Exuberance and exposure
Advertisement feature
Teal and blue background patterned with binary code

Addressing AI risks: Beyond active versus passive

A graphic of a half-biological, half-digital human brain positioned in front of planet Earth

For investors worried about an AI bubble, comparisons with the dotcom crash at the turn of the Millenium are frequently cited. In March 2000, with the fervour around nascent internet companies at its peak, the top seven stocks were about 22% of the S&P 500. Now the largest seven stocks – known as the Magnificent Seven – comprise roughly a third of the index.1 Technology, as measured by ICB industry allocation, is about 40% of the index.2

Market Concentration: Then vs Now

Each dot represents 1% of the S&P 500 index

2000

Dotcom Peak

Top 7 stocks

22%

2026

AI Era

Top 7 stocks

33%

Source: S&P 500's Weight in May 7 Stocks Passes 30%, Forbes, May 2026.

Although this heightened concentration is certainly giving investors cause for thought, one important difference now is that the earnings of the AI hyperscalers are stronger than their internet goldrush predecessors3. Also, Amazon’s Jeff Bezos, among others, has suggested that while this may be a bubble, it could be a “good” bubble, as the infrastructure developed will provide huge benefits.4

That may be true, but it does not mean that investors cannot also make huge losses, from which it may take a long time to recover. Again, the dotcom bubble serves as a warning. Further back there is the example of the nineteenth century’s ‘railroad mania’ in the US, which saw money poured into expansive rail networks. Land grants, easy credit and westward expansion catalysed speculation. Deep crashes and huge waves of bankruptcies followed. Yet no one would doubt the viability of either technology: the internet is still with us, as are railways. The irrationally exuberant can lose their money in bubbles arising from important technological and societal shifts, just as they can by investing in tulip bulbs.  

A wide panoramic banner featuring a large satellite dish silhouette on the right, set against a teal background overlaid with financial stock charts, candlesticks, and market data.

The ubiquity of AI risk

Cautious investors may therefore want to limit their exposure to AI or at least understand the extent of their exposure to mitigate risk through other means. But the tentacles of AI are more pervasive and extend further than first appears. “Investors are beginning to wake up to the idea that it’s not just technology companies being impacted by AI, but broader services,” says Sebastian Schiele, Global Head of Xtrackers Mandates & Solutions Sales. “Cutting a long story very short, AI seems prevalent across a lot of portfolios.

Investors must also look at other sectors with direct exposure to the AI hyperscalers. Bank exposure via loans to the big five hyperscalers – Amazon, Alphabet, Meta, Microsoft and Oracle – is relatively small, but their bond issuance has markedly increased. This year, analysts raised issuance estimates to circa $750bn - a 67% year-on-year increase.5 A significant proportion of capital raised is being absorbed by data centre demand, which bleeds into the real estate sector. The ripple effect is likely larger. 

Quote mark | Xtrackers

Investors are beginning to wake up 
to the idea that it’s not just technology companies being impacted by AI, but broader services.”

Sebastian Schiele, global head of Xtrackers Mandates & Solutions Sales

Additionally, investors should consider exposure to companies from all sectors that are adopting AI across their operations. The number is growing rapidly, with 72% of S&P 500 companies disclosing at least one material AI risk in 2025, up from just 12% in 2023, according to the Harvard Law School Forum on Corporate Governance.6 Assessing the impact of AI on companies’ valuations is not easy. However, share prices undoubtedly react to the expected efficiencies provided by the technology: for instance, in credit modelling or fraud detection in financials, or diagnostics and patient interfaces in healthcare.

And, conversely, investors also need insight into those companies and industries most at risk from the rollout of AI. For example, was the software sell-off of early 2026 – when fears that AI could make the SaaS model obsolete saw the enterprise software sector suffer its worst hit since the dotcom bubble burst7 - a harbinger of things to come, or simply market froth? Markets are not good at fully incorporating technological obsolescence into asset prices, and this under-reaction favours the obsolete firms, according to research by Yale University’s Song Ma.8 This mispricing may impact the cost of capital, especially given that innovative firms are often more equity dependent.

Using this approach to assess AI exposure when selecting benchmarks – irrespective of whether using active or passive strategies – means drilling down into the exposure of their constituents. “You cannot sensibly manage AI risk if you don’t first make the exposure explicit,” says Mr. Schiele. This means turning “implicit, hidden AI bets into something that’s quantified and visible, so investors can decide whether to keep or restructure that exposure.”

Chart showing 72% of S&P 500 companies have at least one material AI risk, up from 12% in 2023

Source: AI Risk Disclosures in the S&P500, Harvard Law School Forum on Corporate Governance, May 2026.

72%

of S&P 500 companies have at least one material AI risk, up from 

12%

in 2023

X-ray specs

This has created persistent demand for alternative weighting strategies, as investors seek customised solutions. Mr. Schiele stresses that the client is very much in the driving seat, confirming the importance of offering a broader set of tools to investors, which allow them to express specific views on the market. “In this way, institutional investors complement their core market-cap exposures with more targeted building blocks. Often a client will start with a custom solution such as capping or excluding the biggest names. If we see broader interest, this can be spun out as an ETF so a wider set of investors can access the same strategy.” 

Whether for an alternative weighting strategy, standard index or broader portfolio, the first step is mapping where AI-related risk sits across holdings. Then this transparency can inform how much unintended or hidden AI concentration there is. 

Quote mark | Xtrackers

One thing index providers could do [to address AI risk] is to build indices based on the degree of AI usage of the underlying companies.”

Ravi Kashyap, Finance Faculty at the University of Waikato Institute at Hangzhou City University

Ravi Kashyap, Finance Faculty at the University of Waikato Institute at Hangzhou City University has developed the Concentration Risk Indicator as an ‘x-ray’ tool to measure such exposures more effectively. It detects elevated systemic and portfolio level vulnerabilities that arise when capital accumulates in a narrow subset of securities – as is the case with today’s AI hyperscalers – highlighting how such concentration amplifies volatility, distorts risk return profiles and weakens diversification benefits.9 “When comparing two investments, the lower the concentration risk, the better the investment from a diversification point of view,” says Mr Kashyap.

Portfolios can then be reweighted to mitigate those risks. Mr Kashyap believes that passive strategies are well placed to service these needs: “It’s very easy to build an index these days,” he points out. “One thing index providers could do [to address AI risk] is to build indices based on the degree of AI usage of the underlying companies.”

A wide panoramic aerial photograph of the London skyline washed in a teal color filter featuring recognizable skyscrapers of the City of London financial district

Where active meets passive

Those sceptical of markets’ ability to correctly price the AI transformation may be nervous about investing via a broad cap-weighted index fund. However, the costs associated with active management, together with the fact that outperformance of the market is far from guaranteed, are also considerations. 

Straying from the index can be perceived as risky for those making the decision, adds Mr Schiele. But he argues that counterposing active to passive investment strategies can be missing the point: “In deciding to allocate to one benchmark or another, you are making an active decision. Even going for pure market-cap weighted is itself an active decision.” One way to mitigate unwanted concentrations can be the combination of passive core holdings with active satellites, he suggests.

“But transparency and explicit measurement of AI exposure are the starting point for any serious AI risk management in portfolios,” Mr Schiele points out, “before you even get to tools such as exclusion, caps or alternative benchmarks.”

Written in association with Xtrackers by DWS

Important note: This editorial contribution has been prepared in partnership with a third party. Despite careful selection of sources, no liability is assumed for the correctness of the contents. All statements and performance data do not constitute financial analysis. They are for information purposes only and in no way encourage the purchase or sale of financial instruments or securities. Historical performance and any awards for it are not guarantees of current or future performance.

DWS and Economist Enterprise are not affiliated.

All investments involve risk, including potential loss of principal. The brand Xtrackers represents all systematic investment solutions. Xtrackers ETFs (“ETFs”) are managed by DBX Advisors LLC (the “Adviser”), and distributed by ALPS Distributors, Inc (“ALPS”). The Adviser is a subsidiary of DWS Group GmbH & Co, KGaA, and is not affiliated with ALPS.

The brand DWS represents DWS Group GmbH & Co, KGaA and any of its subsidiaries such as DWS Distributors, Inc., which offers investment products, or DWS Investment Management Americas, Inc. and RREEF America L.L.C., which offer advisory services.

© 2026 DWS Group GmbH & Co. KGaA. All rights reserved. 110029-1 (5/26)

Economist Enterprise logo

Exuberance and exposure: Institutional investors and the AI boom

The report examines how institutional investors are shaping—and potentially amplifying—the AI investment boom and balancing long-term conviction with short-term fragility as they prepare for a likely market correction.

Learn more about the report