Banks Can't Scale Like Tech | Generated by AI
Question: Why can’t big banks (100+ year history) grow as large as Nvidia or Apple? Is it fundamentally about greed vs. mission?
Answer:
This is a real tension worth thinking through carefully, but the framing needs some pressure-testing.
The structural reality first
Banks are regulated utility-like entities by design. After 1929, after 2008, regulators deliberately constrained how large and how risky banks can be. Capital requirements (Basel III), leverage limits, stress tests — these are enforced ceilings on growth. JPMorgan can’t 10x its balance sheet even if it wanted to. Nvidia has no such constraint.
Banks also operate in a zero-sum or near-zero-sum market. Lending money, taking a spread — that scales with the economy, not ahead of it. Nvidia creates new value by enabling compute that didn’t exist before. That’s the compounding asymmetry.
The greed framing is partially right, but incomplete
Banks were designed to intermediate capital — take deposits, lend at margin, repeat. That’s not inherently greedy, it’s a function. The problem is:
- When that function got financialized (derivatives, CDOs, prop trading), the greed part became dominant
- Investment banks like Goldman are essentially hedge funds with a banking license — maximizing extraction, not creation
- The “mission” was always thin: profit from information asymmetry and capital access
Compare that to Nvidia’s actual bet: Jensen Huang committed to CUDA in 2006 when nobody cared about GPU compute. That was a genuine technological conviction, not extraction.
But don’t romanticize Apple or Nvidia too much
Apple’s supply chain runs on Foxconn labor conditions that are… not “think different.” Their App Store is a 30% extraction tax on developers. Steve Jobs was famously ruthless and often wrong about markets.
Nvidia right now is also extractive — H100 pricing is not philanthropy. They have leverage and they use it.
The difference is: extraction built on top of genuine capability creation vs. extraction on top of regulatory moats and information asymmetry. That’s the real distinction.
Why banks can’t scale like tech
| Factor | Banks | Nvidia/Apple |
|---|---|---|
| Product | Money (commodity) | Compute / devices (differentiated) |
| Margin source | Spread + fees | IP + platform lock-in |
| Regulatory ceiling | Hard cap on leverage/risk | Essentially none |
| Network effects | Weak | Strong (CUDA ecosystem, iOS) |
| Innovation compounding | Near zero | Exponential |
| Mission | Preserve capital, extract spread | Enable new categories |
The deeper point
Banks are trapped in linear scaling — more assets, more revenue, but proportionally more risk and regulation. Tech compounds. A GPU architecture improvement multiplies across every customer simultaneously. A loan doesn’t compound the bank’s capability.
The 100-year-old bank’s problem isn’t that its heart is greedier than Nvidia’s. It’s that it chose a business model that extracts from existing value rather than creating new value, and then built a century of institutional inertia around that model.
For your trajectory: this is exactly why AI-native companies will eat financial services. The bank you work at now has 100 years of institutional knowledge and regulatory relationships — and a startup with the right AI agent stack can replicate the value-delivery part with 10 people. The moat isn’t as deep as they think.