Every time information technology advances, we have the same discussion: Is our new tech somehow going to crash financial markets?
In this regard, artificial intelligence is not so different from everything between it and the abacus… but the discussion around the potential financial risks from AI investment and adoption are so multi-faceted and fascinating, they bubble up frequently in our weekly research—which is why we’ve chosen to look back at some recent news and comments about AI’s impact on financial markets this week.
For financial professionals, the discussion about AI has gone beyond whether it has staying power—it’s been a while since we’ve heard AI called a fad by any serious commentator—or asking which companies will win the AI race. Increasingly, the larger question is whether AI itself has become a source of systemic financial risk.
Is AI becoming a Financial Bubble?
Definitely maybe. A technology can be in an overvalued bubble and be revolutionary and capable of offering enduring value at the same time. The argument supporting the bubble thesis is straightforward. Capital expenditures on AI are reaching levels rarely seen outside wartime mobilizations or national infrastructure programs. Large technology companies are investing hundreds of billions of dollars into GPUs, networking equipment, semiconductor manufacturing, electricity generation and data centers before corresponding revenues have fully materialized. Morgan Stanley estimates that hyperscalers could collectively spend hundreds of billions of dollars annually on AI infrastructure, while broader industry estimates envision trillions of dollars flowing into AI infrastructure over the remainder of the decade.
Equity valuations have reflected those expectations. Semiconductor manufacturers, memory producers, networking firms and AI infrastructure suppliers have substantially outperformed much of the broader market, while private-market valuations for leading AI developers continue climbing despite relatively limited operating histories.That has led numerous commentators to compare today’s environment with the late-1990s internet boom.
The comparison has become sufficiently mainstream that the Bank for International Settlements devoted significant attention to AI in its latest Annual Economic Report. Rather than declaring AI a bubble outright, the BIS warned that enthusiasm surrounding artificial intelligence has become an emerging financial stability concern. The institution noted that investment booms built around transformational technologies often create productive infrastructure that benefits society over the long run while simultaneously producing painful investment corrections when expectations temporarily outrun economic returns.
Reuters documented growing concerns among institutional investors regarding elevated valuations and the enormous borrowing required to finance AI infrastructure. Analysts at major financial institutions have observed widening performance gaps between semiconductor manufacturers benefiting immediately from infrastructure spending and many large software companies expected eventually to generate AI profits. That divergence resembles patterns observed during previous technology investment cycles, when equipment suppliers often outperformed downstream users during the earliest phases of infrastructure construction.
History suggests bubbles are rarely built upon worthless technologies. Railroads transformed transportation despite numerous railroad bankruptcies. The internet reshaped the global economy despite the collapse of hundreds of dot-com companies. Electricity fundamentally altered industrial civilization even though many early investors lost fortunes financing electrical infrastructure. Artificial intelligence may ultimately follow the same historical pattern.
The more relevant question may not be whether AI constitutes a bubble, but whether current market prices correctly distinguish between enduring infrastructure investments and speculative enthusiasm. Financial markets can simultaneously overvalue individual companies while correctly recognizing the arrival of an economy-changing technology.
Market Infrastructure and the New Sources of Systemic Risk
Financial markets are not simply investing in AI. They are increasingly being operated by AI. Banks, exchanges, hedge funds, asset managers, market makers and financial infrastructure providers are embedding machine learning and increasingly agentic AI into nearly every stage of the investment process. Portfolio construction, securities research, compliance surveillance, trade execution, client communications, fraud detection, liquidity forecasting and operational resilience are all becoming progressively more automated. The transformation promises lower costs, greater efficiency and better decision-making. It also introduces a new category of systemic risk, as many financial institutions rely upon similar AI systems that make similar decisions simultaneously. Remember those concerns about robo-advisors and index funds and algorithmic trading? Here we go again.
Yet, these concerns are moving from academic theory to regulatory agenda. During June, the Financial Stability Board (FSB) released a consultation report on sound practices for the responsible adoption of artificial intelligence in financial services, while separately highlighting AI as a potential source of new vulnerabilities affecting global financial stability. The organization stopped well short of warning against AI adoption. Instead, it emphasized that AI introduces new operational dependencies, governance challenges and concentration risks that financial firms must understand before AI becomes deeply embedded in core market infrastructure.
History suggests that seemingly obscure infrastructure issues often become catalysts for major market disruptions. The 1987 stock market crash, the 2010 Flash Crash, the collapse of Long-Term Capital Management, and the 2008 Global Financial Crisis were all amplified by structural characteristics of financial markets rather than simply deteriorating economic fundamentals. AI introduces another layer of interconnected infrastructure whose behavior during periods of stress remains largely untested.
Automation itself is hardly new. Algorithmic trading has existed for decades. High-frequency trading firms already account for a significant percentage of equity market volume. Banks have long employed quantitative models to price securities and manage risk. Artificial intelligence differs because it introduces systems capable of adapting, reasoning and generating novel responses rather than simply executing predefined instructions.
Traditional algorithms generally fail in predictable ways because their decision rules are explicitly programmed. AI models, particularly large language models combined with reasoning systems, can produce unexpected behavior under unusual market conditions. When thousands of institutions deploy similar foundation models trained on similar information, an unexpected market event could produce correlated decisions that amplify volatility rather than dampen it.
Financial systems become more fragile when many participants rely upon similar analytical frameworks. If competing firms all purchase AI services from the same handful of providers, use similar datasets and optimize toward similar objectives, diversity of market behavior may gradually decline.
The concentration issue extends well beyond software. Modern AI depends upon a remarkably narrow technological supply chain. A relatively small number of cloud providers host much of the world’s AI computing capacity. Advanced semiconductor manufacturing depends upon only a handful of firms. GPU production remains concentrated among very few suppliers. Large language models themselves are increasingly dominated by several major technology companies. This concentration creates efficiency during normal operations but raises resilience questions during periods of disruption. Efficiency, in some cases, creates fragility.
AI, Fraud and Cybersecurity
AI is strengthening both sides of the cybersecurity equation simultaneously. Financial institutions are deploying increasingly sophisticated AI systems to detect fraud, identify cyber intrusions and strengthen operational resilience. At the same time, criminal organizations and nation-state actors are using many of those same technologies to automate phishing campaigns, generate convincing deepfakes, identify software vulnerabilities and scale financial fraud at unprecedented speed. The result is an accelerating technological arms race whose outcome could influence not only individual institutions but confidence in financial markets themselves.
For central banks and financial stability regulators, maintaining confidence is nearly as important as maintaining liquidity. Markets function because investors believe transactions are authentic, counterparties are genuine and financial infrastructure can be trusted. Artificial intelligence has the potential to reinforce those assumptions—or undermine them.
Banks increasingly use machine learning to monitor millions of transactions in real time, identifying suspicious behavior that would be virtually impossible for human analysts to detect manually. AI systems continuously analyze login patterns, payment instructions, customer behavior and network traffic, allowing institutions to recognize anomalies before losses accumulate.
Unfortunately, the same technologies improving financial security are also becoming available to attackers. Generative AI has dramatically lowered the cost of producing convincing phishing emails, fraudulent investment solicitations and business email compromise campaigns. Large language models enable criminals to generate grammatically flawless messages tailored to specific industries, executives or institutions. Translation capabilities have likewise removed many of the linguistic clues that once exposed fraudulent communications.
Deepfake technology presents an even more troubling development. Voice cloning systems can now reproduce executive voices with remarkable accuracy after training on only brief audio samples. Video synthesis continues improving rapidly, raising concerns that attackers could eventually impersonate senior executives, investment professionals or government officials during live communications.
Financial institutions have already experienced isolated incidents involving AI-assisted impersonation. Regulators increasingly worry that future attacks could target market-moving events, corporate earnings announcements or payment authorization processes.
So what, right? Historically, most cyber incidents affected individual organizations.
Artificial intelligence raises the possibility of cyber events occurring at systemic scale, enabling a coordinated attack simultaneously targeting multiple financial institutions through AI-generated malware that continuously adapts to defensive measures. Or, AI systems autonomously discovering software vulnerabilities faster than organizations can patch them. Or, sophisticated disinformation campaigns spreading fabricated reports regarding bank solvency, payment system failures or geopolitical crises across social media, financial news platforms and automated information channels.
Simply generating sufficient uncertainty could temporarily impair market functioning. Modern financial markets increasingly depend upon digital information flowing continuously and accurately. Artificial intelligence can dramatically increase both the speed and credibility of false information.
The concern is not hypothetical. Several governments have already warned that generative AI may significantly enhance information operations conducted by criminal organizations and nation-state actors. Financial markets—which respond almost instantaneously to breaking news—represent particularly attractive targets. Recognizing these emerging threats, international regulators are placing greater emphasis on operational resilience than ever before.
The Financial Stability Board’s June consultation on responsible AI adoption stresses that governance, third-party oversight, human accountability and cyber resilience must evolve alongside AI deployment. The objective is not to slow innovation but to ensure institutions understand the operational dependencies they are creating as AI becomes integrated into critical financial functions.
AI, the Global Economy and the Future of Market Stability
Artificial intelligence is beginning to look less like another technology cycle and more like a new general-purpose economic platform—comparable to electricity, the steam engine or the internet. If that assessment proves correct, the implications extend far beyond technology companies or financial institutions. In a mere decade, we’ve gone from talking about cool software tricks and relatively rare supercomputers to hyperscalers and AI factories. AI could fundamentally alter labor markets, corporate profitability, international competitiveness, government finances and, ultimately, the long-term stability of global capital markets.
The most optimistic forecasts surrounding artificial intelligence continue to center on productivity. Transformational economic shifts rarely unfold in a straight line. Periods of extraordinary productivity are often accompanied by significant dislocation, creating years in which markets struggle to distinguish temporary adjustment from permanent value creation.
A growing body of real-world research suggests that AI indeed accelerates software development, improve scientific research, reduce administrative costs, enhance customer service and allow knowledge workers to complete routine tasks dramatically faster than before. The gains are increasingly appearing outside the technology sector.
Manufacturing companies are deploying AI for predictive maintenance and quality control. Pharmaceutical firms are using AI to accelerate drug discovery. Banks are integrating AI into credit analysis, compliance and client servicing. Wealth management firms are using AI to summarize meetings, prepare investment research, monitor portfolios and personalize client communications.
The PwC 2026 AI Jobs Barometer found that industries with higher AI adoption continue to experience faster productivity growth and stronger wage growth for workers possessing complementary skills. Rather than simply replacing jobs, the report argues that AI is reshaping job composition by increasing demand for employees capable of supervising, interpreting and deploying intelligent systems effectively.
Similarly, economists at Goldman Sachs continue to argue that AI’s long-term economic contribution will depend less on the technology itself than on how rapidly businesses reorganize around it. Historically, transformative technologies have generated their largest productivity gains only after organizations redesigned workflows rather than merely automating existing ones. That distinction matters. The biggest beneficiaries of AI may ultimately be companies that successfully reinvent their business models—not necessarily those that first purchase AI software.
Research from the Yale Budget Lab continues documenting AI’s evolving effects on employment, highlighting that exposure varies significantly across occupations. Professional services, finance, law, accounting, consulting and administrative functions appear particularly susceptible to automation, even as entirely new categories of work emerge.
Dan Katz of the International Monetary Fund argues that AI should be understood as an “economics of adjustment” problem rather than simply a technology problem. The greatest challenge may not be whether economies ultimately become more productive, but whether workers, businesses and institutions can adapt quickly enough to avoid prolonged periods of disruption.
If productivity gains arrive gradually while labor displacement occurs rapidly, consumer spending could weaken before economic benefits fully materialize. Corporate earnings may become more volatile as businesses incur significant AI investment costs before realizing efficiency gains. Governments could face rising pressure to expand workforce retraining, education and social support programs.
Artificial intelligence is also reshaping the global competitive landscape. Unlike previous waves of enterprise software, AI has become intertwined with national security, semiconductor manufacturing, energy infrastructure and industrial policy. Governments increasingly view advanced AI capabilities as strategic assets comparable to telecommunications networks, aerospace manufacturing or nuclear technology.
Countries are subsidizing semiconductor fabrication plants, investing in domestic computing capacity, strengthening export controls and competing for access to advanced chips, electrical generation and AI talent. Supply chains may become less efficient as governments prioritize resilience over globalization. Capital allocation may increasingly reflect national security priorities rather than purely commercial considerations. The result could resemble previous periods in which geopolitical competition reshaped entire industries: a cold war.
Will AI Create Greater Inequality?
Several economists argue that artificial intelligence could increase productivity while simultaneously concentrating profits among relatively few firms possessing proprietary models, computing infrastructure or unique datasets. This possibility has become increasingly relevant as AI infrastructure spending concentrates among hyperscale cloud providers and a relatively small number of semiconductor manufacturers.
From a financial stability perspective, concentration matters. Markets become more vulnerable when broad equity performance depends heavily upon a limited number of companies or sectors. Recent years have already demonstrated how AI-related firms can disproportionately influence major stock indexes.
Perhaps the most striking feature of recent discussions—from the Financial Stability Board, the International Monetary Fund, the Bank for International Settlements, central banks, investment banks and academic researchers—is their balance. Few credible observers argue that artificial intelligence represents a passing fad. Equally few argue that its adoption will proceed without meaningful financial, operational and societal disruption.
The principal threat to market stability may not be artificial intelligence itself, but rather the speed with which capital, institutions and societies attempt to adapt to it.
