‘BFSI CIOs Must Stop Pumping Money Into More POCs’

The message emerging from the banking, financial services and insurance (BFSI) sector is becoming sharper: artificial intelligence is no longer faltering because of technological limitations—it is struggling because of how enterprises are attempting to scale it. A recent analysis by HFS Research authors Hansa Iyengar and Niti Jhunjhunwala makes that argument clearly, noting that firms are trying to “bolt probabilistic systems onto operating models built for deterministic work.”

The study draws on a roundtable with BFSI executives conducted by the research firm, where one participant described the situation as “death by 1,000 POCs”. Experimentation is proliferating without convergence, leaving firms stuck between “proof of concept” and “proof of promise”, but with little “proof of value”.

The authors frame the issue as structural rather than incremental. BFSI organisations, they argue, are attempting to scale probabilistic AI within operating models designed for deterministic processes, fragmented ownership and legacy constraints. This has led to a pattern in which enterprises treat AI as a tool rollout rather than a workflow redesign, focus on model performance instead of business outcomes, and attempt to scale without robust governance or clear economic frameworks—highlighting gaps in lifecycle control and ‘tokenomics’.

As a result, the report suggests that BFSI firms are unlikely to scale AI by continuing to fund isolated pilots. Instead, progress is being made where organisations redesign workflows, clarify decision rights, embed evaluation and governance mechanisms, and treat AI as a supervised workforce capability tied to measurable outcomes.

This shift is now beginning to reflect in how IT services firms describe BFSI demand.

At Tata Consultancy Services, CEO K Krithivasan said “client demand across BFSI remained technology-led and outcome focused through Q4,” even as macroeconomic and geopolitical uncertainty shaped cautious spending. Clients are prioritising core and legacy modernisation, data transformation, cloud migration and scaled AI deployments aligned to measurable outcomes, moving toward “industrialised business driven transformation… with strong emphasis on cost discipline, regulatory resilience, and measurable outcomes.”

The numbers reinforce this measured shift. With the IT major, BFSI accounted for 31.6% of revenue in Q4 FY26, up from 31.2% a year earlier, with 0.1% quarter-on-quarter and 0.4% year-on-year growth in constant currency. For the full year, the vertical contributed 32.0% of revenue, underscoring stability rather than acceleration.

At Wipro Limited, BFSI remains the largest vertical, contributing 34.1% of IT services revenue in FY25-26, with minimal variation across quarters. Yet, beneath that stability, execution friction is becoming visible.

As management noted, “...the BFSI sector was impacted by delayed ramp-ups on some large deals that were closed earlier this year and by certain client-specific issues. Europe grew sequentially and has remained flat on a year-on-year basis. In fact, we see good traction in the UK, specifically in the BFSI sector.”

At Infosys, financial services accounted for 28.0% of revenue in Q4 FY26, down from 28.4% a year earlier, even as it posted 5.0% year-on-year growth (2.9% in constant currency).

The sharpest signs of pressure are visible at LTIMindtree, where BFSI has declined from 37.1% to 33.0% of revenue over a year, alongside a 4.9% sequential decline and 3.9% year-on-year drop in the latest quarter. That combination points to tighter discretionary spending and slower deal conversions.

And yet, the trend is not uniform. At Persistent Systems, BFSI continues to grow strongly, with 24.3% year-on-year growth in Q4 FY26, albeit moderating from earlier quarters. This suggests that while BFSI demand remains intact, it is increasingly concentrated in programmes that are already scaled or clearly tied to business outcomes.

The shift towards outcome-led AI is also shaping leadership moves across the sector. Mphasis, on April 28, announced the appointment of Richard Miller as Global Head of Insurance Business.

It signals a sharper focus on scaling AI within core BFSI workflows rather than standalone experimentation. Miller, who brings over three decades of experience in enterprise transformation for global insurers, is expected to drive what the company described as “trusted AI-led transformation”, aimed at delivering measurable business value.

His appointment comes at what he called an “inflexion point for the insurance industry,” where clients are increasingly looking for “practical ways to apply AI responsibly at scale,” reinforcing the broader industry shift away from pilots toward production-grade deployments.

These signals across firms indicate that BFSI is not retreating from technology investment, but is becoming more selective in how that capital is deployed. Spending appears to be shifting towards initiatives that can demonstrate repeatable value, while loosely defined or experimental efforts face greater scrutiny.

The conclusion from the HFS Research analysis is unambiguous. As the analysis puts it, “BFSI CIOs must stop pumping money into more POCs.” The path to scale lies not in expanding experimentation, but in reducing it—focusing instead on a smaller set of governed, repeatable AI deployments embedded into core workflows and tied directly to measurable business outcomes.