Artificial Intelligence has become one of the most powerful investment themes of the decade. From Last updated: June 2026
Ask ten analysts whether AI stocks are still a good investment in 2026, and you’ll get ten different answers — some warning of a bubble on the verge of bursting, others pointing to record earnings and calling it the start of a multi-decade buildout. Both camps are looking at the same data and reaching opposite conclusions, which tells you something important: this isn’t a question with an obvious answer, and anyone selling you certainty in either direction should be treated with caution.
This article lays out what both sides are actually arguing, what the money is doing right now, and which ETFs are capturing the flows if you decide the AI trade still has room to run.
Disclaimer: This article is for informational and educational purposes only and does not constitute financial advice. Nothing here should be read as a prediction of future returns. Always do your own research and consider speaking with a licensed financial advisor before investing, especially in concentrated, thematic sectors like this one.
The bull case: this time, the spending is real
The strongest argument against an AI bubble isn’t optimism — it’s cash flow. Unlike the dot-com era, when many companies were burning investor money with little revenue to show for it, today’s AI buildout is being funded largely out of corporate earnings rather than debt. According to Fidelity’s analysis, AI has pushed valuations for both the S&P 500 and tech stocks above historical averages, but those valuations remain below the extremes seen during the late-1990s dot-com bubble. Fidelity also notes that companies have largely funded their AI capital expenditures from earnings rather than debt, which reduces the risk of a debt-driven collapse if sentiment shifts.
The numbers behind that spending are enormous. Forecasts cited by The Hill suggest the major hyperscalers — Alphabet, Amazon, Meta, Microsoft, and Oracle — will spend around $755 billion on AI-related capital expenditures in 2026 alone. Bloomberg reports similarly aggressive figures, with the six largest U.S. hyperscalers projected to invest more than $500 billion combined in AI infrastructure this year, while TSMC has announced plans for as much as $56 billion in capital spending for 2026.
Money is following the thesis. AI-focused ETF inflows have surged dramatically — Bloomberg reports they jumped to $19 billion in the most recent year, up from just $4.2 billion the year before. Morningstar’s broader tracking of AI-themed ETFs found the category has pulled in more than $19 billion in net flows over the past 12 months, with $13.5 billion of that arriving in just the second half of last year.
The bear case: valuations are stretched, and history isn’t kind to «next big things»
On the other side of the debate, a growing number of voices argue the warning signs are already visible. Capital Economics has gone so far as to predict the AI-driven bubble will burst, arguing that rising interest rates and higher inflation will eventually weigh down equity valuations and trigger a correction.
Valuation metrics for individual AI names have drawn particular scrutiny. The Motley Fool points out that data-mining company Palantir entered 2026 trading at a price-to-sales ratio above 100, and that price-to-sales ratios above 30 have historically not proven sustainable for companies riding «next big thing» trends. More broadly, Fortune notes that the S&P 500’s cyclically-adjusted price-to-earnings ratio is at its highest level ever, outside of the early-2000s dot-com peak, and that just six companies — Nvidia, Microsoft, Alphabet, Amazon, Broadcom, and Meta — now account for almost 30% of the entire S&P 500, meaning a sharp AI selloff would hit the broader index hard, not just tech-specific funds.
Profitability remains the open question. One widely-cited estimate noted in GMO’s research suggests total AI-related revenue currently sits at less than $50 billion against well over a trillion dollars of cumulative investment — a gap that needs to close substantially to justify current valuations across the sector.
The honest middle ground
Perhaps the most useful framing comes from analysts who refuse to pick a side. Research firm Resonanz Capital put it directly: their answer to «is this a bubble» is «probably yes — but also yes, it can keep going, and possibly overheat, into 2026». Their reasoning is that bubbles aren’t defined by an absence of real fundamentals — they’re defined by price becoming increasingly dependent on narrative and positioning rather than fundamentals, with a shrinking set of outcomes that justify current valuations.
That’s a more useful mental model than a binary «bubble or not.» The spending is real. The earnings growth, so far, has mostly kept pace. But valuations leave less room for error than they did two or three years ago, and a disappointing earnings season from just one or two of the AI mega-caps could ripple through the entire sector given how concentrated the market has become.
So where is the money actually going? The ETFs leading the flows
Regardless of which side of the bubble debate you land on, fund flow data tells you where investor conviction currently sits. A few funds have stood out:
VanEck Semiconductor ETF (SMH)
The default choice for investors wanting direct exposure to AI’s hardware layer. ETF research from etf.com notes SMH has pulled in over $9 billion in net inflows over the past year, becoming the go-to vehicle for the chips-first AI thesis. As of mid-June 2026, SMH carries a 0.35% expense ratio, roughly $84.5 billion in total assets, and a year-to-date return near 83% — though that kind of run-up also means the fund is now more exposed to a sharp pullback if AI capex spending slows.
Global X AI & Technology ETF (AIQ)
With roughly $10.9 billion in AUM and a broader mandate spanning software, cloud, and semiconductors, AIQ has captured investors who want diversified AI exposure rather than a pure hardware bet. Its 0.68% expense ratio and roughly 85-90 holdings make it one of the most-discussed «core AI exposure» funds among financial publications in 2026.
Global X Robotics & AI ETF (BOTZ)
The longest-running fund in this category, BOTZ has gathered a cumulative net inflow north of $400 million over the past year alone, on top of its roughly $3.5 billion in existing assets — a sign that even after nearly a decade on the market, investor interest in the robotics angle of AI hasn’t faded.
While these established funds capture the bulk of the market flows, the landscape is constantly evolving. If you want to see the freshest options entering the market, you can check out our analysis on the new ETFs launched this month to see if any are worth adding to your radar.
The broader picture
Morningstar’s tracking of the AI-ETF category as a whole found these funds have collectively gathered a cumulative $28.6 billion in net inflows since the category’s first fund launched in 2016 — but performance has been mixed. Morningstar’s data shows the average AI-themed ETF has outperformed the S&P 500 since 2016, but has not been good enough to beat the Nasdaq 100 Index over that same stretch. That’s a meaningful detail: a basic, low-cost Nasdaq 100 fund has, on average, done just as well or better than dedicated AI-themed funds — without the added concentration risk or higher fees.
What this means if you’re deciding whether to invest
A few practical takeaways from all of the above, without telling you what to do with your own money:
- The spending behind AI is real and verifiable — hundreds of billions of dollars in capex from companies that are, for now, funding it out of profits rather than debt. That’s a meaningfully different setup than the dot-com era.
- Valuations leave less room for disappointment than they used to. When a handful of stocks make up close to a third of the S&P 500, «diversified» index investing is now more exposed to AI-specific risk than many investors realize — even without buying a dedicated AI ETF.
- Thematic AI ETFs haven’t necessarily beaten simpler, cheaper alternatives. Morningstar’s own data shows the average AI ETF has lagged the Nasdaq 100 since 2016. That doesn’t mean AI ETFs are a bad choice — it means the «AI» label alone doesn’t guarantee outperformance over a broad tech index you may already have access to at a lower cost.
- Nobody — not Fidelity, not Capital Economics, not any analyst quoted here — knows for certain whether this is a bubble. The honest answer is that it has bubble-like characteristics and real fundamentals at the same time, which is exactly why position sizing and diversification matter more than ever in this sector.
Frequently asked questions
Is it too late to invest in AI ETFs in 2026? No source can answer that with certainty, and you should be skeptical of anyone who claims otherwise. What’s measurable is that valuations have risen significantly, which raises both the potential reward and the potential downside compared to a few years ago.
Are AI ETFs riskier than a regular index fund? Generally, yes. Thematic, concentrated funds like AIQ, BOTZ, or SMH carry more company- and sector-specific risk than a broad market fund like a total stock market or S&P 500 ETF, which spreads exposure across hundreds or thousands of companies.
How much of my portfolio should be in AI-themed ETFs? There’s no universal answer, and this is a decision worth discussing with a licensed financial advisor given your own timeline and risk tolerance. Many investors who do hold thematic AI funds treat them as a smaller «satellite» position layered on top of a diversified core portfolio, rather than a primary holding.