The Expansion Tax

An operator in 2026 has two ways to think about AI. The first is the one most boards talk about: "How much can we save?" The second is the one that wins: "What can we now do that we couldn't before?"

These sound like the same question with different framing. They are not. They produce opposite operations within twelve months.

The 10x productivity move

Product strategist Nate B. Jones put it most cleanly in a December 2025 episode:

The logic is unsentimental. If a company's effective execution capacity has increased ten-fold and their response is to cut staff, the implicit admission is that the company sees no new work worth doing with the additional capacity. The market — their market — isn't worth investing into. They are not picking efficiency over growth as a strategy. They are signaling that they don't believe growth is available to them, which is a different and worse statement.

The companies running the opposite play are doing something specific. They are taking the capacity gain and pointing it at adjacent work the business previously could not afford to do.

What expansion actually looks like for an operating business

For most small-to-mid market operations, the math runs like this. A two-person customer success function was previously absorbed by handling existing accounts. With AI handling the routine work, those same two people can now run a structured outbound expansion motion to existing customers — cross-sell, upsell, account growth — that the company previously had no capacity for. Same payroll. New revenue line.

A four-person operations team that spent every day fighting same-day issues now has space to redesign the upstream processes that generated the issues, and to take on adjacent operational work for a new product category. Same headcount. New product launch made possible.

A small accounting firm that bills per-hour discovers it can now serve double the client base at the same partner-hour input. They can either pocket the margin or use the expanded capacity to take on a higher-tier client segment that previously required staff they couldn't justify hiring. The firm gets bigger by growing into the work, not by cutting to fit it.

None of these moves are about replacing people. All of them require the people. The AI handles the bulk work; the humans do the higher-order work that the bulk work used to crowd out. The operator who confused "AI productivity" with "fewer people" never sees the second-order opportunity at all.

Why the cost-cut move loses

The company that uses AI to cut headcount is operating on a fixed-pie assumption: the market is what it is, our share is what it is, the only lever is cost. This is a defensible view in industries that are genuinely saturated.

The 2026 reality, in nearly every operating-business category, is the opposite. When execution costs across the entire industry drop by an order of magnitude, customer expectations move too. The same customer who used to wait 48 hours for a quote now expects 4. The buyer who used to accept generic outreach now expects personalized context. The market doesn't sit still while you compress your costs. It expands — in service quality, in response speed, in customizability, in the number of customers an operating model can address.

The cutters are reducing capacity at the exact moment the market is rewarding capacity. The expanders are taking the new capacity and converting it into the service expectations that customers are already starting to require. The cutters won't see the gap until it's the price of a customer they've already lost to a competitor who shipped faster.

The harder problem the cutters skip

Cutting headcount is the easy version of the AI strategy question. There is a spreadsheet. There is a number. There is a board that approves it. Expansion is the hard version — it requires answering "what new work can our team take on now that the bulk work is handled" and that question has no spreadsheet because no one has the answer yet for your specific business.

Jones is direct about this:

The reason this work gets skipped is not laziness. It is that operators do not know where to start. The capacity gain is real but invisible without instrumentation. The adjacent opportunities are real but only nameable by someone who has watched a dozen other operating businesses make the same transition. The team is willing but doesn't have the playbook for what to do with the time AI gives them back.

This is the work an embedded engagement does. The AI install is the first half. The harder, more compounding half is sitting with operators and naming the expansion play their specific business should run. Most consultancies don't do this because it requires shipping the AI first — and they don't ship the AI.