Artificial intelligence is prompting financial institutions to reassess how they source, consume and manage market data, potentially creating new opportunities for providers seeking to challenge established vendor relationships. As AI-driven workflows become more prevalent, data quality, governance and delivery flexibility are emerging as increasingly important considerations alongside traditional requirements for coverage, performance and cost.
These are among the findings of the fourth annual Market Data Study from SIX and Crisil Coalition Greenwich, which surveyed 73 buy-side and sell-side firms across the US, UK and Europe. The research found that 86% of respondents are using or evaluating AI in their investment and trading workflows, while 88% believe market data providers can play an important role in ensuring AI accuracy and reliability.AI Raises the Bar for Data Quality
Data quality has consistently ranked as a leading priority for firms surveyed. However, AI introduces additional considerations as financial institutions begin incorporating automated analysis and decision support into their workflows.
“We have conducted this study for four years now, and quality always trumps everything else,” says Matthew Nurse, Head Market Data, Financial Information at SIX, in conversation with Market & Alt Data Insight. “Now and with the increasing use of AI, it’s even more important. It really does raise the bar for all market data vendors. As firms increasingly rely on AI-assisted workflows and decision support, there is a greater responsibility to ensure the quality, provenance and governance of the underlying data.”
The implications extend beyond the accuracy of individual data points. AI applications need sufficient contextual information to interpret datasets correctly, while providers must ensure that their sourcing, normalisation and quality-control processes can support increasingly automated consumption.
Nurse cautions that AI has the potential to amplify errors when inaccurate or poorly understood data enters automated workflows. This helps explain why regulated financial institutions remain relatively cautious about allowing AI systems to make consequential decisions without appropriate oversight.
MCP Changes the Role of Market Data Teams
The research also highlights growing interest in Model Context Protocol (MCP), which enables AI applications to interact with external data sources and tools through a standardised interface.
Some 58% of respondents are considering or already engaged with MCP, with 57% of those engaged using or testing an MCP server.
As AI agents become capable of discovering and retrieving data directly, some established market-data management processes could become increasingly automated. However, Nurse believes this will increase rather than diminish the importance of human oversight.
“I don’t think it replaces the market data function, but I think it changes its focus. Somebody still needs to determine which datasets are trusted, licensed and approved. Historically, market data teams focused heavily on sourcing, contracts, entitlement management and vendor relationships. Increasingly, their role will extend into data governance, quality assurance, usage controls and AI readiness. The big change is the quality-control mechanism and oversight function which will become even more important.”
The growing involvement of chief data officers, data scientists and governance specialists reflects this evolution. Market data teams will increasingly need to work alongside these functions to establish which datasets can be used by automated systems, under what conditions and with which controls.
Agentic Consumption Challenges Traditional Licensing
The emergence of AI agents also raises questions about how market data usage should be licensed, monitored and charged for.
Traditional licensing arrangements generally distinguish between human users, applications and redistribution activities. Autonomous systems introduce additional complexity, particularly where AI applications combine information from multiple sources, generate derived outputs or use licensed datasets to support other automated processes.“This is one of the defining questions facing the industry,” notes Nurse. “If you look at contracts from even a couple of years ago, they were designed around human use, terminal use, application use and standard redistribution models. The new consumption patterns created through AI introduce autonomous data use and large language models learning from data. I don’t think the intellectual property and licensing model itself is going to change fundamentally. What needs to evolve is how you measure, monitor and audit that usage.”
Data owners will continue to expect compensation for their intellectual property, but existing mechanisms for tracking consumption may need to accommodate new forms of automated access.
One emerging example is SIX Exchange’s recently launched Market Signal product, which provides consolidated best bid and offer information across its exchange portfolio, including Swiss market data outside the scope of the European consolidated tape. Designed primarily for retail consumption, the product explicitly permits AI and large language model use.
Nurse suggests such initiatives could provide useful experience as the industry considers how to accommodate AI consumption within institutional licensing arrangements, although approaches currently vary considerably between trading venues.
Rising Costs Encourage More Selective Consumption
These developments are unfolding against a backdrop of increasing market data expenditure.
The study found that 60% of respondents expect their market data budgets to increase over the next 12–18 months, while 78% are willing to pay for lower-latency delivery. Nevertheless, cost consciousness is rising as institutions face growing expenditure on real-time data, storage, infrastructure and AI applications.
New delivery technologies could provide opportunities to manage those costs more effectively. Rather than subscribing to comprehensive feeds, institutions can increasingly use request-response APIs to obtain information for specific securities, portfolios or applications. Vendors are consequently adapting their commercial models to reflect more targeted consumption.
Nurse also points to the European consolidated tape as an example of the importance of matching data quality and cost to individual requirements. Consolidated indicative pricing may be suitable for retail applications, while trading, execution and regulatory processes continue to require more specialised data sources.
New Opportunities for Market Data Providers
The combination of changing consumption patterns, greater cost scrutiny and new delivery technologies could also encourage institutions to reconsider longstanding vendor relationships.
“Historically, the institutional market data landscape has been dominated by a relatively small number of large providers,” observes Nurse. “Advances in delivery, cloud and now AI driven workflows are encouraging and opening consumers to reconsider provide relationships and a broader range of services. Technological change also creates an opportunity for vendors that didn’t really exist ten years ago, when the prevailing attitude was often that existing systems were in place, they worked adequately and there was little reason to change or challenge them.”
The range of delivery mechanisms now available, from traditional file transfers and streaming APIs to cloud platforms and MCP, is expanding the options available to data consumers.
Established providers retain advantages in aggregation, normalisation and breadth of coverage. However, institutions increasingly have opportunities to evaluate alternative providers according to the quality, accessibility and commercial suitability of individual datasets.
For market data vendors, the challenge will be to accommodate these increasingly diverse requirements while maintaining the reliability and governance standards on which institutional customers depend. As AI adoption progresses, the ability to deliver appropriate data through the right technology and commercial model could become an increasingly important source of competitive differentiation.
Subscribe to our newsletter
