Investing in the AI boom, with Confidence
Artificial intelligence is a real and lasting investment theme. For most people the question is not whether to have some exposure to it, but how to get that exposure without taking on a risk they cannot live with.
Two decisions do most of the work. The excitement around AI is not misplaced. It is one of those broad shifts, like the arrival of the internet, that plays out over years and touches almost every industry. Naturally, a lot of people want part of their portfolio pointed at it, and that instinct is a reasonable one.
What matters is the execution, because the very same theme can make you money or cost you a great deal, depending on two choices: how you hold it, and how much you put in.
How to hold it: own the basket, not the winner Start with how you hold it, because this is where the biggest mistakes are made. The temptation with any new theme is to pick the single company that will win it. It is understandable. A few names have delivered spectacular returns, and it can feel as though the winner is obvious. It almost never is, and the history of new technologies is brutal on those who try.
Consider the motor car. When it arrived, it was obviously the future, and it was. But being right about the technology gave investors no protection at all. The early industry grew from a few pioneers into hundreds of competing manufacturers, and then collapsed back to a handful. In the United States, an industry that once had hundreds of car makers ended up dominated by just three. Almost everyone who backed an individual early manufacturer lost their money, not because they were wrong about cars, but because they were wrong about which company. The technology won. Most of its shareholders did not.
The internet repeated the pattern a century later. The theme was real and it reshaped the world, yet most of the companies that looked like sure things at the time faded or failed. Being right that a technology matters does not tell you which company will capture the value from it, and the market has usually priced in the obvious candidates already. A basket sidesteps the problem entirely. Instead of needing to identify the winner, you hold a spread of the companies exposed to the theme. If the theme delivers, you benefit from it broadly. If any single company stumbles, and in a field that moves as fast as this one some always do, your outcome does not hang on it. You are backing the shift itself, not gambling on the one name that survives it.
The picks and shovels of AI
So a basket, but a basket of what? The most direct answer is semiconductors, the companies that make the computer chips and the ones that make the machines used to build them. To see why, it helps to notice what AI actually runs on. AI does not run on clever ideas alone. It runs on computing power, and computing power is built on chips. Every model that gets trained, every question answered, every data centre built to run AI, depends on enormous quantities of specialised semiconductors: the processors that do the calculations, the memory that feeds them and the parts that connect them.
The recent expansion of AI data centres has driven a sharp rise in demand for exactly these chips, and for the equipment used to make them. This is why semiconductors are called the picks and shovels of the boom. In a gold rush, the steadiest money was rarely made by any single prospector, but by the people selling the tools that every prospector needed. Chips are the tools of this one. Whoever ends up building the winning AI products, and that is precisely the thing nobody can reliably predict, they will all need chips to do it, and they will buy them from the same few industries. It is also a broad field rather than a single company. The chip designers, the manufacturers, the memory makers and the equipment firms together form a whole supply chain, which is part of what makes semiconductors well suited to being held as a basket in the first place.
Not all baskets are diversified
A basket is only diversified if it is built to be. The word fund, or ETF, makes people assume a spread of risk, but the label guarantees nothing on its own. Plenty of thematic funds are far more concentrated than they look, and a poorly built one can leave you with the very risk you were trying to avoid, exposure to a single company.
A few things separate a real basket from a concentrated bet dressed up as one. Count the holdings. A genuine basket needs enough companies in it to spread the risk. Some thematic funds hold only a handful of names, sometimes fewer than ten, which is not diversification, it is a concentrated bet with a fund wrapper around it. A properly diversified fund will typically hold thirty or more companies drawn from across the theme.
Check the concentration. The number of holdings is only half the story. A fund can hold dozens of names and still be dominated by one or two of them, if it simply weights each company by its size. When the largest holding is a fifth or a quarter of the entire fund, you are mostly betting on that one company, whatever else sits alongside it. The funds that are built well cap how much any single name can take up, so no one company decides your result. Look across the value chain. The point of backing a theme through a basket is to capture it broadly.
A good fund reaches beyond the obvious headline names into the suppliers and the equipment makers behind them, so you end up owning the theme itself rather than just its most famous stock.
A fund that is large, liquid, low in cost and open about what it holds is easier to own well and cheaper to carry over the years that a theme takes to play out. This is the sort of check we run on any fund before it goes near a client's portfolio, and it is where a good many thematic products quietly fall down.
How much: Enough to matter, not enough to hurt
The second decision is how much to hold, and here the theme's biggest attraction is also its main hazard. Investments tied to AI, and semiconductor shares in particular, are volatile. That is not a criticism of them, it is simply their nature. A fund in this area can double over a strong run and then give back half or more when the cycle turns, and both of those things will happen at some point. The technology can be advancing exactly as promised while the share prices fall sharply, because markets move on expectations, valuations and sentiment, not only on progress. That volatility is manageable if the position is the right size, and punishing if it is not. The real danger is not the fall itself. It is being forced to sell into the fall, either because you need the money or because you cannot bear to watch it.
Selling at the bottom is the one thing that turns a temporary drop into a permanent loss. For that reason we would treat AI as a satellite position and cap it at around 10% of a portfolio. That is enough to make a real difference to your returns if the theme delivers, without being so large that a sharp drawdown derails your plan or your nerves. The test is a simple one: choose a size you could watch fall by half and still hold, because at some point you may have to.
The discipline is the edge
None of this is especially exciting, and that is rather the point. The excitement is in the theme. The returns, if they come, will come from the discipline: spreading the money so you are not relying on any one company, and sizing it so you can live through the swings. Get those two things right and you can take part in the AI boom without it testing you in ways you never signed up for. This article is general information only and is not personalised financial advice. It does not take account of your particular circumstances, goals or risk tolerance, and overseas investments such as these can carry tax consequences for New Zealand investors that depend on your situation. Before acting, please speak to a licensed financial adviser about what is right for you.