Neuberger’s Jonathan Bailey Urges Investors to Treat Artificial Intelligence as a Core Systemic Risk

Ask a chief investment officer about “system-level investing” and you will probably get a puzzled look. The term often floats through sustainable finance circles as a high-level concept, frequently detached from the immediate realities of portfolio management.

Ask the same CIO about heavy concentration in a handful of artificial intelligence stocks, thin governance at the technology companies involved, and the possibility that an AI bubble becomes a systemic problem rather than a bad year for returns, and you will get a nod of absolute recognition.

That gap is the central problem facing modern investment management. Sustainable investing professionals are talking a great deal about system-level investing and system-level stewardship. The terms are not always clearly defined, and the conversation often returns to the familiar themes the field has worked on for years: climate, biodiversity, and the workforce. Those risks are real, and the ongoing work to address them matters deeply to the long-term health of capital markets.

However, one system-level risk has, arguably, gone beyond the rest in terms of sheer urgency, and institutional CIOs already recognize it. It is moving much faster than the slow-building risks the debate usually centers on, and it is intensely concentrated in a small number of firms. Financial regulators, too, now treat its rapid adoption as a structural stability issue. If system-level thinking is to mean something practical in the current market environment, AI is precisely where it should be tested.

None of this is an argument against the advancement of artificial intelligence. The potential economic and societal gains, ranging from broad productivity growth to breakthrough drug discovery and the earlier diagnosis of complex diseases, are exceptionally large, and adoption will undoubtedly continue at a rapid pace. As Goldman Sachs chair and CEO David Solomon has put it, the question is not whether AI will advance, but whether it advances well.

For an institutional investor, navigating this landscape starts with absolute precision about what artificial intelligence actually represents. In the view of market veterans, it is simultaneously a powerful source of return and a profound source of risk. It is far better understood as a market-wide risk factor than as a narrow thematic trend—and crucially, it is one that investors can choose to hold in greater or lesser amounts depending on their mandates.

Every asset in a modern portfolio has some sensitivity to AI. Short-dated government bonds sit at the lowest end of this sensitivity spectrum. Conversely, data-centre developers, AI-exposed growth equities, and segments of the private equity market sit firmly at the high end. Credit and real assets sit somewhere in between, interacting with the technology in ways that are often difficult to discern from asset-class labels alone. Measured against the total investable universe, every portfolio carries an inherent AI beta, whether or not anyone actively or consciously chose to take that exposure.

AI: A Systemic Risk

The sheer size of the economic opportunity is part of the problem itself: it makes the associated risks remarkably easy for market participants to discount. In the narrower sense that matters for financial stability, a systemic risk is a malfunction that can cascade through the entire financial network, rather than simply a market risk that moves up and down with general economic cycles.

In an official letter addressed to G20 finance ministers and central bank governors, Financial Stability Board (FSB) chair Andrew Bailey singled out the potential for advanced AI models to conduct sophisticated cyberattacks as the most immediate concern for a highly interconnected financial network. He also warned that many regulatory jurisdictions still lack the necessary protocols to properly manage the development, release, and deployment of frontier models.

The Bank of England’s latest Financial Stability Report points in the exact same direction, highlighting a significant increase in operational and cyber risks, increasingly stretched market valuations, and AI-focused firms drawing on credit markets at an unprecedented pace.

Not everyone across the global financial landscape agrees with this assessment. At a closed-door investor conference, US Treasury Secretary Scott Bessent reportedly stated that he sees no systemic problem stemming from the AI boom and that it does not represent a repetition of the historic dot-com bubble. Those remarks, however, speak primarily to equity valuations and traditional market pricing.

The concern raised by the Financial Stability Board is fundamentally different: it focuses on how tightly artificial intelligence is woven into a highly interconnected financial system. A fundamentally sound economic boom can still carry severe systemic cyber, operational, and concentration risks, meaning the case for treating AI as a systemic risk does not rest on calling a market bubble.

For an asset owner, the exposure runs deeply through both sides of the balance sheet. On the asset side, it reaches well beyond public equity weightings into investment-grade and private credit, large-scale data-centre construction, commercial real estate, and essential infrastructure. On the liability side, one highly plausible channel of transmission is lower future contributions if job displacement from automation falls unevenly across particular demographic cohorts or specific industrial sectors.

Beyond these tangible channels sits a low-probability, high-severity tail risk involving the potential loss of control over powerful autonomous systems—a tail risk that no individual portfolio can truly diversify away.

What Asset Owners Can Do

No single asset owner can mitigate these complex risks in isolation, and neither can any individual AI developer. A technology laboratory that decides to slow down its development pace on its own simply cedes competitive ground to those that do not. The FSB chair made this exact point from the perspective of global regulators: these risks will not respect national borders. Effective mitigation will ultimately require a careful, system-level mix of industry self-regulation and robust public policy.

This is precisely where the case for system-level stewardship is strongest, and where it is also easiest for investors to overreach. Simply recognizing that AI is a systemic risk does not automatically tell an investor what concrete steps to take. The appropriate answer differs significantly depending on the specific role an institution plays in the market.

What Active Managers Can Do

An active asset manager is typically judged and paid for generating alpha over a defined time horizon set by the client—ranging from a single quarter to five years—while simultaneously looking through to the long-term underlying strength of a business model over a decade or more.

When applied to artificial intelligence, three distinct principles follow from this reality: active managers must adjust their AI beta deliberately, favor businesses with durable long-term economic moats over riskier counterparts, and understand how evolving self-regulation or government policy might impact each individual company.

Debt markets provide a clear example of this dynamic. Hyperscaler borrowing has grown so large that market analysts argue it directly affects broader government bond yields, and a manager who thoroughly understands who is buying long-dated debt is far better placed to judge those macroeconomic impacts.

The underlying investment opportunity is undeniably real. The core professional skill lies in successfully separating the businesses that will compound real value over time from those that will not, and in ensuring that investors know precisely what they are being paid for the risks they assume.

Because AI beta differs so widely across various portfolios and client mandates, it is not an active manager’s place to push a specific political or social policy goal on behalf of their clients. Instead, stewardship should aim to strengthen and de-risk the specific companies the manager owns, one by one, while deeply understanding the operational and financial risks those companies are actually running. That process produces practical insights that asset owners can directly utilize in their own strategic decisions.

System-level investing will ultimately be judged by whether it genuinely helps investors navigate the systemic risks standing directly in front of them. On the subject of artificial intelligence, that means asking much narrower and more disciplined questions than the broad public debate usually features. How much AI risk do we actually hold, and did we intentionally choose it? What are the companies we own actually risking in the process? And where can we credibly move the system, and where should we leave those interventions to others?

Investors who can answer those three fundamental questions will find themselves far better positioned than those who maintain a superficial view on everything while holding a concrete plan for nothing. Artificial intelligence will continue to advance, and its potential economic upside remains vast. The defining question for investors is whether they can help it advance well, and whether they are fairly compensated for the risk they take while it does so.

Jonathan Bailey is global head of stewardship and sustainable investing at Neuberger.