The numbers are hard to ignore, and they are landing with enough regularity now that calling it a trend feels like an understatement.
Block cut roughly 4,000 roles, representing around 40% of certain divisions, with executives explicitly framing the decision around AI-driven efficiency. Atlassian followed with 1,600 cuts, over 900 of them in software R&D and engineering. HSBC is now reportedly considering up to 20,000 redundancies as part of a multi-year AI-led overhaul. Goldman Sachs and Citi are said to be exploring similar restructurings. Morgan Stanley analysts estimate that AI could eliminate up to 10% of European banking jobs by 2030, with around 200,000 roles across 35 large banks at risk.
What makes this moment different from previous waves of financial services automation is the speed and the seniority of the roles affected. Past rounds of technology-driven efficiency typically targeted the most routine, lowest-skill work. The current wave is moving into software engineering, R&D, compliance analysis and mid-level operations , categories that were previously considered relatively safe. Atlassian cutting over 900 roles in software development at the same time as expanding its AI product capabilities is a signal.
The question nobody in a boardroom wants to answer out loud is how much of this is a genuine productivity transformation and how much is a cost-correction cycle that AI is providing convenient cover for. In 2025, AI-driven automation was cited as a factor in over 50,000 layoffs globally, though analysts have noted that some companies had already over-hired and would have restructured regardless. Klarna, often cited as a poster child for AI-driven efficiency, has since walked back some of its more aggressive headcount reduction claims.
The situation is complicated, and the people with the most to say about it are the ones closest to the decisions.
The Efficiency Case Stacks Up, So Does The Scepticism
AI tools in finance are already being credited with roughly 30% efficiency gains in areas like report generation, reconciliation and risk monitoring.
For a CFO under margin pressure, that makes workforce reductions look like rational operational decisions rather than cost cuts dressed up as strategy. The math is straightforward: if a compliance function that once required a dozen people can produce the same output with a fraction of that headcount and an AI-assisted workflow, the business case writes itself.
But a 40% workforce reduction is not the output of a careful workflow redesign. It is a blunter instrument than that, and the AI framing is doing a lot of work to make it look strategic rather than reactive. The distinction matters: ‘we’ve redesigned these workflows around AI and some roles are no longer needed’ is a strategy. ‘We’re cutting headcount by 40% and AI is how we’ll manage the gap’ is a bet, and a risky one.
The data on who bears the cost of these bets is also starting to emerge. Tracking data on AI-exposed occupations in the US shows the employment impact is sharply skewed by age. Workers in their early twenties have seen material declines in employment in these roles since 2022, while those in their thirties and above in the same occupations have seen growth. AI-related hiring outpaced AI-driven job losses in 2024 overall, but the distribution is deeply uneven.
The people losing out are disproportionately early-career workers in fintech and financial services, the same people who would historically have built the institutional knowledge that makes senior talent valuable.
The Junior Talent Problem Nobody Is Talking About
There’s a longer-term problem embedded in this wave of cuts that the quarterly numbers don’t capture.
Junior analyst roles, associate positions, entry-level compliance and operations work were never just about cheap labour – they were the training ground. You developed judgment about credit risk by processing loan applications for two years before anyone let you make a recommendation. You learned to read a balance sheet by reading hundreds of them.
When AI absorbs those tasks, the immediate efficiency gain is visible. The longer-term question, of where the next generation of senior talent comes from, is not so apparent.
We asked five experts where they think this is heading.
Our Experts:
- Leigh Coney, Founder and Principal Consultant, WorkWise Solutions
- Gershon Goren, Founder and CEO, Cangrade
- Noah Kenney, Founder and Principal Consultant, Digital 520
- James Lloyd, Digital Strategy Lead, THE LINE, NEOM
- Kelsey Szamet, Partner, Kingsley Szamet Employment Lawyers
Leigh Coney, Founder and Principal Consultant, WorkWise Solutions
“Both, and the honest answer is that most executives making these decisions can’t fully separate the two in their own heads.
“AI is absolutely eliminating specific tasks. The Decidr US AI Readiness Index found that 60% of businesses cite efficiency and cost reduction as their primary driver for AI adoption. When a compliance team that needed twelve people to review transaction alerts can now handle the same volume with four people and an AI triage layer, that’s a genuine productivity shift.
“But the speed and scale of the cuts tell a different account than pure operational logic. When Block eliminates 40% of its workforce and points to AI, that’s not a measured redeployment plan. That’s a cost restructuring with AI as the narrative frame. My research on skill erosion in AI-augmented teams points to a risk that doesn’t show up in the quarterly numbers. When you remove large portions of your experienced workforce quickly, you don’t just lose labour capacity. You lose the institutional judgment, the pattern recognition, the informal knowledge that no AI system currently replaces.
“The practical question for founders is: have you actually redesigned your workflows around AI, or have you just cut people and handed the survivors a ChatGPT login? Those produce very different outcomes. The first builds a more capable organisation. The second creates a thinner team running the same broken processes faster, with fewer people available when something goes wrong.”
Gershon Goren, Founder and CEO, Cangrade
“Honestly? Both. And that’s what makes this moment so hard to read. Some of what Block, Atlassian, and the banks are doing reflects real productivity gains from AI automating work that used to require headcount. But a significant portion is a correction that was coming regardless. These companies over-hired and AI-driven efficiency makes the rebalancing easier to explain. The danger is that organisations conflate the two and convince themselves every cut is strategic. Klarna tried that and they walked it back.
“Stop asking ‘what can AI do that my team does?’ and start asking ‘where does my team create value that AI can’t replicate?’ Those are different questions with very different answers. Founders who cut first and ask those questions later will find themselves rebuilding sooner than they expect.
“The roles being eliminated are the ones built around pattern-matching and high-volume routine work — what AI does well. What’s growing is demand for people who can exercise judgment, work effectively alongside AI tools, and bring genuine adaptability.”
Noah Kenney, Founder and Principal Consultant, Digital 520
“In fintech, trust is the product. If AI adoption introduces opacity, inconsistency, or compliance risk, it erodes enterprise value, even if it improves short-term margins. Fintech founders shouldn’t be asking where to cut headcount with AI, but instead where AI can improve trust, accuracy, and responsiveness.
“The companies that win long-term will be those that use AI to increase revenue per employee while simultaneously strengthening governance, audibility, and customer confidence. Deploying AI tools and reducing staff is not a competitive advantage in itself. The competitive advantage is upskilling existing teams in areas like AI oversight, security, and regulatory alignment.
“What we would currently consider an entry-level role in financial services will likely be redefined rather than eliminated, with the baseline expectation shifting from execution to oversight. The next generation will need to evaluate AI outputs, identify edge cases, and make judgment calls where automation breaks down. As AI handles more of the technical workload, differentiation shifts toward relationship-building, communication, and the ability to represent the institution credibly in high-stakes contexts.”
James Lloyd, Digital Strategy Lead, THE LINE, NEOM
“It is both, but not in equal proportion. In the first half of 2025 alone, 77,999 US tech job losses were attributed to AI. But the displacement is concentrated at the bottom. Entry-level workers aged 22 to 25 have seen a 13% employment decline in AI-exposed occupations since 2022, while workers over 30 in the same fields actually saw growth.
“In 2024, AI-related hiring in the US actually outpaced AI-driven job losses, which supports the cost-cutting cover argument, since companies are shedding cheaper junior roles while adding expensive AI specialists. So there are layoffs being disguised as AI-driven efficiency when in reality they’re cost corrections. The two things are happening simultaneously, and conflating them leads to bad policy decisions both inside companies and in the regulatory response.”
Kelsey Szamet, Partner, Kingsley Szamet Employment Lawyers
“From my perspective representing employees, I see no change in the employer’s legal responsibilities whatsoever. What I see is a change in how those decisions are being made and, in some instances, how they’re being rationalised.
“The risk of using AI in these decisions is that there is a risk of amplifying a bad decision on a larger scale. If there is a flawed process for evaluating employees or determining which roles are eliminated, there is a risk of a larger number of people being negatively impacted — and the employer is still legally accountable for every one of those decisions, regardless of whether a human or a machine was part of the process.
“For those in fintech: if you cannot explain and justify why certain employees were chosen for termination, there is a problem. For those entering the next generation of financial services: there are still opportunities, but they are less certain. Employees should be paying closer attention to how they are being measured and be willing to ask questions if those measurements don’t make sense.”
