S03.9 · Financial Services & Capital Markets
Exchange operators and market infrastructure providers, where AI automates surveillance but core network moats hold.
Exchange operators and the market infrastructure beneath them generate roughly $50-60B in global operating revenue (2024/2025E, WFE Market Highlights) against a global listed market capitalization of roughly $115-120T (2025, WFE). The fee pool is a tiny fraction of the asset value it supports, and that asymmetry is part of why the moats hold. The commercial story is the migration away from cyclical trading-fee revenue toward recurring data, index and post-trade revenue — and this is one of the few segments in financial services where the core moats, network effects and exchange-operator licensing, remain structurally intact even as AI automates the compliance function around them.
Revenue rose mid-to-high single digits in 2025 on trading-volume growth plus expansion of the data and index businesses, and forward growth projects at ~5-7% as exchanges continue diversifying beyond trading fees specifically. The diversification is the point: trading fees swing with the cycle, data and index licensing recurs, and the revenue mix is shifting toward the latter. US exchanges hold the largest listed market capitalization globally, Asia has the fastest-growing trading volumes, and Europe's exchange landscape has consolidated into a smaller number of larger groups than a decade ago. By the standards of financial services generally, concentration is extreme. A handful of groups — ICE, CME, Nasdaq, LSEG, Deutsche Börse and HKEX among them — dominates developed-market trading and listing activity, and in many other jurisdictions the national exchange operates as an effective monopoly.
Upstream, the value chain runs through listed issuers and liquidity providers; downstream, it serves brokers, asset managers and a growing base of pure data and analytics consumers who never touch the trading venue directly. Core trading economics are high-margin and capital-light, underpinned by the strongest network effect in finance — liquidity attracts more liquidity — and increasingly supplemented by recurring data and index licensing revenue that behaves like a subscription business rather than a transactional one. Exchange-operator licensing is heavy, as befits systemically important infrastructure, and the licensing burden itself ranks among the strongest moats in the sector.
The adjacencies sit along the exchange's own value chain: data and index licensing, post-trade clearing and settlement, and trading technology. Building out data and index capability diversifies away from trading-fee cyclicality — trading volume swings with market volatility in ways a recurring licensing business does not, so this is defense against the cycle rather than pursuit of new demand. Vertical integration into clearinghouses and central securities depositories extends the exchange's economic footprint further along the post-trade value chain it already touches, and that is where genuinely incremental fee capture lives.
Agentic AI's clearest application is market surveillance and compliance monitoring — work that benefits from processing enormous volumes of trading data in real time to flag anomalous or manipulative activity. The compliance-staffing cost line compresses accordingly, across exchange operators. What does not move is the core competitive structure. Network effects and licensing gating are not the kind of moat AI erodes; they rest on regulatory status and accumulated liquidity, not on labor-intensive processes that automation can substitute for. That makes this segment the outlier in financial services: AI is a cost-line story here, not a moat story.
The longer-term change is the emergence of tokenized, 24/7 markets operating as parallel infrastructure alongside traditional exchanges. Tokenized securities and round-the-clock settlement cut against the batch-based, business-hours model that has defined exchange infrastructure for decades, and they pose a real question — do incumbent exchanges extend into the new category, or does a separate infrastructure layer capture it instead? Surveillance and compliance automation is already underway. Tokenization's impact on market structure is a nearer-term call than most of the disruption discussed elsewhere in financial services — roughly 2-5 years — because so much of the regulatory and market-structure groundwork is already in motion.