The AI Paradox in Asset Management: Widespread Adoption, Elusive Returns

Part 1 of 5 — Gradial FinServ Intelligence Series

Fifty-five percent of asset managers have now integrated AI into at least one investment process, and 91% plan to expand its use over the next 12 months. By any conventional measure of adoption, the industry has moved decisively. Yet only 8% of firms report a measurable improvement in investment returns from that same AI investment. The gap between how widely AI has been adopted and how much value it has actually produced is one of the more interesting paradoxes in asset management today — and understanding why it exists says a lot about where the real opportunity lies.

The explanation isn’t really about the models themselves. Sixty-nine percent of firms cite data quality or access as a material barrier to broader AI adoption, Fragmented fund data is also limiting operational efficiency. Only a minority of firms describe their AI use as broad or fully integrated. In other words, the technology is often capable of more than the underlying infrastructure allows it to do. Portfolio accounting systems, custodial data, market feeds, and analyst research typically live in separate systems, reconciled on separate schedules — and even a highly capable model can only reason as well as the data it can actually see.

What makes 2026 notable is a shift in why firms are investing in the first place. For the first time in three years, competitive differentiation (55%) has overtaken both operational efficiency (33%) and cost control (44%) as the leading driver of technology investment in asset management. That’s a meaningful change in framing — from “are we keeping pace with the industry” to “is this changing our competitive position.” Early data on agentic workflows, which can move across fragmented systems the way a very fast, very thorough analyst might, suggests real headroom here: firms applying these approaches to cross-system tasks report capacity gains in the 55–65% range and meaningful reductions in operational cost.

Many asset managers say AI will only deliver meaningful value once data quality and structure improve. They are pointing to the actual next step, not a hypothetical one. There’s a parallel version of this fragmentation problem one layer up the stack, in how asset managers manage the content built on top of that data: fund fact sheets, product pages, and client-facing disclosures that need to reflect pricing, performance, and regulatory language accurately across dozens or hundreds of pages, on a schedule that doesn’t wait for a production queue to clear. That’s the specific problem Gradial’s agents are built to solve — coordinated updates to regulated content across a firm’s CMS, with brand, accessibility, and compliance checks built into every change and a full audit trail behind it, so marketing and compliance teams aren’t the last function catching up while the rest of the firm modernizes.

Up next in Part 2: why fraud defense in payments is becoming a speed problem, not just a detection problem.

• Mercer: How AI Is Shaping Asset Management
• Mercer: Asset Managers' Use of AI
• InvestmentNews: Most Asset Managers Are Using AI, But Few Let It Call the Shots
• SimCorp: Two-Thirds of Investment Managers Adopting AI (2026)
• BCG: Rebuilding Asset Management for an AI-First World (2026)