Investing & portfolio
How to Invest in the AI Era Without Chasing Hype
How should you invest in the AI era?
Treat AI as a long-term shift that can reward patient investors, not as a reason to move your whole portfolio into one theme. The simplest approach is to keep most of your money in broad, diversified holdings and give AI a limited, deliberate slice, sized so that you could watch it fall by half and still stay on track with your goals.
A great technology and a great investment are two different things. The internet changed the world, yet investors who bought early internet winners at any price waited fifteen years just to get back to even. This article gives you a framework: define the terms, understand why AI matters, size your exposure with real numbers, buy in stages, and check a company before you buy.
Key terms: what do these words mean?
A few terms make the rest of this article easier to follow.
- Core-satellite: a large core of diversified, low-cost holdings plus a small satellite of higher-conviction themes.
- Concentration risk: too much of your wealth depending on one company, sector or idea.
- Valuation: how much you pay for each unit of earnings or sales. A great company bought at too high a price can still be a poor investment.
- Drawdown: the fall from a peak to a later low, for example -50%.
- Capex (capital expenditure): money companies spend on buildings, chips and equipment. Much of the AI boom is driven by capex.
- Dollar-cost averaging (DCA): investing a fixed amount at regular intervals so you buy more units when prices are low.
- Expected value: the probability-weighted average of possible outcomes, used to compare uncertain bets.
Why is AI treated as a long-term shift?
AI lowers the cost of tasks that used to need skilled people: writing, coding, analysing documents, designing, customer support, and parts of drug discovery and engineering. When a general-purpose technology makes many activities cheaper, it can raise productivity across almost every industry, much like electricity, the railway and the internet did.
Value, however, does not stay in one place. Some firms build the technology, some supply what it needs (chips, data centres, electricity), and many simply use it to cut costs or sell more. Others lose business to it. A diversified portfolio is already exposed to all three groups without you choosing individual winners, which is why the first step is to check what you already own.
How much AI exposure do you already have?
Most global and US index funds are weighted by company size, so the largest technology companies, several of them central to AI, take a big share. In many popular indexes the ten biggest companies make up more than a third of the fund. If you already hold such an index fund, you own AI exposure before adding a single AI-themed stock.
Adding a thematic AI fund on top often doubles up on the same companies. Before buying, look at the top ten holdings of every fund you own and note the overlap. If the same names appear everywhere, your real concentration is higher than your fund count suggests.
What does the dot-com crash teach us about AI investing?
The Nasdaq Composite peaked in March 2000 and then fell roughly 78% by 2002. It did not regain that peak until 2015, about fifteen years later. The internet itself was real and kept growing; what collapsed were prices that assumed perfect execution for every company, and many leaders of that era never returned to the top.
The maths of losses explains why this hurts so much. A 78% fall needs a gain of 1 / (1 - 0.78) - 1, about 355%, just to get back to even. At -50% you need +100%, at -33% you need +50%, and at -20% you need +25%. The deeper the fall, the longer the road back, so limiting how much you hold in any one hot theme matters more than predicting the winner.
Two differences today are worth noting. Many of the biggest AI spenders are highly profitable companies, unlike many 1999 start-ups with no revenue. But heavy capex still needs to earn a return, and prices can still run ahead of earnings. Both facts can be true at once.
How much of your portfolio should go into AI?
A common range is a satellite of 10% to 20% of investable money for a single theme like AI, with the rest in a diversified core. Younger investors with stable income and a long horizon can lean to the higher end; people who need the money within five years, or who already hold a lot of technology through index funds, should stay lower.
A simple test: write down the largest fall you could accept without changing your plans, for example 15% of your whole portfolio. Divide it by the fall you fear in the AI slice, say 50%. The answer, 15% / 50% = 30%, is the most you should put in that slice. Many people find their honest answer is lower than they first thought.
What does a core-satellite split look like with real numbers?
Take 10,000 to invest. Option A puts 8,000 in a diversified core and 2,000 (20%) in AI-related holdings. Option B puts 5,000 (50%) in AI.
If the AI part falls 50%: Option A loses 1,000, so the whole portfolio is down 10%. Option B loses 2,500, so the portfolio is down 25%, and that part needs a 100% gain just to recover. If the AI part instead doubles: Option A gains 2,000 (+20% overall) while Option B gains 5,000 (+50% overall).
Option B pays more if you are right but turns an ordinary correction into a loss many people cannot sit through. Option A keeps you invested for the upside while leaving you able to stay calm. Staying invested through a fall is what actually earns long-term returns, so the best position size is the one you can hold.
How do you think about AI outcomes with scenarios?
Nobody can forecast AI precisely, but you can test your plan against scenarios. Suppose your 20% AI slice could end in one of three ways over five years: a strong case (+100%, 25% chance), a base case (+30%, 45% chance) or a bust (-50%, 30% chance). The expected value is 0.25 x 100 + 0.45 x 30 + 0.30 x (-50) = +23.5% on the slice, but the range is wide.
Look at what each scenario does to the whole portfolio when the slice is 20%: +20%, +6% and -10%. A -10% hit is painful but survivable; at a 50% slice the bust becomes -25%. The probabilities are made up for illustration; the point is to see whether the worst case fits your plan, not to trust the average.
Should you invest in AI all at once or in stages?
Staged investing reduces the risk of putting everything in at a peak. Split the AI allocation into equal parts and invest monthly or quarterly, and add a rule for putting in extra if prices fall by set amounts.
Example of DCA: invest 500 a month for four months when the price is 100, 90, 80 and 100. You buy 5.00, 5.56, 6.25 and 5.00 units, 21.81 in total, for 2,000. Your average cost is about 91.72, so when the price returns to 100 your holding is worth about 2,181, a 9% gain even though the price ended where it started.
A staged-fall plan works similarly. With 10,000 reserved, you might buy 2,500 after a 10% fall (price 90), 3,500 after 20% (80) and 4,000 after 30% (70). Your average cost becomes about 77.72, so you break even once the price recovers just 11% from the low and you are in profit above 77.72, well before the original 100. If it does return to 100, the plan returns about 28.7%. The Drawdown planner does this arithmetic for any levels you choose.
How do you check an AI company before buying?
Ask these questions about any AI-related stock or fund:
- Is revenue growing, and is it real? Look for sales, not just announcements or pilot projects.
- Does it make money, or when will it? Check profit margins and cash flow, not only user growth.
- How much does it depend on one customer or one product? Concentrated customers raise risk.
- What is the capex burden? Heavy spending is fine if returns follow; check the trend over several years.
- What is the price? Compare valuation with growth and with the company's own history.
- What is the moat? Ask what stops a competitor, or a cheaper approach, from copying it.
How do you protect against AI disruption and a bust?
AI can hurt businesses as well as help them. A company whose main product is easily automated may lose pricing power, so avoid owning only one industry. Spread across sectors, regions and asset types such as bonds, gold and cash, which respond differently to technology cycles. Rebalance once or twice a year: if the AI slice has grown from 20% to 30%, trim back to target so one theme cannot dominate.
Your career is also an investment. If AI changes your field, learning to use it is the most direct hedge. Keep an emergency fund of three to six months of expenses so a job disruption never forces you to sell investments at a bad time.
What mistakes do investors make with AI stocks?
Most mistakes come from emotion and from ignoring price.
- Buying only because a stock is up and everyone is talking about it.
- Putting a large share of savings into one company or one theme.
- Ignoring valuation: a great business does not mean any price is fine.
- Owning several funds with the same top holdings and calling it diversification.
- Checking prices daily and selling after the first big fall.
- Assuming today's leaders will stay on top for decades; leaders change.
Try it: set your AI allocation in Eyrie
Use the Asset allocation calculator to see a sensible split across stocks, bonds, gold and cash for your risk level, then reserve your AI slice from the equity part. To plan staged buying, try the Drawdown planner, and to see what a given return means over time use the CAGR calculator. For which parts of the AI chain to focus on, read [Best AI sectors for long-term investing](/blog/best-ai-sectors-for-long-term-investing).
What to read next
Terms and topics worth searching to go deeper:
- Core-satellite portfolio construction
- Diversification and correlation
- Dollar-cost averaging vs lump sum
- Valuation: price-to-earnings and price-to-sales
- Technology bubbles: the dot-com era
- Rebalancing bands
- Market-cap weighted vs equal-weight indexes
Is this investment advice?
No. This article is for education only and is not investment, tax or legal advice. Allocation ranges, scenario probabilities and examples are illustrative, past performance does not guarantee future results, and you can lose money. Consider your own goals and risk tolerance, or speak to a licensed adviser, before investing.
Frequently asked questions
How much of my portfolio should be in AI?
A common range is 10% to 20% as a satellite, with the rest diversified. Lower it if you need the money soon, already hold a lot of tech through index funds, or cannot stomach a 50% fall in that part.
Is AI a bubble?
Nobody knows in advance. Prices can run ahead of earnings even when the technology is real, as in the dot-com era, so size positions so a deep fall would not derail your plans.
Should I buy AI stocks or an AI fund?
A diversified fund spreads the risk of picking wrong winners; individual stocks can pay more but carry much higher risk. Many investors use a fund for most of the AI slice.
Do I already own AI through index funds?
Very likely. Large technology companies are among the biggest holdings in most global and US index funds, so check your existing exposure before adding more.
Is it better to invest in AI all at once or gradually?
Gradual investing reduces the risk of buying at a peak and is easier to stick with. Lump sums can do better in rising markets, but staging is a sensible default for a volatile theme.
Get new posts in your inbox
We email you whenever a new post goes live. No spam, unsubscribe anytime.
Track your portfolio with Eyrie
Get the Android app to manage holdings, trades and market research in one place.
