AI Research Spending Jumps 40% While Asset Managers Keep…

AI Is Taking a Larger Share of Existing Budgets

AI-enabled research analytics and tooling now account for approximately 7% of the average research budget in Substantive Research’s survey. That spending competes with allocations traditionally directed toward investment banks, specialist research houses and independent analysts.

The commercial distinction is important. A fund manager buying an AI-powered research platform may be paying for software that retrieves, organizes, compares and summarizes information produced by other providers. The technology supplier captures a share of the research budget even when the underlying intellectual content originated with a broker.

This changes how value is distributed across the research supply chain. Investment banks historically monetized access to analysts, written reports, financial models, conferences and corporate management. AI-based workflows allow institutional clients to consume some of that information through search, retrieval and automated analysis rather than direct interaction with its original producer.

The underlying research may remain valuable, but its delivery becomes less visible to the end user. Providers therefore face the challenge of preserving attribution, licensing control and pricing power when their content is incorporated into larger analytical systems.

Can Banks Charge More for Research Delivered Into LLMs?

Substantive Research found that sell-side firms have not yet succeeded in generating meaningful additional revenue from delivering research directly into buy-side large language models.

Mike Carrodus, CEO of Substantive Research, identified the central commercial question as whether delivery into a client’s AI system represents another distribution channel or a distinct use of intellectual property that warrants additional payment.

The answer could affect the economics of research licensing. A traditional institutional research subscription may authorize named users to read reports through a portal. Machine ingestion introduces different questions concerning storage, indexing, extraction, redistribution and the generation of derivative analytical outputs.

Providers may seek additional fees for machine access, particularly where automated systems make research available across a larger group of portfolio managers and analysts. Buy-side firms, however, already face rising expenditure on market data, infrastructure and analytics.

With overall research budgets effectively flat, additional licensing charges would probably require reallocations from existing suppliers unless investment managers obtain approval for higher expenditure.

US Research Budgets Remain Larger Than Europe’s

The survey found that North American research spending declined 6% as a proportion of assets under management, reversing a recovery trend seen over the preceding three years. American research budgets remain more than 50% below their 2018 levels on the study’s comparison.

European research spending increased by just over 1%, but budgets remain more than 60% below their 2018 levels. Substantive Research estimates that European asset managers still spend approximately 40% less on research than their American counterparts.

The regional difference reflects, in part, the consequences of MiFID II’s research-payment reforms. The original European framework restricted the practice of funding research through dealing commissions and pushed asset managers toward explicit research budgets.

That policy improved visibility into research costs but also created pressure to reduce the number of providers and the amount spent on external analysis. Earlier FinanceFeeds coverage of Substantive Research documented concerns about the research market following MiFID II and changes to the SEC’s treatment of research payments.

The current figures indicate that the structural spending gap has persisted even as AI creates new demand for investment information.

Will the FCA’s Research Payment Reforms Change Spending?

The UK Financial Conduct Authority has introduced greater flexibility for asset managers to pay for research through arrangements combining dealing commissions and research charges.

Its PS24/9 policy statement introduced new payment options for investment managers, while subsequent changes extended the framework to pooled funds.

The objective is to address concerns that strict research unbundling reduced investment in external analysis, particularly research covering smaller companies and less liquid securities.

However, regulatory permission to use a payment method does not guarantee that investment managers will adopt it. Asset owners may question the transfer of costs back to investment portfolios, especially when managers have spent years presenting explicit research budgeting as an efficiency improvement.

Substantive Research expects a more substantial transition toward commission-sharing arrangements, but the latest survey suggests adoption has not yet produced a material increase in overall expenditure.

Research Spending Is Concentrated Among Established Providers

The survey found that the 10 largest research providers receive 56% of the average institutional research budget, with the three largest accounting for 23%. Independent Research Providers receive only 8% of average expenditure.

This concentration presents an additional obstacle for smaller research firms. Large investment banks can combine research with corporate access, sector coverage, global distribution and other institutional relationships. Smaller providers must demonstrate differentiated expertise while competing for a relatively limited portion of spending.

AI could create opportunities for specialist research firms by making their work easier to discover and integrate into investment processes. Equally, it could strengthen larger providers with the resources to build proprietary data systems, negotiate enterprise licenses and distribute research directly into client technology environments.

The outcome will depend on how asset managers evaluate the value of original analysis compared with tools that organize and process it.

What the Numbers Mean for the Sell Side

For investment banks, the immediate challenge is that technological consumption of research is expanding faster than the budgets available to pay for it. Research departments may face higher costs for content preparation, data infrastructure and machine-access controls without a corresponding increase in client revenue.

For asset managers, AI may improve the productivity of research teams, but it does not eliminate the need to assess source quality, methodology and potential conflicts. Automated synthesis can accelerate the processing of information without necessarily improving the accuracy of the original analysis.

The survey’s central finding is therefore a redistribution of expenditure within a constrained market. AI vendors are capturing a larger share, traditional providers are seeking new licensing revenue, and the buy side has shown little willingness to expand the total budget.