AI boom didn’t increase market chaos but quietly reshaped financial power

Since the arrival of generative AI systems such as ChatGPT, companies linked to AI infrastructure and machine learning have attracted surging capital flows and unprecedented market attention, raising growing questions about whether the AI boom is fundamentally changing how financial shocks spread across the global economy.

A study titled "Artificial Intelligence and Financial Market Connectedness: Evidence from AI-Related Equities, Cryptocurrencies, and Global Assets," published in the journal FinTech, investigates the impact of GenAI on interconnectedness across global financial markets. Using daily Japanese market data from January 2021 through December 2025, the research analyzed spillover dynamics among AI-related equities, cryptocurrencies, gold, crude oil and the S&P 500 through a time-varying parameter vector autoregression framework.

The findings show that the AI boom did not create a simple increase in overall financial interconnectedness. Instead, generative AI triggered a structural transformation in how shocks move across markets, while traditional equity markets continued to dominate global spillover networks.

AI boom changed market structure rather than intensifying overall spillovers

The study examined financial connectedness across four major asset categories representing different transmission channels within global markets. AI-related equities, including NVIDIA, Microsoft, Alphabet, Meta Platforms and AMD, represented technological innovation and AI-driven economic growth. The S&P 500 served as a proxy for broad market conditions and macro-financial sentiment. Bitcoin and Ethereum represented speculative and decentralized financial activity, while gold and crude oil reflected macroeconomic fundamentals and safe-haven behavior.

Researchers selected the launch of ChatGPT on 30 November 2022 as a major turning point in the AI economy. According to the paper, the release of ChatGPT dramatically accelerated global attention toward generative AI technologies, triggering sharp increases in valuations, trading activity and investor focus on AI-linked firms. The study therefore divided the analysis into pre-ChatGPT and post-ChatGPT periods to examine how the emergence of generative AI influenced financial market spillovers.

Using a time-varying parameter vector autoregression model, researchers measured how shocks originating in one asset class transmitted across the broader financial system over time. The model allowed the study to track changing relationships dynamically rather than assuming stable market interactions throughout the sample period.

The findings revealed that overall market connectedness remained high throughout the study period, with more than half of forecast error variance across markets explained by cross-market spillovers. However, contrary to expectations that AI enthusiasm would sharply intensify market interconnectedness, researchers found that total connectedness actually declined modestly after ChatGPT's release.

The total connectedness index fell from 57.42 percent during the pre-ChatGPT period to 54.69 percent in the post-ChatGPT period. According to the study, this decline suggests that generative AI did not create a broad amplification of financial contagion. Instead, AI altered the structure and direction of spillovers across markets in more complex ways.

The initial emergence of generative AI likely intensified investor attention and accelerated information diffusion across markets, temporarily boosting spillovers. However, as financial markets gradually absorbed information about AI's economic implications, connectedness stabilized and became more dispersed rather than continuously intensifying.

The study describes this process as a structural transformation of financial interconnectedness. Rather than concentrating risk around a few dominant AI firms, the AI boom gradually produced a more distributed spillover structure across global markets.

Researchers also suggest that financial markets may be entering a phase of partial segmentation where AI-related sectors evolve into increasingly differentiated market clusters. As investors gain a clearer understanding of AI business models, productivity expectations and technological risks, AI-related assets may begin reacting more independently from traditional asset classes.

S&P 500 remained dominant shock transmitter despite rise of AI-linked equities

Among others, one key finding was the continued dominance of traditional equity markets in transmitting financial shocks. Despite rapid growth in AI-related firms and increasing global attention toward generative AI, the S&P 500 remained the largest net transmitter of spillovers throughout the sample period.

The study found that the S&P 500's role as a dominant spillover source strengthened further after ChatGPT's release. Its net spillover measure increased from 18.84 in the pre-ChatGPT period to 24.52 afterward, reinforcing the central role of broad U.S. equity markets within the global financial system.

While AI-related firms have gained enormous market influence, broader market structures still anchor financial interconnectedness. Major macroeconomic conditions, investor sentiment and overall equity market dynamics continue shaping global spillover transmission more strongly than individual technological sectors alone.

Among AI-related equities, NVIDIA, Microsoft, Alphabet and AMD initially acted as strong net transmitters of shocks during the early AI expansion period. NVIDIA and Microsoft in particular played major roles in spreading financial shocks across markets as investors increasingly associated them with the future growth of generative AI infrastructure, cloud computing and semiconductor demand.

However, the influence of these AI-linked firms weakened over time. NVIDIA's net spillover measure dropped sharply after ChatGPT's emergence, while Alphabet shifted from a net transmitter to a net receiver of shocks. According to the researchers, this suggests that AI-related equities gradually lost some of their exceptional spillover dominance as markets internalized AI-related expectations and adjusted to the new technological landscape.

Meta Platforms also consistently behaved more as a receiver than a transmitter within the spillover network, indicating that not all AI-linked companies occupied equal positions in financial connectedness dynamics.

The study further found that cryptocurrencies remained highly sensitive to broader equity market developments rather than functioning as independent sources of financial contagion. Bitcoin and Ethereum consistently behaved as net receivers of shocks, suggesting that digital asset markets continued reacting heavily to developments in technology stocks and broader financial conditions.

Cryptocurrencies' dependence on external market conditions reflects their speculative nature and strong sensitivity to investor sentiment surrounding technological innovation. During periods of heightened AI enthusiasm, cryptocurrencies absorbed spillovers originating from equity markets rather than driving contagion independently.

Gold and crude oil also acted primarily as shock absorbers throughout the study period. Gold remained a strong net receiver despite its traditional role as a safe-haven asset, while oil markets reflected broader macroeconomic and global demand conditions rather than transmitting major spillovers themselves.

The findings reveal a highly asymmetric financial network where traditional equity markets remained the system's primary transmission hub even during the height of the generative AI boom.

AI-driven markets may reshape systemic risk and investment strategies

The study argues that the rise of AI-driven markets has important implications for financial stability, risk monitoring and investment strategy. Researchers emphasize that policymakers and regulators should focus not only on the overall level of market connectedness but also on the direction and structure of spillovers across sectors.

According to the paper, directional spillover measures may help regulators identify emerging transmission channels associated with rapidly evolving technological sectors. Monitoring how shocks move from AI-linked equities into broader financial markets could provide early warning signals for systemic vulnerabilities and cross-market contagion risks.

As for portfolio diversification and risk management, researchers argue that investors may no longer rely on traditional assumptions about stable market correlations because AI-driven sectors are changing how financial assets interact dynamically over time. As spillover structures become more complex and dispersed, diversification strategies based solely on historical correlations may become less effective.

Researchers additionally suggest that AI-related financial dynamics may evolve through several phases. Initial technological breakthroughs often trigger intense investor attention and strong interconnectedness as markets rapidly process new information. Over time, however, markets may begin internalizing expectations, stabilizing spillovers and creating more segmented financial structures where AI-linked sectors behave increasingly independently.

The paper further notes that technological innovation may not always increase systemic risk uniformly. While AI expansion clearly altered market relationships, the decline in total connectedness after ChatGPT's release suggests that technological revolutions can sometimes redistribute and diffuse spillover structures rather than simply intensifying financial contagion.

Studying financial networks dynamically instead of assuming constant relationships among assets is also crucial. Traditional rolling-window approaches often rely on fixed parameter assumptions, while the time-varying framework used in the study allowed researchers to capture gradual structural changes associated with evolving market conditions and technological innovation.

The findings remained stable across multiple robustness checks involving alternative lag lengths and forecast horizons, reinforcing confidence that the observed patterns were not driven by technical modeling choices.

Researchers acknowledge several limitations, including the exclusion of government bond and foreign exchange markets from the analysis and the difficulty of fully isolating AI-related effects from broader macroeconomic developments occurring simultaneously during the study period. The paper also notes that classifications of AI-related firms may continue evolving as technological ecosystems change rapidly.