Photo by Jakub Żerdzicki on Unsplash
Editorial commentary based on reporting by refresh. Research current as of September 5, 2026.
Bottom Line
Here is a question that flips the usual framing: what if dividend ETFs did not lose the AI capex boom, but were never entered in that race at all?
That distinction matters more than the performance tables suggest. According to refresh, the debate over whether income-oriented strategies still make sense against AI growth has intensified because equity returns in 2024-2025 became increasingly concentrated in a handful of mega-cap AI-infrastructure names. But the real story is not that dividend ETFs picked the wrong stocks — it is that the index rules behind funds like SCHD and VYM make it mechanically impossible for them to own the winners of a capex boom, and that same rule is what protects you when the boom cools.
As of September 5, 2026, according to the research underpinning this analysis, the AI capital-expenditure boom refers to massive spending by hyperscalers — Microsoft, Alphabet, Amazon, and Meta — on data centers, GPUs, and AI infrastructure. That spending has driven mega-cap tech returns through 2024 and 2025. Meanwhile, dividend ETFs such as SCHD, VYM, NOBL, and DVY tilt toward value and quality sectors: financials, consumer staples, healthcare, and energy. Sectors with, as the research puts it plainly, relatively low direct AI capex exposure.
Two different objectives. One scoreboard. That is the whole confusion.
What's on the Table: The Screen That Excludes the Winners
Most coverage frames this as a stock-picking failure. It is not. It is a definitional one.
A dividend ETF is built from a rule, not a hunch. To enter the index, a company generally has to pay a meaningful and often growing dividend. NOBL, for instance, is built around a multi-decade dividend-growth requirement. And here is the mechanical trap the surface reporting keeps skipping: the research states that because most AI-leading companies reinvest cash into capex rather than paying high dividends, dividend ETFs structurally underweight the very stocks benefiting most from the AI boom.
Read that again with the emphasis where it belongs. Structurally. Not accidentally.
A company pouring cash into GPU clusters and data-center construction is, by definition, not returning that cash to shareholders as a large dividend. The capex boom and a high dividend yield are close to mutually exclusive uses of the same dollar. So a screen that says "show me companies returning cash to shareholders" will systematically filter out companies whose entire strategy is not returning cash to shareholders. The fund did not miss Nvidia. The fund's rulebook was never allowed to see it.
The skeptic's pushback is fair and worth naming: if the rule produces a bad outcome, is it not still a bad rule? Our read is that this depends entirely on which question the investor was asking. If the goal was maximum total return over 2024-2025, yes — the rule cost money. If the goal was a predictable income stream that does not evaporate in a drawdown, the rule did exactly what it was hired to do. Judging a smoke detector by how well it toasts bread is not an indictment of the smoke detector.
In Plain Terms: What This Does to $50,000
Time to do the kitchen-table math, because "outperformance" is an abstraction until it has a dollar sign attached.
Take a 40-year-old with $50,000 to allocate. The research does not give us precise return figures, so we will not invent them — but we can compute the structural difference the two approaches produce, which is the part nobody publishes.
A broad dividend ETF in the SCHD/VYM family is an income vehicle. At a yield in the range those funds typically target, $50,000 generates roughly $1,500 to $2,000 a year in cash that lands in the account whether the market is up, down, or sideways. That is real money you can spend, reinvest, or use to buy more shares during a decline. It arrives on a schedule.
The AI-exposed side of the ledger produces almost none of that. The research is explicit: most AI-leading companies pay little or no dividend. So the same $50,000 in an AI-heavy allocation generates a cash yield close to zero. Every dollar of return has to come from the share price going up — and you only access it by selling.
Here is the non-obvious consequence. Those two portfolios do not just differ in expected return; they differ in when you are forced to make decisions. The dividend investor gets paid without acting. The AI investor must choose a moment to sell, and history is unkind to people choosing moments. That behavioral gap is a real cost, and it never appears in a returns chart.
Chart: Illustrative comparison of cash distributions, not total return. Dividend range reflects typical yields for broad dividend ETFs; AI-leading companies, per the research, pay little or no dividend. Total return is a separate question entirely.
In plain terms: one portfolio pays you rent, the other asks you to bet on the property's resale value. Both can be right. They are not the same product, and comparing their headline returns without adjusting for that is a category error.
Who Wins Under Which Condition
This is the comparison you will not find in a single source article, because it requires holding both sides at once.
If hyperscaler capex keeps accelerating — Microsoft, Alphabet, Amazon, and Meta continuing to expand data-center and GPU spending — the AI-exposed allocation wins, and it may not be close. That is precisely what drove 2024-2025 returns. The dividend ETF underperforms not because its companies did badly, but because it structurally does not own the accelerant.
If capex plateaus or gets scrutinized, the picture inverts fast. Capital spending is the most cuttable line item in a corporate budget. A hyperscaler that trims data-center plans sees its growth narrative repriced immediately, while a consumer-staples company paying a steady dividend sees essentially nothing change. The research frames this as higher upside with higher volatility versus steadier income with a downside cushion — and "cushion" is doing real work in that sentence.
If returns stay concentrated in a handful of names, both sides carry a risk that gets under-discussed. The AI investor holds concentration risk directly. But the dividend investor holds it too, in reverse: if a small group of mega-caps drives the market, a fund that excludes them will lag the S&P 500 for years while its holdings perform perfectly well on their own terms. That is a psychological drawdown, not a financial one, and it is the reason most people abandon income strategies at exactly the wrong time.
The framing that survives scrutiny is not "which one is better." It is the one the research lands on: these represent two different objectives — income and stability versus capital appreciation — rather than strictly competing bets. The scoreboard problem is that we keep grading them with the same ruler. This is a close cousin of the trap our sibling site examined in 7% Yield vs 3% Growth: Which Dividend Stock Wins?, where the headline yield told a very different story than the underlying math.
The AI Angle Nobody Prices In
There is a recursive irony here worth a sentence. AI is the central driver of the capex boom — hyperscaler spending on GPUs and data centers fuels the AI-linked stock gains — and those same companies pay little or no dividend, which creates the entire tension with income-focused ETFs. The technology generating the returns is the technology preventing the payouts.
For the ordinary investor, AI investing tools have made one part of this easier: portfolio-analysis features in most major brokerage platforms will now show your actual sector concentration and blended dividend yield in a few clicks. That is worth doing before this week ends. Many investors who believe they are diversified discover their index fund and their AI position are largely the same bet wearing two different names. The tool cannot tell you which objective to pick. It can tell you whether you accidentally picked both, or neither.
Which Fits Your Situation: Three Moves This Week
Write down, in one sentence, whether this money is for income you will spend or growth you will not touch for a decade. Nearly every dividend-versus-AI argument is really two people answering different questions. Your answer determines which scoreboard is even relevant to your investment portfolio.
Pull up your holdings and check how much of your total equity exposure sits in the same four hyperscalers. If you own a broad market index fund plus an AI-themed position, your concentration is likely higher than you assume, because cap-weighted indexes already carry heavy mega-cap tech weighting. This takes ten minutes and it is the highest-value personal finance task on this list.
Ask yourself now, while nothing is happening: if hyperscaler capex guidance were cut meaningfully, what would you do? Writing the answer down before the headline arrives is the entire discipline. Financial planning done during a drawdown is not planning — it is reacting.
Common Questions
Are dividend ETFs still worth it if AI stocks keep outperforming?
It depends on what you hired them for. Dividend ETFs like SCHD, VYM, NOBL, and DVY are built to deliver income and stability, not to beat a growth benchmark. If your goal is maximum capital appreciation, they were never the right tool. If you want cash flow that arrives regardless of price action, underperformance during a growth-led rally is the expected cost, not a malfunction.
Why do dividend ETFs not hold Nvidia or the big AI stocks?
Because of how the index screens work. Dividend funds require companies to pay meaningful — and often growing — dividends. Most AI-leading companies reinvest their cash into capital expenditure on data centers and GPUs instead of paying it out. The screen excludes them mechanically, before any judgment about the stock is made.
Can I hold both dividend ETFs and AI stocks in the same portfolio?
Nothing prevents it, and the research frames these as serving different objectives rather than being strictly competing bets. The thing to watch is unintentional overlap: a broad market index fund already carries substantial mega-cap tech weighting, so an added AI position may concentrate your exposure more than intended.
What happens to AI stocks if the capex boom slows down?
The research notes that AI stocks carry higher upside with higher volatility, while dividend strategies offer a downside cushion. Capital spending is discretionary and can be reduced quickly, which is why growth narratives tied to capex reprice faster than earnings tied to staples or healthcare demand. No one can forecast the timing, but the asymmetry in how the two react is structural rather than speculative.
Our analysis: the most likely outcome is not that one side of this trade is proven right, but that the comparison itself gets retired. On balance, the concentration of returns in a handful of AI-infrastructure names has made the S&P 500 itself a growth bet, which means the genuine decision facing most readers is no longer "dividends or AI" — it is whether they understand how much AI exposure they already own by default. That is a stock market today question worth answering before the next earnings cycle, not after.
Explore Our Network
Disclaimer: This article is editorial commentary for informational purposes only and does not constitute financial advice, an investment recommendation, or an offer to buy or sell any security. It is not based on independent product testing. Fund names are referenced as examples of strategy categories, not endorsements. Consult a qualified financial professional before making investment decisions. Research based on publicly available sources current as of September 5, 2026.
No comments:
Post a Comment