The Wrong 2030 Debate?

The Future of Finance Isn’t About AI. It’s About Role Design

Cognizant, a global technology consulting firm, announced on April 30 that it would lay off around 4,000 employees as part of “Project Leap“, a programme to reduce costs and reinvest the savings into AI capabilities and workforce upskilling.

Late last year, Accenture announced 11,000 layoffs under what its CEO called “exiting people on a compressed timeline where reskilling is not a viable path.” Two data points among many in the past year.

These headlines are more than just “cost stories”. It is a fundamental re-architecture of the future operating model and org design. The cuts are not falling on entry-level work, where AI use was supposed to hit first.

They are falling in the middle. The mid-management knowledge workers. The people who are doing coordination, business partnering, and synthesis. Simultaneously, firms are hiring thousands of fresh graduates into roles built around AI from day one. The pyramid is not shrinking. It is, as Cognizant’s CEO has put it, being rebuilt with a different shape — broader at the base, hollow through the middle.

This is exactly how finance teams of 2030 are being shaped up.

Finance is primarily a coordination function. It translates operations into numbers, analysis, risks, and reports. The middle layer exists to do that translation, and this is precisely what AI is able to replace now. Agentic AI can autonomously collect, organise, and synthesise information, without the need for a human layer.

The current upskilling narrative assumes both future roles and AI are static phenomena and that new skills can bridge the two. However, in a likely future where these coordination roles will disappear, you are effectively upskilling for something that will not exist. And those roles are where your next CFO is supposed to come from.

The pipeline that produces finance leaders runs through exactly the layer being compressed. Once this layer gets thinner, you don’t lose one cohort of future CFOs. You lose the structure that produces them at all.

Now, this is not a workforce readiness problem, which assumes one broad stroke “upskilling” as a solution. It is a role design problem. And it cannot be solved by retraining for roles that won’t exist. It can only be solved by designing roles that the future finance function will need.

The 2026 State of AI in Enterprise Study by Deloitte found that 53% of leaders are investing in AI fluency and 48% in upskilling.

However, only 33% are redesigning roles and career paths. The first two are the visible moves. They are easy to announce and easy to measure. The third is the one that addresses the actual problem; it is also the one most firms are overlooking.

JP Morgan is amongst the 33%. Through its AI rollout and adoption, its headcount has stayed roughly steady. However, the composition of roles has shifted. Operations roles have fallen by 4% and support roles by 2%, while client-facing roles have grown 4%. Their CEO, Jamie Dimon, described it plainly: “We have displaced people from AI, and we offer them other jobs.”

That is not retraining. It is redesigning. Roles are being closed, but different ones with an entirely different value proposition are being opened, and people are being moved across the gap deliberately.

For CFOs to be able to stay ahead of this, the redesign starts now or it doesn’t start at all. By 2028, boards will be making these decisions under pressure. The window for deliberate redesign is shorter than it looks.

Redesign is harder than upskilling because it forces a decision that most firms defer. It requires asking the tough questions, like which roles will not exist in five years, and which new ones will. CFOs end up making these decisions under pressure rather than redesigning thoughtfully.

The 2030 debate about which AI models will dominate, which agents will mature, and which capabilities will arrive in which order is the wrong debate to have. None of it matters if the roles around the technology are not redesigned. A more capable AI dropped into an unredesigned function does not transform it. It just exposes how much of the function existed to do work that AI now does.

The finance team of 2030 will not be defined by the technology available to it. It will be defined by which CFOs decided, in 2026 and 2027; which roles will not be relevant, which new ones will open, and which people to move across the gap. The rest will be running succession plans for a function that will look nothing like the one they are leading now.

If we step back, the problem is more fundamental.

It is whether you can recognise which roles you are protecting today versus the roles your function actually needs.

Whether, if you were designing this function from scratch with the technology now available, you would design it as it stands.

Whether you would hire yourself into the role you currently hold.