AI strategy has a bad habit of sounding bigger than it is.

Every few months, someone declares that moats are dead. Then someone else declares that the old moats are stronger than ever. Both takes are usually too neat. AI does not abolish strategy. It changes which advantages compound, which decay, and which were never real advantages in the first place.

Hamilton Helmer's Seven Powers remain a useful lens: scale economies, network economies, counter-positioning, switching costs, branding, cornered resource, and process power. Useful, not sacred. The danger is treating the framework like a checklist where every company gets to claim three powers by Friday.

In the AI era, that mistake gets expensive.

A company can ship faster and still have no moat. It can have proprietary data and still be unable to use it. It can add an agent to the product and still be a feature inside someone else's workflow. It can have customers who hate switching and still lose when the replacement is cheap enough, good enough, and bundled into a system they already use.

The real question is not "Do we use AI?" Most companies will. The useful question is: what does AI make harder for competitors to copy without paying a real price?

The compression problem

AI compresses capability gaps.

A small team can now produce software, content, analysis, support, and automation that used to require larger teams. That is wonderful for execution. It is brutal for weak strategy. If your advantage was merely that you had more people doing ordinary work, AI is coming for that spread.

This does not mean scale stops mattering. It means the shape of scale changes. Scale based on headcount or basic production capacity gets weaker. Scale based on distribution, infrastructure utilization, customer reach, workflow depth, and learning loops can get stronger.

The same pattern shows up across the Seven Powers. AI weakens shallow versions and strengthens deep versions.

Generic data gets cheaper. Workflow data tied to real decisions can matter more. Basic software features are easier to copy. Embedded operational systems are harder to dislodge. Brand slogans matter less. Trust under uncertainty matters more.

The old moat test is too soft

Executives often ask whether an advantage is defensible. That sounds rigorous, but it is often mushy.

A better test: can a serious competitor copy the move without hurting themselves?

If the answer is yes, you probably have execution, not power. Execution matters. It can produce years of good outcomes. But it is not the same thing as structural advantage.

Counter-positioning only works when the incumbent would damage its current model by matching you. Switching costs only matter when the customer would lose something material by leaving. Process power only compounds when the way you work produces better results than competitors can get by buying the same tools.

AI raises the bar because more tools are available to everyone. Your competitor can buy the same foundation model. They can hire from the same talent pool. They can use the same coding assistant. They can automate the same support queue.

So the moat has to live somewhere else: in the loop, the context, the behavior change, the customer relationship, the operating cadence, the rights, the distribution, the trust, or the accumulated decisions that are hard to reverse.

Strategy becomes more empirical

The upside is that AI makes some strategy questions easier to test.

You can simulate workflows. You can prototype products faster. You can instrument customer behavior in more detail. You can run smaller experiments before committing a full team. You can measure whether AI improves retention, margin, cycle time, win rate, or product quality.

That does not make strategy easy. It makes lazy strategy more obvious.

If a company says AI creates scale economies, ask where the fixed cost sits and whether unit economics improve with volume. If it says it has network effects, ask what gets better for one user when another user joins. If it says it has switching costs, ask what exactly the customer loses by leaving.

If the answer is a cloud of words, there is probably no power there.

The useful posture

The right posture is skeptical without becoming smug.

AI will create durable companies. It will also create a lot of businesses that look powerful during an adoption spike and ordinary once the market catches up. The job is to tell those apart early enough to make better decisions.

This series uses the Seven Powers as a set of pressure tests. Not to prove that every AI company has a moat. Quite the opposite. The goal is to separate real power from expensive theater.

That is the work now. Less prophecy. More diagnosis.

This is part 1 of 10 in Seven Powers in the AI Era.