--- title: Was the Cost of That Task Your Moat? synced_from_vault: true vault_source: 03-living-docs/patterns/Was-Cost-of-Task-Your-Moat.md public: true name: Was the Cost of That Task Your Moat? description: >- When automation makes a previously-expensive task cheap or free, the existential question for an incumbent is whether the cost of that task was the very thing protecting your business. If yes, your moat is now a puddle. type: pattern category: strategy origin: >- Benedict Evans, "AI eats the world" (May 2026 deck). Distilled by Dave May 18, 2026 from the Spring 2026 update. created: 2026-05-18T00:00:00.000Z tags: - pattern - ai - strategy - moats - disruption --- > **Core question:** When a new technology makes a previously-expensive task cheap or free, ask: *was the cost of that task what was protecting your business?* If yes, automation isn't a productivity win for you — it's an existential event. --- ## The articulation Benedict Evans, May 2026: *"The internet removed physical distribution costs that protected many industries from competition."* CDs, printing, cables, retail stores — for music, journalism, TV, retail — these weren't operating costs to be minimized. They were the *moat*. Once distribution went to zero, anyone could enter, and the incumbents' cost base collapsed into commoditization. The diagnostic question runs forward: - **Task gets automated to zero** → you have a productivity win. - **Task was your moat** → you have an existential threat. Most analyses stop at the first. The second is where strategic ruin lives. --- ## How to apply When evaluating an AI capability landing in your industry, run this filter on every task it touches: 1. **Was this task expensive enough to deter new entrants?** (If small teams couldn't afford to do it before, the cost was a barrier.) 2. **Did your customers pay you mostly because you absorbed that cost for them?** (vs. paying you for proprietary data, network effects, integration depth, brand, or compliance posture) 3. **If a new entrant could do this task for free tomorrow, would they need anything else you have?** If the answer to (3) is "no" or "very little," your moat *was* that cost base. The new entrant doesn't need to be better — they just need to exist. --- ## What survives when the moat-task gets automated The cost-moat collapses, but other moats can hold: - **Proprietary data** that the new entrant can't recreate - **Network effects** (two-sided, marketplace, data-network) - **Integration depth** in regulated or fragile systems (banking cores, EMRs) - **Compliance & audit posture** (cost+time to clear) - **Brand trust** in domains where errors are catastrophic - **Implicit knowledge / taste** (see [Task-vs-Job](/patterns/task-vs-job)) If any of these are real, the incumbent has a real shot at riding the wave. If the moat-task was *all* you had, the wave breaks on you. --- ## Failure modes - **"AI will make us faster"** — true but irrelevant if the work being made faster was your moat. Speed for everyone = no advantage for you. - **Defending the wrong layer** — pouring R&D into AI-assisting the task that's being commoditized. Defending the cost base instead of building the next moat. - **Missing the unbundling vector** — the new entrant doesn't have to *replace* you; they can split out the one piece that mattered and let the rest atrophy. --- ## Counter-question If your moat-task is being automated, what *new* moat does the automation enable that your scale, data, or position lets you build that the new entrant cannot? See [What-Was-Impossible-Now-Cheap](/patterns/what-was-impossible-now-cheap). --- ## Cross-References - Companion frames: [Task-vs-Job](/patterns/task-vs-job), [What-Was-Impossible-Now-Cheap](/patterns/what-was-impossible-now-cheap), [Absorb-Innovate-Disrupt](/patterns/absorb-innovate-disrupt) - Related: [Build-AI-Run-Deterministic](/patterns/build-ai-run-deterministic), [Bottleneck-Shifts-Upstream](/patterns/bottleneck-shifts-upstream)