Two Tearsheets Worth of Warning Signs
Our July 13–17 Weekly Tearsheet flagged a semiconductor bear market: the VanEck Semiconductor ETF (SMH) down 9%, the Philadelphia Semiconductor Index down more than 20% from its high. Our July 20–24 Tearsheet flagged something adjacent: Tesla down 14.5%, Alphabet down 7% after raising its own capital expenditure guidance to $195–205 billion — money largely earmarked for AI infrastructure.
Both moves trace back to the same underlying industry, and we've now referenced it twice without fully explaining what it actually is. That's the gap this issue closes.
"You don't need to pick a side in the AI trade. You do need to understand the machine underneath the index funds you already own."
What a Chip Actually Is, in Plain Terms
A semiconductor — a "chip" — is a small piece of engineered material, almost always silicon, etched with billions of microscopic switches called transistors. Those switches turn electrical signals on and off in patterns that add up to computation: every calculation your phone, your car, your bank's servers, or an AI model runs happens because transistors on a chip are flipping on and off, billions of times per second.
What makes this industry different from almost any other global supply chain is how few companies can actually build the most advanced chips. Designing a chip and manufacturing it are two entirely separate skills, done by different companies, often on different continents — and that separation is the whole reason this trade is global instead of domestic.
The Map of Money — Who Actually Holds the Value
As of mid-July 2026, the global chip industry carried a combined market value of roughly $18.8 trillion. That number is concentrated in ways most investors would not guess.
| Region / Company | Share of Global Chip Value | Approx. Value | Role |
|---|---|---|---|
| United States | ~62% | — | Design leadership (Nvidia, AMD, Broadcom) |
| Nvidia alone | ~27% | ~$5.0T | One company, over a quarter of the entire industry |
| Taiwan | ~14% | TSMC ~$2.1T | Advanced manufacturing (foundry) |
| South Korea | ~11% | Samsung + SK Hynix ~$2.0T combined | Memory chips (DRAM) |
| Rest of World | ~13% | — | Equipment, materials, smaller foundries |
Two things jump out. First, Nvidia's single-company share of the entire global chip industry is larger than South Korea's combined national share — a level of single-name concentration that's unusual for an industry this size. Second, the country that actually manufactures the most advanced chips in the world, Taiwan, isn't the country that captures the largest share of the industry's market value. Design captures more value than manufacturing does, and that split is the key to understanding who the real players are.
The Four Jobs in the Chip Supply Chain
Nearly every company in this trade does one of four jobs. Knowing which job a company does tells you what actually drives its stock.
1. Designers — Nvidia, AMD, and Broadcom design the chip architecture but don't own factories. Nvidia's dominance in AI-specific chip design (graphics processing units, or GPUs, repurposed for AI computation) is the single biggest reason it captures 27% of the entire industry's value.
2. Foundries — Taiwan Semiconductor Manufacturing Company (TSMC) actually manufactures the chips that Nvidia and others design. TSMC doesn't design its own competing chips; it builds other companies' designs at a scale and precision almost no one else can match. This is why Taiwan matters so much geopolitically — an outsized share of the world's most advanced chip manufacturing runs through one company on one island.
3. Memory Makers — Samsung and SK Hynix build DRAM (dynamic random-access memory), the fast, short-term memory chips that feed data to processors. AI workloads have driven a memory price squeeze — demand has outpaced supply — which is a direct part of why South Korea's combined share has been climbing.
4. Equipment Makers — companies that build the machines the foundries use to actually manufacture chips. Without this small handful of specialized suppliers, none of the above three groups can produce anything at all — arguably the single most irreplaceable link in the entire chain.
Why Hyperscaler Capex Is Suddenly the Story
The other half of what moved Tesla and Alphabet in our July 20–24 Tearsheet isn't the chipmakers themselves — it's the companies buying the chips. Microsoft, Alphabet, Amazon, and Meta — often called "hyperscalers" for the scale of the data centers they run — have been spending enormous, and rapidly rising, sums building out AI infrastructure. Alphabet's guidance raise to $195–205 billion is capital expenditure: money spent building data centers, buying chips, and constructing the physical infrastructure AI models run on.
Is This Investment, or Is This a Bubble?
There are two honest readings of hyperscaler capex right now, and reasonable people disagree on which is correct. One view: this is the early infrastructure-building phase of a genuinely transformative technology, similar to the buildout of railroads or the internet's physical backbone — expensive now, foundational later. The other view: spending is running ahead of proven revenue, hyperscalers are locked into a capacity arms race with each other more than with actual demand, and depreciation on this hardware will eventually catch up to earnings. Both readings are held by serious, informed investors. Neither is settled, and we're not going to pretend otherwise.
What is settled: when a company that size raises or cuts capex guidance, it moves not just its own stock, but the stocks of every chip designer, foundry, and equipment maker that capex is headed toward — and by extension, every index fund holding those names.
Where This Touches an Income-Focused Portfolio
We're deliberately not leading with this section — understanding the trade matters more than reacting to it, and that's the point of this issue. But it's worth a brief, honest note: the chip trade isn't confined to funds with "semiconductor" or "tech" in the name. Broad growth and technology index funds — the Invesco QQQ Trust (QQQ), the Technology Select Sector SPDR Fund (XLK), the Communication Services Select Sector SPDR Fund (XLC) — all carry meaningful, sometimes heavily concentrated, exposure to the handful of companies discussed above.
Concentration Risk Is a Diversification Question, Not a Panic Question
The dividend engine approach — SCHD, VYM, and the broader four-engine thesis — carries meaningfully less single-industry concentration than a pure growth-index approach, simply because dividend-quality screens tend to spread across sectors rather than chase whichever industry is compounding fastest right now. That's not a prediction about where the chip trade goes next. It's a structural reason an income-first portfolio experiences this kind of story differently than a growth-heavy one does.
A full breakdown of exactly how much chip-sector exposure sits inside specific funds is a worthwhile follow-up — not something to compress into the tail end of an explainer piece. Consider this the setup; the sleeve-by-sleeve version can be its own issue if there's appetite for it.
Where This Leaves You
The global chip industry isn't a side story anymore — at $18.8 trillion, with a single company representing more of it than an entire country, it's load-bearing infrastructure underneath most growth portfolios, whether an investor chose that exposure directly or picked it up through a broad index fund. You don't need to bet on Nvidia, TSMC, or the hyperscalers to be affected by what happens to them.
Understanding the machine doesn't tell you where it goes next. It tells you what you're actually holding, which is the more useful thing to know first.