The surge into AI-themed ETFs over the past two years marks a clear shift from “speculative fad” to “priority allocation” for many investors. In a market still digesting higher rates and uneven tech valuations, AI stands out as one of the few themes attracting sustained, global, data-backed interest. That resilience is notable given recurring 7-8% pullbacks in broader tech indices that have tested investor conviction.
From niche idea to global allocation
A decade ago, most investors expressing an AI view did it indirectly—through broad tech funds or single stock bets on mega-cap names. Today, AI is a defined asset class inside the thematic ETF universe, with its own dedicated products, benchmarks and flow patterns.
AI has evolved from indirect tech proxy to a defined, benchmarked thematic ETF asset class.
Global assets in artificial intelligence and big data funds have grown more than sevenfold in five years, reaching about $38.1 billion by the end of the first quarter of 2025. A striking detail in that number is the role of China: Morningstar data show that record inflows from Chinese investors, partly linked to domestic AI successes, have been a major driver of that asset growth. This is not just a U.S. story—it is a theme with genuine international breadth.
In Europe, the thematic ETF ecosystem has matured rapidly. Invesco reports over 200 European thematic ETFs with more than $50 billion in assets, with AI highlighted as one of the highest‑growth segments within that universe. That breadth matters: it suggests AI exposure is increasingly being built through diversified, rules‑based products rather than ad hoc stock picking alone.
Flow dynamics: AI pulls ahead of other themes
The clearest signal of investor conviction is in the flow data. Bloomberg analysis shows AI-focused ETF inflows jumped to $19 billion last year, up sharply from $4.2 billion in 2024. That is an almost fivefold increase in a single year—remarkable in an environment where many other thematic strategies are still seeing cautious or negative flows.
Within the broader disruptive technology segment, which includes AI, investor attention has been even more concentrated. Market commentary from ETF specialists notes that AI-related disruptive tech attracted nearly $20 billion in inflows year-to-date, with roughly $15 billion of that going to ETFs that actually have “AI” in their names. That detail is important: it suggests that investors are deliberately seeking targeted AI exposure rather than hiding the theme inside general “innovation” or “future tech” baskets.
At the same time, the AI trade is not limited to pure-play thematic products. In the U.S., robotics and AI ETFs collectively hold around $18 billion in assets and had already drawn about $4.5 billion in net inflows year to date through mid‑July 2025. Yet flows data show that many investors still default to broad vehicles like QQQ for AI exposure, using dedicated AI and robotics ETFs as satellites rather than core holdings. That hybrid approach helps explain why AI themes can have strong inflows even when investor risk appetite more broadly is mixed.
It is also worth keeping this AI inflow surge in historical context. JPMorgan’s Global ETF Handbook points out that U.S.-listed AI-related thematic ETFs added around $3 billion of new assets over the most recent year in its study. That is far below the roughly $16 billion that poured into ARKK alone in the twelve months through March 2021, during the peak of the earlier innovation boom. In other words, AI flows are strong and accelerating, but they have not yet reached the euphoric extremes seen in the last cycle—a nuance that matters for any serious risk assessment.
Thematic ETFs: from boom, to bust, to AI-led recovery
The AI story sits inside a broader reset in thematic investing. Invesco’s work on thematic flows highlights that many themes endured a difficult two‑year stretch prior to 2024, with rising interest rates and style rotations away from high‑growth narratives weighing heavily on performance and demand. Investors who piled into clean energy, genomics or “disruption” more generally in 2020–2021 often spent 2022–2023 nursing double‑digit drawdowns.
Flows have started to turn positive again, and AI has been central to that recovery. Invesco identifies a “Digital Future” category—combining AI, robotics and related technologies—as the top‑flowing theme during the recent rebound in thematic ETFs. State Street’s thematic dashboard echoes this pattern: robotics and AI led the pack in Q3 2025, drawing about $7.4 billion of inflows, the largest share among major themes.
More recent data reinforce the momentum. Thematic ETFs globally saw their strongest early-year intake since 2021, with roughly $2.4 billion of net inflows in the first two months alone; nearly half of that—around $1.1 billion—came from robotics and AI ETFs. Those numbers point to something structurally important: even as investors remain selective after the last boom‑and‑bust cycle, they are clearly willing to re‑engage where they see long‑run technological platforms rather than short‑term stories. AI fits squarely into that category.
Performance and positioning: not just narrative, but numbers
Flows tend to follow performance, and AI ETFs have delivered enough returns to keep the narrative grounded in hard data rather than marketing alone.
Across a broad universe of AI and big data ETFs, aggregate 1‑year performance has been in the neighborhood of +40%, significantly ahead of many traditional equity benchmarks and much of the wider thematic complex. Drill down further and the dispersion is large: analyses of the “best‑performing” AI ETFs in 2025 show several strategies posting year‑to‑date gains between roughly 30% and 50%, with some generative‑AI‑focused products even higher.
Active strategies have participated as well. One example is an AI and robotics‑focused active ETF from ARK that gained about 31.7% in 2025, reflecting concentrated exposure to names at the intersection of automation, defense and AI software. Meanwhile, leveraged and inverse products linked to AI-related stocks have multiplied rapidly: by August 2025, around 112 new leveraged and inverse ETPs tied to AI names had been launched, bringing the U.S. total close to 196 and pushing AI-linked products to more than half of assets in the leveraged ETP universe. That growth underlines both how popular the trade has become—and how much risk some investors are willing to take to express it.
The performance pattern is not uniform, though. While leading AI ETFs and related names have outpaced broad tech, many general thematic funds have lagged benchmarks, especially those exposed to themes that proved more cyclical than structural. This divergence reinforces a key point for anyone allocating to AI: the label “thematic” does not guarantee future relevance. AI—and particularly the infrastructure and software layers enabling it—appears more deeply embedded in corporate investment plans than many themes that rode the 2020–2021 wave.
Why investors are treating AI as a “durable pillar”
From an industry perspective, the inflow and performance data tell a coherent story. JPMorgan’s ETF research frames AI-related ETFs as an area of sharp growth despite a tough environment for markets more broadly, where valuations and risk appetite have repeatedly been tested. Combined with Morningstar’s evidence of expanding global assets and strong Chinese participation, the picture is one of AI moving from “interesting niche” to “core long‑term bet” for many institutional and retail portfolios.
Several structural factors explain this:
- Enterprise adoption is now visible, not hypothetical. Generative AI and large language models, which hit public awareness in 2023, are translating into tangible spending—on cloud infrastructure, data centers, specialized chips and software subscriptions. Large asset managers expect cash flows for leading AI companies to flip from heavy investment outflows toward meaningful revenue inflows over the next few years as AI applications scale across sectors.
- The theme spans hardware, software and services. AI ETFs typically hold mixes of semiconductor designers, cloud platforms, enterprise software vendors and robotics manufacturers. This breadth provides multiple revenue “legs” for the theme: even if one sub‑segment slows, others—such as data center infrastructure or industrial automation—can continue to grow.
- Global policy and competition are reinforcing the trend. From strategic AI funding in China to industrial policy in the U.S. and Europe, governments see AI as a competitive necessity. That raises the probability that AI investment will persist across cycles, even if valuations and specific winners change.
Risks and misconceptions: what a sober view looks like
Experience with previous thematic booms makes it clear that strong data today does not eliminate risk. Several points deserve emphasis for readers trying to build a responsible AI allocation:
- Valuation and concentration risk. Many AI-themed ETFs are heavily exposed to a small group of high‑multiple stocks, particularly in semiconductors and cloud platforms. A handful of names can dominate index-level returns. If market sentiment toward those leaders turns, ETF holders may experience significant drawdowns even if the underlying technology continues to advance.
- Leverage amplifies both upside and downside. The rapid growth of leveraged and inverse AI-related ETPs—now representing over half of assets in the leveraged product universe—should be treated as a warning sign as much as a data point. These vehicles are designed for short-term trading, not long-term compounding, and they can magnify volatility and path dependency in ways that are often misunderstood by retail investors.
- Theme purity versus diversification. Flow data show a clear investor preference for ETFs explicitly labeled “AI”, but the same data also highlight the ongoing use of broad tech funds like QQQ as the core AI proxy. Pure‑play AI exposure can be powerful, but it is also narrower and more vulnerable to policy shocks, regulatory changes or technical setbacks in specific AI subfields. Blending targeted AI funds with diversified tech or global equity allocations remains a more robust approach for most investors.
- Geopolitical and regulatory uncertainty. Morningstar’s finding that Chinese inflows account for a substantial share of global AI and big data assets underlines both opportunity and risk. Geopolitical tensions, export controls on advanced chips, and evolving rules around data and AI safety could all affect the profitability and investability of certain parts of the AI value chain.
Takeaways: how to think about AI-themed ETFs now
Putting all the evidence together, a few conclusions stand out.
First, AI-themed ETFs have moved into a clear growth phase, with inflows accelerating from $4.2 billion in 2024 to $19 billion in the following year, and disruptive AI strategies attracting nearly $20 billion year-to-date in fresh capital. This is happening in a market still wary of high‑beta themes, which makes the trend more noteworthy than the raw numbers alone suggest.
Second, the theme is genuinely global. From the $38.1 billion in AI and big data assets worldwide, driven in part by Chinese investors, to Europe’s 200‑plus thematic ETFs where AI is a standout growth segment, the data show a geographically diversified investor base. That breadth reduces the likelihood that AI is simply another localized bubble.
Third, performance has, so far, justified renewed attention. Across categories, AI and big data funds have delivered strong one‑year returns, often ahead of major indices and other themes, while leading AI ETFs have posted substantial double‑digit gains. This outperformance is a key reason flows have persisted even as macro conditions stayed challenging.
Finally, the lessons from the last thematic cycle argue for disciplined enthusiasm. AI looks more like a long‑run technology platform than a passing story, but the vehicles used to access it—especially concentrated or leveraged ETFs—can be unforgiving when sentiment turns. Treating AI as a “durable pillar” of a portfolio does not mean abandoning diversification, valuation discipline or risk management.
For investors, businesses and policymakers, the message is similar: AI is no longer just a bet on the future—it is an active, evolving capital market theme shaping how money is allocated today. The ETF data simply make that reality impossible to ignore.
Conclusion
JPMorgan’s latest ETF guide is capturing a paradox that says a lot about where the AI story is right now: investors are pouring money into AI-themed exchange-traded funds even as the broader AI equity complex has gone through a choppy, sometimes painful quarter. This divergence between performance and flows is a strong signal that many investors now see AI as a long‑term structural theme, not a speculative trade to abandon after a few volatile months.
Why this matters now
Over the past decade, AI has moved from being a niche “future tech” allocation to one of the core drivers of global markets, corporate earnings and portfolio construction. JPMorgan’s new “Guide to ETFs” and its midyear ETF report show that AI-focused strategies have become one of the top investment themes by assets under management, even as the sector has endured sharp drawdowns since late 2025. That mix of strong inflows and rough performance is exactly what you see when a theme shifts from hype cycle to long‑duration capital cycle: capital doesn’t disappear when prices wobble; it rotates across segments of the ecosystem.
At the same time, the guide reinforces a broader industry shift: money is steadily migrating out of traditional mutual funds and into ETFs, with AI and other secular growth themes increasingly being expressed through ETF wrappers rather than legacy vehicles. For asset managers, advisors and allocators, that isn’t just a product trend; it is a change in the operating system of portfolios.
The numbers behind JPMorgan’s ETF guide
JPMorgan’s ETF work highlights several concrete data points that help explain why AI thematics are staying in focus:
1. AI is now a top‑five ETF theme by AUM
The firm’s guide shows AI-themed ETFs ranking among the top five thematic categories by assets under management, despite elevated volatility in the group in the second quarter. Thematic ETF assets overall grew roughly 33% in the first half of 2026 to around $430 billion, with inflows tracking the second‑highest annual total on record. AI is explicitly identified as a key driver of this resurgence in thematics because it cuts across sectors—from semiconductors and robotics to data infrastructure and software.
2. ETF flows are breaking records, with AI at the center
U.S. ETF flows surpassed $1 trillion in the first half of 2026 at the fastest pace on record, putting full‑year flows on track to be 35–40% higher than in 2025. Within that, about 33% of all ETF flows went into active ETFs, pushing active equity and fixed income ETF flows to their highest levels on record. AI-specific and AI-adjacent strategies sit squarely in that active bucket, as investors seek managers who can navigate fast‑moving technology and valuation cycles rather than simply owning static indexes.
3. Mutual funds are losing ground
JPMorgan’s analysis shows mutual fund inflows meaningfully tapering off, with net outflows across many categories in recent years, while ETFs capture growing shares of new capital. Other industry work echoes this, noting that active ETFs have grown roughly 50% organically over the past five years while active mutual fund assets have declined about 4%, with expectations that global active ETF assets could triple to roughly $4.2 trillion by 2030. That structural shift is critical: AI is being plugged into portfolios through the vehicles that are gaining share, not those slowly shrinking.
From mutual funds to ETFs: a structural migration
The ETF guide and related midyear work frame the growth of AI ETFs as part of a broader migration from “legacy vehicles” into more flexible wrappers. After early momentum from mutual fund-to-ETF conversions, JPMorgan notes that separately managed account (SMA) conversions have accelerated, positioning ETFs as a preferred “operating system” for implementation and advisor workflows. BlackRock’s product trends analysis reaches similar conclusions, highlighting how new share‑class structures and rule changes may further merge mutual funds and ETFs, but still leave ETFs as the more scalable, convenient wrapper for most investors.
For AI specifically, that matters for two reasons:
- It widens access. ETF wrappers allow retail and smaller institutional investors to participate in specialized AI exposure—whether generative models, robotics, data centers or semiconductor supply chains—without needing complex structures or high minimums.
- It shortens the feedback loop. As AI-related fundamentals and sentiment shift, ETFs make it easier to rebalance or rotate across the stack (hardware, infrastructure, software, applications) in a single trade.
In other words, the rise in AI-themed ETFs is not just about enthusiasm for AI; it is about the financial plumbing evolving to deliver AI exposure more efficiently.
AI as a long‑term structural theme, not a fad
Multiple strands of research now support the idea that investors treat AI as a multi‑year “supercycle” rather than a short‑term boom. Nuveen characterizes AI as a multi‑year capital cycle driven by real constraints—compute, power, cooling, and infrastructure—rather than a purely speculative technology bubble. JPMorgan’s own AI work estimates plausible productivity gains of roughly 1.4–2.7% from generative AI and other AI technologies over the next few years, on top of an already‑expected 1.5% annual productivity trend from the broader economy. That kind of incremental productivity shock is precisely what long‑term investors seek when they allocate to structural themes.
Institutional sentiment data back this up. Natixis finds that about 65% of institutions expect AI to “supercharge” growth again, even as many express concern about valuations and bubble risk. The same survey reports that more than half of institutions are focused on AI‑adjacent opportunities—areas like data infrastructure, power and enabling technologies—rather than only pure AI software names. Other institutional outlooks describe AI as “close to a universal macro and equity driver,” with teams treating it as a long‑term theme even when they remain cautious about near‑term price action.
JPMorgan’s market outlooks mirror these views. Its Q2 2026 review notes that earnings growth remains robust and broad‑based, led by technology sectors where strength is tied to AI capital expenditure and margin discipline, contributing to six consecutive quarters of double‑digit EPS growth in the U.S. AI and commodities are described as the two most important themes supporting a constructive equity outlook despite elevated volatility.
Volatility, concentration risk and bubble fears
The headline in JPMorgan’s ETF guide—sharp growth in AI-themed ETF assets despite a difficult quarter—is not an accident. The AI sector has been volatile. Since November 2025, parts of the software-heavy AI segment are down more than 30%, according to JPMorgan’s JTEK strategy commentary, prompting the managers to hold a roughly 20‑percentage‑point underweight to software. That decline has unfolded alongside uneven demand shocks across the “AI stack,” where bottlenecks in compute, memory and networking are amplified by expectations that a handful of large cloud and hyperscale players will spend more than $600 billion on capital expenditure.
Several risks stand out:
1. Concentration in mega‑cap tech
Recent market gains have been heavily powered by dominant U.S. technology companies whose AI spending and productivity investments underpin stronger profit trajectories, even at rich valuations. This raises concentration risk: a narrow group of names drives index‑level returns, and AI-themed ETFs can end up heavily exposed to those same leaders, especially in market‑cap‑weighted designs.
2. Valuation and bubble concerns
Institutional investors explicitly worry about a bubble in parts of the AI complex, particularly segments where revenue models are unproven and barriers to entry are lower. Elevated multiples, combined with rapid retail inflows into AI-branded products, can create an uncomfortable mix of structural optimism and speculative behavior.
3. Cyclical and supply-side risks
Hardware and infrastructure beneficiaries—semiconductors, memory, networking, data centers, power systems—are enjoying extraordinary top‑line growth, with some companies expected to grow revenues by 80–160% in 2026. Yet these are cyclical industries, sensitive to inventory corrections, policy shocks and changes in hyperscaler capex budgets.
The ETF guide does not dismiss these risks. Instead, it implicitly argues that the right way to hold AI as a structural theme is to broaden exposure across the ecosystem and use active selection and diversification to manage concentration and cyclicality.
Implications for technology, businesses and portfolios
The surge in AI-themed ETF assets, despite a tough quarter, is more than a market curiosity. It has real consequences for how technology is funded, how companies are evaluated and how portfolios are built.
1. Capital allocation across the AI value chain
As AI-themed ETF AUM climbs, capital is distributed beyond front‑page software names into enabling layers: semiconductors, data centers, power grids, cooling systems, specialized equipment and robotics. This supports long‑dated investment in physical infrastructure that AI systems need to function at scale, aligning with the view that digital and physical investments offer distinct but complementary risk‑return profiles.
2. Pressure on companies to demonstrate real productivity gains
With asset managers quantifying plausible productivity gains from AI in the range of 1.4–2.7% above baseline trends, there is growing pressure on companies to justify AI spending with measurable efficiency and margin improvements. Earnings narratives increasingly separate firms that turn AI into better workflows, products and customer outcomes from those that merely rebrand existing tools as “AI.”
3. Evolution of product development in asset management
The ETF guide and related industry work show AI playing two roles in product development: as an investment theme and as a tool inside the investment process. Product teams are building AI-themed ETFs, but they are also using AI to enhance research, risk management, portfolio construction and client reporting. BlackRock highlights “new ways of using AI in investment processes” as one of the key 2026 trends shaping investment products, alongside democratization of private markets and more convenient wrappers.
4. Democratization and responsibility
The ETF wrapper makes AI exposure more accessible to a wide range of investors, but it also raises responsibility questions. Investors who buy AI ETFs may be indirectly funding large-scale data usage, compute intensity and energy consumption that have social and environmental implications. Institutions increasingly factor sustainability, data governance and regulatory risk into their AI allocations, which is likely to influence index construction and active selection over time.
Opportunities and risks in context
When viewed against earlier technology cycles—dot‑com, mobile, cloud—the current AI supercycle looks both familiar and different. The familiar pattern is rapid capital formation, hype, valuation spikes and subsequent drawdowns. The difference lies in the breadth of the ecosystem and the speed at which AI is being embedded into real business processes, from credit underwriting and logistics to healthcare diagnostics and industrial automation.
Opportunities:
- Sustained earnings growth: Forecasts still point to double‑digit earnings expansion through 2026, with technology—largely AI-driven—acting as a key engine of profits and innovation.
- Diversified exposure: Investors can now access AI not only through U.S. mega‑cap tech, but also through international hardware and infrastructure names, credit, private markets and real assets that finance AI deployment at scale.
- Better portfolio tools: Active and thematic ETFs offer more targeted ways to express views on parts of the AI stack, helping investors manage concentration and factor exposures.
Risks:
- Over‑reliance on a narrow cohort of leaders, making portfolios vulnerable to idiosyncratic shocks or regulatory actions.
- Potential mismatch between long‑term AI adoption curves and short‑term investor expectations, which can lead to periods of sharp repricing.
- Structural issues such as energy intensity, data privacy and geopolitical tension around AI supply chains, which can influence valuations and access.
Key takeaways and what to watch next
JPMorgan’s ETF guide, combined with midyear flows data and institutional surveys, paints a clear picture: AI-themed ETFs are attracting strong inflows and climbing the ranks by assets under management even as the underlying sector navigates volatility and drawdowns. This contrast suggests investors increasingly treat AI exposure as a long‑term structural bet—anchored in expected productivity gains and earnings power—rather than a short‑term momentum trade.
For investors and practitioners, several forward‑looking points are worth tracking:
1. How ETF construction evolves
Expect more AI ETFs that go beyond simple “tech baskets” to include infrastructure, power, data, and specialized industrials, as managers respond to both concentration risk and demand for diversified exposure.
2. The balance between active and passive AI exposure
With a growing share of ETF flows heading into active strategies, AI themes may increasingly be implemented via active ETFs rather than purely passive index trackers, allowing for more nuanced risk management and fundamental selection.
3. The real-world impact of AI productivity
As the projected 1.4–2.7% productivity uplift from AI is tested in practice, markets will likely reward companies and strategies that convert AI investment into tangible margins, while penalizing those that fail to deliver.
4. Regulatory and societal responses
Policymakers and regulators are moving closer to AI-related rules on data, safety and competition. Those decisions will shape which business models are scalable and how capital flows into different parts of the AI ecosystem.
The core message behind the sharp growth in AI-themed ETFs, even in a difficult quarter, is that AI is no longer seen as a passing fad. It is becoming a foundational layer in both the real economy and the architecture of modern portfolios—and investors are structuring their capital to reflect that, even when the path is volatile.
