The 100-Year Pattern — Part III: The Half-Life of Your Business

Every business model has a half-life. For most of the twentieth century, that half-life was measured in decades. AI is compressing it. The uncomfortable question is not whether this applies to your industry. It is which quadrant you are in — and whether you are seeing it clearly.

Part II ended with a number. 88% of organizations use AI. 5% have integrated it at scale. That gap isn't about tools. It's about where the compression lands — and how fast it finds you.

Where it finds you depends on two things: how much of your core work AI can automate, and how strong a structural position you hold over AI-enabled competitors. The intersection produces four positions. Each with a different half-life.

Before you see the map: most people reading this will self-select into Sweet Spot or Protected before they reach the quadrant descriptions. Surveys that measure self-reported AI maturity against operational reality consistently find a gap of 40 to 60 percentage points. The most exposed categories are usually the most confident. The map doesn't take your self-assessment into account. It uses two variables. The answer may not be the one you expect.

If you placed yourself in Sweet Spot or Protected while looking at the map — read Under Attack and Countdown carefully before confirming. The self-assessment bias runs in one direction.

Sweet Spot. AI does the work. Your moat keeps competitors out. This is the only position where the wave carries you rather than hits you. Apple is the clearest example: hardware, software, and brand compound into a moat no competitor can cheaply copy, while AI reduces every layer of operating cost. A diagnostic clinic built on ten years of proprietary patient data. A niche marketplace where the data compounds with every transaction. What these have in common: AI reduces operating cost while the moat makes that cost advantage permanent. These are the businesses compounding while others debate. They are the 5%. Or they are actively becoming them.

Protected. AI assists. It cannot replace. Electricians and plumbers who work on-site with liability attached to every job. Surgeons. Architects on complex briefs. Regulated advisory where professional accountability sits between the output and the client. These businesses have a physical presence or a regulatory structure AI cannot currently cross. That layer holds. What it cannot do is generate upside. The AI wave that compounds margins in Sweet Spot doesn't reach here. Protected is not the same as surfing. It is standing on dry land, watching Under Attack fight for air in the waves below.

Under Attack. AI-enabled competitors are cutting your price floor. Full-service performance marketing agencies without proprietary audience data. Generalist recruiters without exclusive candidate relationships. Law firms billing by the hour for standard contract review and due diligence. Accounting firms processing standard compliance and tax filings without advisory depth. Consulting projects that are 80 percent slide assembly and 20 percent insight. These businesses are not being eliminated. They are being undercut. AI-enabled competitors can now deliver equivalent output at 40 to 60 percent lower price. The asymmetry that made them valuable, "we have expertise you don't," is closing. Sam Altman: 95% of what marketers use agencies for will be handled by AI, nearly instantly, at almost no cost. Satya Nadella on SaaS: "Applications are essentially databases with business logic. In the future, this logic will migrate to AI agents." These aren't predictions. They are descriptions of decisions already made and being executed. The diagnostic: if you cannot state clearly why a client could not get the same output from a different provider using current AI tools at 40 percent lower cost, you are likely here. The window to reposition is narrowing. Based on enterprise AI deployment rates, it is measured in quarters.

Countdown. The assumption is that AI replaces the cheap work first. It doesn't. Anthropic's economic research, drawn from over a million real Claude conversations, shows the tasks AI handles most efficiently are mid-complexity, high-volume tasks equivalent to 14 years of education on average. Not the bottom of the expertise stack. The center. Generic press release writing. Standard market reports. First-tier IT support. Translation at volume. Graphic design for standard marketing assets. Journalism covering templated beats — earnings reports, match results, weather. These are not low-skill tasks. They are the revenue core of every Countdown business. AI is not coming for the floor. It is removing the ceiling. The center is already under pressure. Gradual is not a strategy. Not a new plan. A different business.

Under Attack and Countdown are not fixed outcomes. The businesses that look back at this period as an opportunity are the ones that used the pressure as a forcing function — not to optimize the existing model, but to build a different one. That move is only available now.

Why the window is narrow. Goldman Sachs identified a peak displacement period in 2026 for administrative functions, customer support, and first-level professional services. Peak displacement is the year when the half-life of most service businesses expires — when AI handles the highest share of those categories and forces the largest number to reprice or reposition. It is not a finish line. It is the year the pricing pressure goes from theoretical to structural. Peak displacement in 2026 produces pricing pressure in 2027, when AI-native competitors can undercut at scale and force consolidation. By 2028, the new cost floor is the market baseline. Businesses that exist after that transition crossed to a different model before the floor was set.

Here is the timeline.

In a compressed wave: 2026 is 1996. 2028 is 2006. In the internet era, businesses that waited until 2001 were five years behind those who moved in 1996. The ones who moved in 2006 were catching up to a market that had already restructured. The window is not a decade. It is months.

Your position doesn't tell you what to do. It tells you how fast, and how radically.

Protected: build the AI layer now, before the junior pipeline dries up. Sweet Spot: the wave is compressing your competitors' margins. Use that window to deepen your proprietary layer before the advantage becomes table stakes. Under Attack: the move is toward proprietary. Exclusive data, client relationships that transfer, process IP. Not faster-generic. Structurally different. Countdown: what this means in practice is walking away from a revenue model, not adding a line item to it.

I ran this analysis on my own situation in late 2025. The map gave me a clear answer. I acted on it.

The window is not closing slowly. Your half-life started before you read this. The question is whether you have used it.

Sources: Goldman Sachs AI labor market analysis, McKinsey Global Institute, WEF Future of Jobs Report 2025, OECD Employment Outlook 2023. CEO quotes from public interviews and published statements. Timeline estimates are the author's own based on historical GPT adoption patterns and publicly available institutional forecasts. Not financial advice.