AI Labor Displacement: Investment Valuation Analysis
How We Built This Model
Forecasting AI’s impact on employment requires navigating considerable uncertainty. We’ve attempted to ground this analysis in where AI capabilities actually stand today and where they’re realistically heading over a 15-year horizon.
What we considered for each job category:
- Current AI capability vs. the full scope of the role
- The “corner case” problem—routine tasks automate easily, but exceptions and edge cases dominate what remains
- Public acceptance—will customers and businesses actually adopt AI in these roles?
- Physical vs. cognitive tasks—robotics lags behind software AI
- Regulatory and liability constraints—some industries move slowly regardless of technical capability
- Legacy infrastructure—new systems deploy AI easily, retrofitting old ones is slow and expensive
Where this led us:
Rather than accept headline figures of 30%+ displacement (which often measure “exposure to automation” rather than actual job losses), we’ve worked through each category individually. Many roles already experienced automation waves—bookkeepers to QuickBooks, bank tellers to ATMs, travel agents to online booking. What remains are the harder tasks.
The Conclusion
- 400 million jobs displaced globally over 15 years
- $670 billion/year in new savings each year, compounding
- $10 trillion/year ongoing savings at full deployment (Year 15)
- $80 trillion cumulative savings over the transition
At 8% value capture with 30% net margins, the AI ecosystem reaches fair value (20x P/E) by Year 12-13—roughly 2037-2038.
| Segment | Current P/E | Year to 20x |
|---|---|---|
| Memory | 25x | Now |
| Infrastructure | 30x | Year 3-4 |
| Networking | 35x | Year 5-6 |
| Cloud | 45x | Year 8-9 |
| Chips (Nvidia) | 60x | Year 11-13 |
| Application Layer | 80x | Year 10-12 |
| Pure-Play AI | 100x+ | Year 14+ |
Displacement by Category
- Customer Service (55% displaced) — routine calls automated, angry customers still want humans
- Administrative (45%) — already gutted by software, AI takes the remainder
- Retail (40%) — checkout goes, floor staff stay
- Manufacturing (45%) — new plants automated, legacy slow to retrofit
- Transport (35%) — highways yes, last-mile and regulation lag
- Professional Services (30%) — junior roles compress, seniors gain leverage
- Content/Creative (45%) — commodity automated, premium stays human
The Numbers
Fair value timeline: Memory and infrastructure now. Cloud by Year 8-9. Nvidia by Year 11-13. Pure-play AI requires the full 15-year thesis.
Discover more from Priory House
Subscribe to get the latest posts sent to your email.