MCP adds a new dimension to AI
Agentic AI is already revolutionizing the information services industry. Now its power is being further boosted by MCP (Model Context Protocol) connectors. These allow users to launch agentic solutions from their preferred large language models (LLMs), combining proprietary data and capabilities with LLM-driven intelligence.
Together, MCP applications and agentic AI are set to illustrate the Jevons Paradox: they are technological advances that increase efficiencies, yet lead to more use, rather than less. By linking AI agents to data sources, MCP is set to accelerate data consumption.
The combination has disruptive potential. Companies are meeting customers where they are, by hooking their data assets directly into third-party platforms. LLMs are partnering with many information services companies to incorporate their data into usable assets.
These companies should benefit through increased pricing, improved retention, and opportunities for cross-selling and upselling. By automating workflows while protecting brand identity, MCP and agentic AI will help to capture substantial value.
Partnership model drives opportunities
Further revenue opportunities are emerging. Anthropic’s launch of vertical platforms, such as Claude for Financial Services, Life Sciences, Healthcare, and Legal, initially raised fears of disintermediation for established players. However, information services companies are finding advantages through the “Bring Your License” partnership model.
The renegotiation of customer contracts to incorporate LLM-delivered solutions creates new opportunities for companies that have prioritized AI-readiness.
Again, these companies are benefiting from incremental usage charges, enhanced pricing power, improved retention, and expanded upsell and cross-sell potential through optimized product packaging.
“The renegotiation of customer contracts to incorporate LLM-delivered solutions creates new opportunities for companies that have prioritized AI-readiness.”
Ashish Sabadra, U.S. Business and Information Services Analyst, RBC Capital Markets
Who’s exposed to increased competition?
One impact of MCP adoption has been to lower switching costs dramatically. MCP servers can be deployed in minutes, replacing months of API and data integration.
This creates pricing pressure for data that lacks strong moats. By allowing enterprises to fragment their data budgets across multiple providers, it also intensifies competition and fragmentation.
Benchmark and proprietary datasets offer the most resilient competitive moats. Benchmark data commands the highest value, due to superior growth, incremental margins, and defensibility from brand and regulatory moats.
Among proprietary data, self-generated data and contributory data networks are best protected, followed by proprietary datasets built from multiple licensed sources.
By contrast, publicly available data faces erosion, as LLMs enhance their abilities to link disparate sources. Systems of record no longer offer genuine moats, since customers retain ownership of their data and can switch platforms.
Development accelerates—and so do costs
Rather than developing software step by step—the traditional “waterfall” model—companies are turning to more agile methodologies. Many are now moving towards the AI Software Development Lifecycle (ASDL), which integrates AI tools into each development phase to improve speed, quality, and decision-making.
ASDL operates with smaller teams and accelerated delivery cycles. This represents a fundamental cultural shift, as companies reallocate incentive compensation toward their top performers, aiming to drive adoption and maintain competitive velocity.
As companies accelerate this transformation, and commit to uncapped AI token consumption agreements, their token costs have risen substantially. However, operational efficiency gains and workforce optimization will ultimately offset these costs.
Rapid LLM evolution should lead to further automation in code generation and review, where the need for development input currently curbs efficiencies. In future, companies will focus on outcomes deploying disciplined token controls and sophisticated governance frameworks, as they unlock substantial opportunities from AI-powered development tools.
Ashish Sabadra authored "RBC ImagineTM: Disrupt or Get Disrupted - Data Moats in the Agentic Economy" published on June 4, 2026. For more information on the full report, please contact your RBC representative.
