Marsh Talks Up AI, but the Money Is in Infrastructure

Marsh reported fourth-quarter revenue of $6.6 billion, up 9% year over year, and full-year revenue of $26.98 billion, a 10% increase.

On its fourth-quarter earnings call, Marsh said investments in AI and digital infrastructure would support future growth alongside continued margin expansion and cost discipline, framing technology spend as part of a longer-term operating strategy rather than a near-term revenue driver.

Despite repeated references to AI during the call, Marsh did not present AI as a distinct revenue category. Executives disclosed no AI-specific bookings, pricing, or client adoption metrics, instead describing AI as embedded within internal operations, restructuring, and productivity initiatives tied to scale and efficiency.

That positioning contrasts with how some peers have chosen to market AI tools more explicitly. Marsh’s disclosures suggest a deliberate decision to treat AI as internal infrastructure rather than as a standalone commercial product, even as the company increases spending in the area.

AI as Operating Leverage

Marsh’s primary vehicle for deploying AI is Thrive, a multi-year restructuring and growth program launched in 2025. The company said Thrive is expected to deliver approximately $400 million in annual savings over three years, while requiring roughly $500 million in total charges to execute. Marsh recorded $112 million in Thrive-related costs in the fourth quarter alone.

John Doyle, Marsh’s president and CEO, described AI as part of a broader operating-model change rather than a commercial offering. Through its Business and Client Services (BCS) unit, Marsh is “building a data and technology ecosystem that harnesses AI and advanced analytics to improve client outcomes and drive operational excellence,” Doyle said.

Doyle said “dozens of AI-driven productivity tools” have already been introduced to employees. He framed those tools as a way to increase efficiency and output, not to reduce headcount. “These tools are going to make our people better and more efficient and able to serve clients in a better way,” he said.

Management did not provide guidance on AI-driven job reductions. That stance differs from some competitors that have publicly linked AI adoption to automation-related workforce changes or appointed senior executives specifically to lead AI commercialization.

At Marsh Management Consulting, executives were explicit about the limits of AI monetization. CEO Nick Studer said the firm is not yet being paid for AI capabilities themselves. “We’re not really being paid for the things that AI can do at this stage,” he said. “We’re paid for helping clients deliver outcomes.”

Studer said roughly 30% of consulting work already draws on advanced analytics and AI, including techniques used before large language models became widely available. He added that AI tools have improved productivity but have not reduced demand for junior consultants, noting that the firm expects to hire the same or more entry-level staff who are able to use those tools effectively.

Taken together, the earnings call positioned AI primarily as a margin and scale enabler, embedded inside restructuring, technology consolidation, and consulting delivery, rather than as a product line expected to generate near-term revenue growth.

Digital Infrastructure is Where Marsh Expects Growth

While AI itself was framed as internal infrastructure, Marsh was far more specific when discussing digital infrastructure. Doyle said the company expects roughly $3 trillion in global digital infrastructure investment over the next five years, citing it as a core growth opportunity.

Martin South, CEO of Marsh Risk, said between 2,000 and 3,000 data centers are expected to be built worldwide over that period. He said Marsh already holds a leading position in the segment and claimed that Marsh U.S. had the leading market share of an estimated $205 billion in data center construction packages in 2025.

That activity did not translate into outsized growth in the most recent quarter. Marsh Risk reported fourth-quarter revenue of $3.7 billion, up 10% on a reported basis and 3% on an underlying basis. Doyle said digital infrastructure contributed to results in 2025 but added that “there’s much more in front of us than is behind us” as the build-out continues.

The call made clear that the primary monetization path lies in capital structuring, not insurance placement alone. Guy Carpenter CEO Dean Klisura said the scale of data center risks is forcing new approaches. “It’s not going to just be traditional reinsurance capital,” he said, pointing to third-party capital, sidecars, and securitization.

Klisura described this as “the single biggest new business opportunity in 2026.” Marsh said its Nimbus insurance facility supporting digital infrastructure risks recently doubled its capacity to $2.7 billion.

Capital markets activity already reflects that shift. Marsh reported that 86 catastrophe bonds were issued in 2025, totaling more than $24 billion in limits, while casualty sidecars and structured reinsurance solutions attracted significant new capital.

Other Marsh businesses are aligned with the same infrastructure theme. Mercer reported fourth-quarter revenue of $1.62 billion, up 9%, and executives linked demand to workforce planning, global mobility redesigns, and skills development tied to data center and semiconductor projects.

Across the call, executives repeatedly linked AI adoption to increased demand for data centers, power generation, and complex risk aggregation, but stopped short of presenting AI itself as a revenue catalyst. For 2026, Marsh guided to underlying revenue growth similar to 2025, despite what it described as a historically large infrastructure investment cycle.

The disclosures show a company investing heavily in AI while resisting the urge to sell it as a product. Marsh’s growth narrative instead centers on advising, insuring, and financing the physical systems that make large-scale AI deployment possible, and on using AI internally to operate that business more efficiently.