The No-Hype Investor

Simple investing insights for long-term investors

This Week’s Big Idea

This Week’s Big Idea: Don’t Bet the Entire AI Boom on One Company

When people think about investing in AI, they usually start with the most obvious winners.

That makes sense. Companies like Nvidia have become the face of the AI boom because they provide the chips that power a huge part of the AI buildout. But the bigger lesson for long-term investors is this: AI is not being built by one company. It is being built by an entire supply chain.

That is why I think it is important to avoid looking at AI as a one-stock opportunity.

The AI boom has several bottlenecks. First, companies need compute. Then they need advanced chips to be manufactured. Then they need semiconductor equipment, memory, networking, custom silicon, power, cooling, and the infrastructure to support massive data centers.

If one part of that chain cannot keep up, it can become the next major opportunity.

That is where an ETF like SMH becomes interesting. SMH is not just a bet on one AI company. It gives investors exposure to several parts of the semiconductor ecosystem, including chip designers, manufacturers, equipment companies, memory companies, and networking/custom silicon businesses.

The main idea is simple: instead of trying to guess the one AI winner, investors can own several companies tied to the bottlenecks that make AI possible.

That does not mean SMH is risk-free. It is concentrated, semiconductor stocks are cyclical, and the ETF can be volatile. But for investors who believe AI demand will continue to grow, it can be a useful way to get exposure to more than one part of the AI supply chain.

The question is not just, “Which AI company wins?”

The better question is: which bottlenecks have to be solved for AI to keep growing?

-Alex

What I’m Watching: The AI Bottlenecks Inside SMH

When I look at SMH, I do not just see a semiconductor ETF. I see a basket of companies tied to different parts of the AI buildout.

1. Compute

This is the most obvious bottleneck.

AI models need a massive amount of computing power, and that is why Nvidia has become one of the most important companies in the market. AMD is also trying to compete in AI accelerators, while other companies are designing custom chips for specific workloads.

This part of the market still matters, but investors should remember that compute is only one piece of the AI story.

2. Chip Manufacturing

Even if demand for AI chips keeps growing, those chips still need to be manufactured.

That is why companies like TSMC are so important. The world can design advanced chips, but if there is not enough advanced manufacturing capacity, the entire AI buildout can slow down.

This is one of the reasons I like looking beyond the obvious AI names. Sometimes the company enabling the winner is just as important as the winner itself.

3. Semiconductor Equipment

Before chips can be manufactured, companies need the equipment to make them.

This includes the businesses involved in lithography, deposition, etching, inspection, and process control. These companies may not always get the same attention as the biggest AI chip designers, but they are critical to the entire semiconductor industry.

As chips become more advanced, the equipment needed to produce them becomes even more valuable.

4. Memory and Networking

AI is not just about raw processing power.

These systems also need memory, bandwidth, and the ability to move massive amounts of data quickly. As AI models get larger and data centers become more complex, memory and networking can become major constraints.

That is why companies tied to memory, networking, and custom silicon deserve more attention. The next big AI bottleneck may not be the GPU itself. It may be everything around the GPU.

ETF of the Week:

VanEck Semiconductor ETF

Ticker: SMH

  • What It Does:

    SMH owns a basket of semiconductor companies.

    That includes businesses involved in chip design, chip manufacturing, semiconductor equipment, memory, networking, and other parts of the chip supply chain.

    In simple terms, SMH gives investors exposure to the companies helping power AI, data centers, cloud computing, and other major technology trends.

  • Why it Matters:

    AI is not built by one company.

    It needs GPUs, foundries, equipment, memory, networking, and custom chips. SMH gives investors exposure to several of those bottlenecks in one ETF.

    That is the main appeal: instead of betting only on one AI winner, SMH gives you a broader way to invest in the semiconductor side of the AI boom.

  • Bull Case:

    If AI infrastructure spending keeps growing, semiconductor demand should remain strong.

    SMH could benefit because it owns companies across multiple parts of the chip ecosystem, not just one stock.

  •  Bear Case:

    SMH is still concentrated in one industry.

    Semiconductors can be cyclical, volatile, and sensitive to high expectations. If chip stocks pull back, SMH can fall hard too.

    There is also overlap risk. If you already own Nvidia, AMD, Broadcom, TSMC, or other chip stocks, SMH may make your portfolio more concentrated than you realize.

  • Long Term View:

    I would think of SMH as a satellite position, not the core of a portfolio.

    It can be a useful way to add AI semiconductor exposure, but I would still want the foundation of the portfolio to be broad and diversified. Long term though I can see continued success for SMH.

Reader Question

Question: “How much should I invest in SMH?”

My answer: I would think of SMH as a satellite position, not the foundation of the portfolio.

Because SMH is concentrated in one industry, I would not want it to become too large of a position. Semiconductors can be one of the most important long-term growth areas in the market, especially with AI, but they are also cyclical and volatile.

For most long-term investors, the core of the portfolio should still be broad and diversified. That could include something like an S&P 500 ETF, a total market ETF, or another broad-market fund.

Then SMH can sit around that core as a targeted bet on the semiconductor industry.

Personally, I would not want SMH to be more than 20% of a portfolio. And for many investors, something smaller may make more sense depending on risk tolerance, time horizon, and how much semiconductor exposure they already have through other stocks or ETFs.

The key is to avoid accidentally doubling up.

If you already own Nvidia, AMD, Broadcom, TSMC, or other semiconductor stocks individually, then adding SMH gives you even more exposure to the same theme. That can be fine if it is intentional, but it should not happen by accident.

So my framework would be:

Use broad-market ETFs as the core. Use SMH as a focused AI semiconductor satellite. And keep it at 20% or less so one sector does not control the entire portfolio.

James’ Corner

When I first started investing, I had this grand idea: I was going to build my own custom ETF. I’d pick the stocks, weigh the ratios, and manage it all myself. And sure, you can do that—but it is a massive headache to manage. What I eventually realized is that finding a few existing ETFs in sectors I already believe in does the exact same thing, without the full-time job of managing it. You still get the risk and the reward, but it smooths out the extreme volatility. For me, it's the perfect way to keep my portfolio in check

I- James