Source: The TRADE — 2026-07-16
Summary
A survey of 35 of the world's largest asset managers, conducted by Substantive Research and Aiera, found that 77% now have organization-wide generative AI deployments in place — but adoption is being throttled by data licensing, not technology. The single most valuable input respondents want feeding their internal AI systems is broker research (also cited by 77%), followed by earnings transcripts (57%) and market data (42%). The dominant barrier isn't model quality or integration effort — it's that broker/data licensing restrictions (69%) and compliance/entitlements friction (54%) prevent firms from legally piping that content into their AI systems as machine-readable feeds.
Key Takeaways
- 77% of surveyed buy-side firms have organization-wide GenAI platform deployments already in place — adoption itself is no longer the bottleneck at large asset managers.
- The content buy-side firms most want as a machine-readable feed into their AI systems is broker research (77%), ahead of earnings transcripts (57%) and market data (42%).
- The top-cited barrier is broker/data licensing restrictions (69%), followed by compliance and entitlements friction (54%) — a legal and commercial problem, not a technical one.
- Onboarding timelines are still long even where licensing isn't the blocker: 37% report 4-6 months to approve and onboard a new AI model, 20% report over 6 months.
- Just over a quarter of respondents are evaluating or have implemented specialized finance/investment-research AI platforms, with 44% viewing them as potential long-term strategic partners over general-purpose tools.
- The finding reframes the "AI adoption gap" narrative for finance: the ceiling isn't model capability, it's who owns the rights to feed proprietary research content into an AI pipeline.