You have five stocks. They’re different companies — Apple, Microsoft, Google, NVIDIA, Meta. Different tickers, different boardrooms, different products. How concentrated can you really be?

Pretty concentrated, as it turns out. And the danger isn’t in the number of holdings — it’s in what those holdings have in common.

Two Types of Concentration

Most investors understand ticker-level concentration: having 40% of your portfolio in a single stock is obviously risky. A bad earnings report, a product recall, a regulatory fine — one event, one massive hit.

Far fewer investors understand thematic concentration: your portfolio being exposed to a single macroeconomic theme across multiple positions that look, on the surface, completely unrelated.

This is the harder one to see. It requires you to look at your holdings not as companies, but as bundles of economic exposures. And it’s exactly what happened in a mega-cap tech portfolio we analyzed last month.

The AI Infrastructure Portfolio: A Case Study

Imagine a portfolio with five positions:

  • NVDA (3% weight) — semiconductors
  • MSFT (6% weight) — cloud infrastructure
  • GOOGL (4% weight) — cloud infrastructure
  • AAPL (7% weight) — consumer tech
  • META (2% weight) — social media

Twenty-two percent of the portfolio. Five tickers. Looks reasonably diversified — or at least not reckless.

But when you map each position to its economic themes, a different picture emerges. NVDA makes AI chips. MSFT and GOOGL are the largest buyers of those chips for cloud compute. META is deploying $60 billion into AI compute this year. AAPL is integrating AI into every hardware layer from iPhone to iPad to Mac.

All five are structurally linked to a single macro theme: AI infrastructure spending.

The actual thematic concentration across those five positions: 73%.

Why This Matters

When a single macro theme drives 73% of your portfolio, a single market event can move everything simultaneously — in the same direction. A few examples:

AI capex correction: If Microsoft, Google, or Meta slow their data center buildouts — even by 10-15% — NVIDIA’s revenue guidance falls immediately. The hyperscalers are NVDA’s biggest customers. Their budgets compress, NVDA’s revenue follows. Apple’s AI hardware roadmap also compresses. All five positions move down together.

Chip export restrictions: NVIDIA’s H100 and H200 chips are subject to US export controls on China. An escalation doesn’t just hit NVIDIA — it delays Google and Microsoft data center builds that depend on next-generation silicon. Correlated downside, again.

Rate hike cycle: Higher-for-longer rates compress growth multiples across all five names. Consumer discretionary (AAPL hardware cycles) also slows under rate pressure. The correlation holds through different mechanisms, producing the same result.

The Correlation Coefficient Problem

A healthy, diversified portfolio has low correlation between positions — typically under 0.35. When themes are well-separated, one position getting hit doesn’t mean others follow.

That mega-cap tech portfolio had an intra-portfolio correlation of 0.84. That means the five holdings were moving together more often than not — which is the opposite of diversification. It’s essentially five positions that behave, in aggregate, like one very large position in AI infrastructure.

How to Know If You’re Hidden-Concetrated

The question isn’t “how many stocks do I own?” It’s “what economic themes are driving my returns?”

A few signals that your portfolio may have hidden thematic concentration:

  • Your portfolio has 8+ stocks but they’re all in the same 2-3 sectors
  • Your holdings share a major customer-supplier relationship
  • You hold multiple ETFs (SPY, QQQ, VGT) alongside individual tech stocks, which amplifies tech exposure rather than diversifying it
  • Your sector allocation chart shows 60%+ in a single sector

Thematic analysis — looking at what themes your holdings are exposed to and how much — is the only way to see concentration risk that ticker-level analysis misses.

The Fix

Reducing thematic concentration doesn’t mean abandoning tech. It means being intentional about theme separation:

  • Add positions with inverse or uncorrelated economic drivers (energy infrastructure that benefits from AI power demand; defensive sectors with recurring revenue)
  • Distribute single-theme exposures across the supply chain rather than concentrating in one name
  • Analyze what’s inside your ETFs, not just the ETF itself — a QQQ holder often has more NVDA exposure than they realize

The goal isn’t zero AI exposure. It’s knowing how much you have, and whether that’s intentional.