Wall Street's valuation models are pricing in last decade's productivity growth while companies are already running on this decade's AI infrastructure.
The Summary
- HSBC's global CIO Willem Sels says US stocks look undervalued because current valuations don't reflect the scale of AI-driven productivity and earnings gains already underway
- Corporate resilience has exceeded expectations, with AI productivity gains making Sels more bullish on US markets than Europe
- Thesis: The market is still using old metrics to price companies experiencing new-paradigm efficiency
The Signal
Willem Sels thinks the market is looking at the wrong numbers. While analysts debate whether tech stocks are overpriced, HSBC's global chief investment officer argues the opposite: US equities are actually cheap relative to what's happening inside these companies. The productivity surge from AI deployment isn't showing up properly in traditional valuation models yet. Those models were built for incremental improvements. AI is delivering step-function changes.
This isn't theoretical. Sels points to corporate resilience that's already surprised markets. Companies are doing more with less, margins are holding, and the productivity gains are compounding. The gap between US and European markets reflects this reality. American companies moved faster on AI infrastructure. They're seeing returns now.
"The economy and corporates have proved more resilient than people thought."
Key factors driving the thesis:
- AI productivity gains are measurable, not speculative
- Earnings growth is accelerating from efficiency, not just revenue
- Traditional valuation multiples don't capture structural productivity shifts
The timing matters. We're past the "AI will change everything" phase and into the "AI is changing everything" phase. The companies that spent 2023-2025 building agent infrastructure and workflow automation are now reporting the numbers. Revenue per employee is climbing. Customer service costs are dropping. Software that used to require teams now requires oversight. These aren't projections. They're trailing twelve-month figures.
Sels is essentially saying the market is mispricing the present, not the future. When analysts say tech stocks are expensive, they're comparing current prices to historical earnings patterns. But the companies aren't following historical patterns anymore. They're operating on different economics.
The Implication
If Sels is right, the next year will reveal which companies have real AI productivity gains versus which ones have AI press releases. Watch earnings calls for specifics: headcount trends relative to revenue growth, margin expansion stories, automation metrics. The companies showing measurable productivity improvements will reprice upward as the market catches up to the new operational reality.
For builders and workers, this confirms what's already visible on the ground. The agent economy isn't coming. It's here, it's working, and it's starting to show up in corporate financials. Position accordingly.