasoning can reveal a treatment need without creating an appointment. A capable robot can still wait for a technician, a safe workspace, or a component.

Our 2025 WTW work on AI hardware examined how changes in computation could alter hardware demand and competition.⁵ The question I carry forward is broader: after an improvement, where does the cost of the completed outcome move?

This is a way to avoid investing in a slogan. “AI needs energy” is a starting observation. A company still has to solve a specific constraint at a cost a buyer will accept, capture a defensible part of the benefit, and survive the capital required to deliver it.

Three essential outcomes¶

Domain	The outcome I care about	The constraint to investigate	The ordinary-life consequence
Energy	Dependable service at an affordable total cost	Connection, generation, storage, maintenance, financing	A cooler home, a functioning business, fewer disruptive outages
Health	More healthy, capable time	Prevention, useful diagnosis, treatment access, delivery capacity	Less waiting and avoidable illness; more freedom to participate
Safety	Less harm at a given level of exposure	Detection, prevention, response, resilience	Travel, work, and daily activity with fewer catastrophic losses

The unit of energy value is not a panel, battery, or turbine by itself. A household needs the service when it matters, and a bill it can pay. The IEA’s account of grid constraints makes that distinction tangible: useful capacity depends on connections and system operation as well as generation.¹

In healthcare, counting more visits or tests can miss both progress and waste. Research on medical productivity that incorporates longevity and quality of life shows why the definition of output matters.² More useful health from a given resource budget is an achievement. More billable activity is an ambiguous signal. A lower-cost diagnostic tool can still increase total spending if it triggers large amounts of low-value follow-up.

Safety has a similar measurement problem: the successful event is often the one that never happens. Automatic emergency braking offers a bounded example. IIHS’s updated analysis associates the technology with substantially fewer rear-end crashes in the vehicles studied.³ It does not establish the safety of every automated system. It shows why a specific, repeatable reduction in harm deserves attention even when it is less spectacular than a promise of complete autonomy.

Manufacturing a future cash flow¶

A future cash flow becomes plausible through work done now: build the capacity, prove useful performance, find the payer, agree on a contract, and maintain the ability to deliver. The spreadsheet records that chain. It cannot substitute for it.

Power purchase agreements make the mechanism concrete. The EPA explains how long-term contracts with predictable revenues can help developers attract financing.⁴ A financial PPA settles differences against market prices; it does not itself supply physical electricity to the buyer. Even this apparently simple phrase—buying future power—contains distinct physical, commercial, and financial arrangements.

Credible future payments can fund useful capacity before a developer has earned enough to build it from retained profits. This is one of finance’s valuable capabilities: connecting a present construction problem with a future stream of service people expect to need.

The word “credible” carries most of the weight. The technology must work, the counterparty must pay, maintenance must be funded, and contract terms must survive adverse conditions. A long-duration agreement can make a project financeable while creating risks that somebody else now bears. Calling a payment recurring does not make it secure.

Two very different ways to get paid again¶

A customer can keep paying because the company keeps doing useful work. A customer can also keep paying because leaving is difficult, alternatives have been removed, or the underlying problem is never allowed to diminish.

These mechanisms can look similar in a financial model. They deserve different judgments.

For a hypothetical maintenance business, repeat revenue may reflect genuine ongoing labor that keeps equipment dependable. For a hypothetical platform that makes users less capable of leaving each year, repeat revenue may reflect increasing dependence. Neither conclusion follows from the subscription format alone. We need to understand what produces retention.

My test is counterfactual: if customers became healthier, safer, more efficient, or more capable, what would happen to this company’s economics? Some excellent businesses would earn less per customer but serve many more people. Some would earn the same fee for delivering more value. Others would need to change their business model. That last possibility is worth confronting before making a mission statement.

What it means for an investor¶

A real need, a useful product, a good company, and a good investment are four different propositions. They require four different kinds of evidence.

Competition can pass most of a technical breakthrough’s benefit to customers. That can be a tremendous social outcome and a mediocre return for a particular investor. Conversely, market power can produce impressive margins while little improves for the user. Valuation adds another distinction: an excellent company purchased at an excessive price can still disappoint.

I therefore want to trace both the useful outcome and the retained cash. How much capital is consumed before the next installation works? What must be replaced? Which costs fall with repetition? Does the customer keep some of the gain? Does a cheaper alternative undermine the contract? What happens if the claimed bottleneck stops being scarce?

The attractive position is often an organization that keeps finding and removing the next constraint. Owning a bottleneck and being good at eliminating bottlenecks are different capabilities. The second is the one I would rather help build.

The inheritance is more than the portfolio¶

For a family, a lower bill is useful. So is a smaller chance that one failure will destabilize the month. Dependable energy, effective care, and safer infrastructure create room for decisions that are not dominated by the next emergency.

Over a lifetime, that room can compound. Better health, more reliable education, a safer journey, and time available for learning or play all change what a future can contain. These are possibilities to pursue, not promises that technology automatically delivers. Access, distribution, and ownership determine who receives them.

We should count liabilities with the same seriousness as benefits. Pollution, brittle infrastructure, preventable illness, and systems that require permanent dependence can also be passed forward. A cash flow that looks manufactured today may be financed by costs someone else will inherit tomorrow.

The world I want to leave behind has fewer essential things hanging by a thread. I want businesses that become valuable by making that world more dependable. The test is what people can do with their lives after the company has been paid.