Companies cutting staff to fund AI projects generate no better returns than those that have not – and the firms producing the strongest results are investing in human skills alongside the technology, not instead of it. That is the central finding of a Gartner survey published on 5 May 2026, covering 350 large enterprises across global markets that had already deployed or were piloting AI agents, intelligent automation, or autonomous technologies. For financial services firms in Geneva and Dubai currently weighing AI-driven restructuring, the data is a direct challenge to a widely held assumption.
The findings are set out in Gartner's press release "Autonomous Business and AI Layoffs May Create Budget Room, but Do Not Deliver Returns" (5 May 2026). Of the 350 executives surveyed, 80 percent reported workforce reductions linked to AI or automation. The workforce reduction rate was nearly identical among those reporting high return on investment (ROI) and those reporting modest gains or negative outcomes – no measurable link between cutting headcount and generating AI value.
How the study was conducted
Gartner conducted the survey in Q3 2025, targeting organisations with annual revenue of at least $1 billion that had been piloting or had already deployed at least one of: AI agents, intelligent automation, or robotic process automation (RPA). The sample of 350 executives spanned multiple industries and geographies. The research was led by Helen Poitevin, Distinguished VP Analyst at Gartner, whose focus covers AI adoption and workforce strategy.
One caveat worth noting: the methodology relies on self-reported executive data. Outcomes reflect business leadership perception as much as audited financial performance. The directional finding is clear; the precise magnitude of the gap is harder to establish with certainty.
Why the wrong question is being asked
Most executive conversations about AI begin with cost. How much will this save? How many roles can it absorb? The framing is understandable – AI has been positioned primarily as an efficiency instrument, and efficiency has a direct line to the cost base.
But Gartner's research points to a different variable: the quality of human judgement applied to autonomous systems. Organisations that build oversight capability – people who understand what AI is doing and why, who can interrogate its outputs and catch its errors – outperform those that simply remove the human layer.
This is not an argument against AI adoption. It is an argument against lazy adoption.
What the high performers actually do differently
The distinction is not between companies that use AI and those that do not. It is between those that use it thoughtfully and those that deploy it primarily to justify a reduction in headcount.
High performers invest in technical literacy at management level, build explicit review mechanisms into AI-assisted workflows, and treat model outputs as inputs to human decisions rather than as decisions themselves. They are notably less likely to frame AI transformation as a cost story.
For financial services firms in particular, this matters beyond performance metrics. FINMA has flagged AI-related risk in its supervisory communications, and the EU Artificial Intelligence Act (EU AI Act) imposes transparency obligations on high-risk deployments. A governance gap is not just a performance drag – it is a compliance exposure.
"Workforce reductions may create budget room, but they do not create return." – Helen Poitevin, Distinguished VP Analyst, Gartner
The talent question nobody is asking
There is a secondary finding in the Gartner data that deserves more attention: the firms cutting to fund AI are also the ones struggling most with retention.
Experienced staff who understand the business deeply – who know the client, the context, the regulatory history – are exactly the people needed to supervise AI effectively. When those people leave, the oversight capacity goes with them. What remains is the automation, operating with less correction, less contextual understanding, and more risk.
The organisations generating the best returns recognised this before it became a problem.
What this means for your IT infrastructure
AI-assisted workflows require reliable, secure, auditable environments. The ability to interrogate what a model did, when, and on what data, is not a nice-to-have. It is a governance requirement.
The underlying infrastructure – whether cloud, private hosting, or hybrid – needs to be built for accountability as much as for performance. Logging, access controls, data residency, auditability: these are not features that can be added when a regulator asks a question.
The honest conversation
The honest question is not "how much can AI save?" but "what kind of organisation do we want AI operating in?" The answer to the second determines the answer to the first.
Penta works with financial services firms across Geneva and Dubai navigating exactly this conversation. The infrastructure decisions being made now – on hosting, security architecture, data governance – will define how much control organisations retain as AI becomes more embedded in their operations.
Key findings from the Gartner survey (Q3 2025, n=350)
- 80 percent of organisations piloting or deploying autonomous technologies reported workforce reductions
- Workforce reduction rates were nearly equal among high-ROI and low-ROI respondents – no correlation between layoffs and returns
- High performers invest in skills, roles, and operating models that allow humans to guide and scale autonomous systems
- Gartner predicts: by 2027, 50 percent of enterprises without a people-centric AI strategy will lose their top AI talent
References
Gartner, "Autonomous Business and AI Layoffs May Create Budget Room, but Do Not Deliver Returns," 5 May 2026: gartner.com
FINMA Annual Report 2025: finma.ch
EU AI Act, Regulation (EU) 2024/1689: eur-lex.europa.eu
