Financial Services · Enterprise AI

AI Adoption in financial services in 2026: 81% deploy it, only 14% are transformed by it

AI adoption in financial services is rapidly progressing, though still in mid-transition. The 2026 Global AI in Financial Services Report from the University of Cambridge, covering 628 organizations across 151 jurisdictions, finds that 81% of surveyed financial services firms (out of 352 industry respondents) are adopting AI at some level, with 14% describing it as transformational to their strategy and competitive advantage.

What does that spread really mean? Most institutions bought the technology but did not change the way they operate. The report's own profitability data says as much: only 40% report an increase, while 43% report no change at all.

The profitability divide among AI use cases in financial services

The report compares use-case adoption with reported profitability. The most widely adopted AI use cases (combining pilots and full deployment) are focused on generic operational tasks:

Process automation (79% active adoption; 32% fully deployed)
Data visualization (75% active adoption; 34% fully deployed)
Software development (75% active adoption; 42% fully deployed)
AI-powered customer support (73% active adoption; 27% fully deployed)

Higher profitability shows up elsewhere: in firms embedding AI into complex, specialized domain workflows like new product creation, trading and portfolio intelligence, treasury management, Financial Planning & Analysis, insurance risk, claims management, and climate risk.

The report states this is correlation, not causality, and offers a direct reading: firms able to embed AI into complex, domain-specific workflows extract a higher financial premium than those applying it to generic automation.

That reframes the roadmap question. The race in financial services isn't about who uses the smartest AI model—it's about who owns the workflow that is hardest to replicate. Don't ask "which AI model do we deploy?"—ask "which of our core business workflows is hard enough to become our ultimate competitive moat?"

Three reasons most institutions never get there

The study names the core adoption bottlenecks—and model performance is not one of them:

Legacy and data

Data availability and quality remains the leading barrier for 49% of traditional financial institutions (versus 34% of fintechs), heavily compounded by legacy architecture and siloed environments.

No proof of value

55% of industry respondents cannot readily measure the financial impact of AI deployment—a difficulty that leaps to 76% among large financial institutions.

Supervision and explainability gap

55% of industry respondents cite loss of human oversight as a top risk, yet model opacity remains a major hurdle as institutional governance struggles to keep pace with deployment.

What changes when delivery is built around outcomes

None of these blockers are solved by buying a better model. They are solved by changing how the work is delivered: whether the workflow encodes your domain expertise, how value is measured, and who is accountable for each output. That is a delivery problem, not a technology one.

Built on two decades of enterprise delivery, Globant's answer is the AI Pod: a human-supervised delivery unit scoped to one business problem, built on documented workflows and our domain expertise. Agents execute. A Forward Deployed Engineer supervises every output and answers for it. Clients are charged by consumption or outcome, not by the hour.

Mapped against the three blockers, the AI Pod model matters for specific reasons.

On legacy

AI agents document the legacy logic before rebuilding it which is where traditional projects lose most of their time. A leading bank completed a COBOL migration in 2 months, against the 14 projected with a traditional approach.

On proof of value

Paying by outcome rather than by hour moves the measurement burden to the provider. If the result isn't evidenced, it isn't billed—a direct answer to the 55% who cannot prove returns.

On supervision

Every output has a named Forward Deployed Engineer accountable for it, and Glob.AI OS keeps a record of which model ran, on what data, and under which policy.

In financial services, that delivery model is already applied through Business Reinvention, Process Management and AI Agents for Modernization, with cloud and partner integrations supporting scale.

"AI is expanding the possibilities for technology services like never before. We're not just making the same projects faster, but unlocking 1,000 new ones that were previously not viable."

Guibert Englebienne, co-founder, Globant (Press Release, August 2026)

Choosing a partner for AI Transformation in Financial Services

The sector's average AI portfolio is now a commodity. The report's evidence points to the return sitting one layer deeper, in workflows that carry your domain, your data and your controls and those do not scale without supervision and governance underneath.

Stop Piloting. Start Delivering.

Explore Globant's Financial Services practice and let's scope where AI creates the most value in your institution.

Get in touch

FAQ

Frequently Asked Questions

What is the difference between agentic AI and generative AI in financial services?

Generative AI produces content and responses within a defined task. Agentic AI executes work across processes, with systems that plan and act rather than respond. Adoption reflects that difference in maturity: generative AI has reached 71% among surveyed industry respondents, while agentic AI stands at 52%, with 29% piloting and 23% already Scaling or Transforming. Fintechs lead traditional financial institutions in agentic adoption, 57% to 45%

How mature are generative and agentic AI in financial services?

Generative AI has reached 71% adoption among surveyed industry respondents and agentic AI 52%, with 23% already Scaling or Transforming their agentic deployments.

Is AI already improving profitability in financial services?

Not consistently. 40% of respondents report an increase in profitability and 43% report no change. Productivity gains are more widely perceived which suggests efficiency is arriving faster than financial return