Early indicator of AI labor impact
Speakers
Deirdre Bosa — TechCheck Anchor/Reporter
— CNBC anchor, tech reporter
Summary
- The "Halo Trade" Rotation: Capital is aggressively rotating out of "Knowledge Economy" stocks and into the "Physical Economy" (Heavy Assets, Low Obsolescence). Goldman Sachs' basket for this trade has outperformed capital-light names by 25% YTD.
- AI Exposure Data: Anthropic's internal data reveals high AI penetration in coding (75%), customer service (70%), and financial analysis (60%). Conversely, construction, agriculture, and transportation have near-zero penetration.
- The "Silent" Labor Crisis: Unemployment isn't spiking because companies aren't firing experienced staff; instead, they have frozen entry-level hiring. Hiring for 22-25 year olds in AI-exposed sectors has dropped 14% since ChatGPT's launch.
- Enterprise Winners: The displacement of human labor in legal, sales, and finance maps directly to enterprise software platforms (Salesforce, Intuit, etc.) integrating AI to automate those specific workflows.
Trade Ideas
Bosa cites the "Halo Trade" (Heavy Assets, Low Obsolescence), noting that capital is fleeing the "Knowledge Economy" for the "Physical Economy." She explicitly states sectors like "Construction, Agriculture, Transportation" have near-zero AI penetration. As AI uncertainty creates volatility in services and tech labor, investors are seeking safety in tangible industries where human labor cannot be digitized. Caterpillar (Construction), Deere (Ag), United Rentals (Equipment), and Union Pacific (Transport) are the blue-chip proxies for this "Physical Economy" safety trade. LONG. These sectors are insulated from the deflationary pressures of AI labor displacement. A broader economic recession would hurt cyclical industrials regardless of their AI immunity.
The Anthropic study shows "Customer Service Reps" have 70% of their tasks covered by AI today. Bosa highlights this as one of the highest displacement risks. TTEC and Concentrix (CNXC) are Business Process Outsourcers (BPOs) whose primary revenue model relies on billing for human customer service agents. If 70% of those tasks are automated, their "seat-based" revenue model faces an existential crisis. SHORT. This is the direct counter-party to the efficiency gains seen by software companies. These companies successfully pivot to becoming "AI managers," though their margins would likely compress significantly.
Bosa notes that AI's theoretical capabilities in "Legal, Sales, Finance" map "almost perfectly to where enterprise deals are actually landing." She explicitly names Salesforce, Intuit, DocuSign, and Thomson Reuters as companies doing deals with AI labs. While AI destroys *jobs* in these sectors (admin, junior analysts), it accrues *value* to the software platforms that host the AI agents doing the work. These companies are not the victims of displacement; they are the vendors selling the efficiency. LONG. These companies are effectively capturing the wages previously paid to entry-level knowledge workers. AI models becoming good enough to bypass these legacy software "wrappers" entirely (e.g., AI writing code directly rather than using a SaaS tool).
