Something important is happening to professional careers in 2026, and it is happening in two directions simultaneously — from the top down, and from the inside out.
The first force is institutional. Companies under financial pressure are discovering that rigid return-to-office mandates produce a convenient side effect. When you require everyone to commute five days a week to an office they moved away from, a predictable number of people will leave voluntarily. They won't require severance packages. They won't trigger regulatory disclosure requirements. They'll just go.
The second force is technological. The tools that once required months of specialist training — building a polished presentation, synthesising a research brief, drafting a client proposal — can now be assembled in minutes by anyone with a subscription and a clear thought.
Together, these forces are producing what might be called the Big Churn: a structural recalibration of what professional work looks like and what it actually rewards. It is happening fast enough that most organisations haven't caught up with it yet, and quietly enough that most individuals haven't either.
The Mathematics Behind the Mandate
Return-to-office policy is rarely only about the office.
The financial logic is straightforward. A traditional reduction in force carries real costs — severance, legal exposure, and the reputational weight of a public headcount announcement. Voluntary attrition, by contrast, is clean. The company achieves the same balance sheet outcome while the narrative stays intact. The employee absorbs the friction of the decision.
The research on what actually happens is consistent. Companies with strict five-day mandates are seeing 13% higher turnover than those offering flexibility. 53% of remote-capable workers say they would start looking for a new role if forced back full-time. And in a period when Amazon cut approximately 30,000 corporate roles between October 2025 and early 2026 — explicitly linking the reductions to AI handling more routine work — and Dell shed over 25,000 positions across two years as part of a combined RTO and AI restructuring programme, the direction of travel is clear enough.
None of this is a conspiracy. It is the normal operation of institutional incentives. Organisations do what produces acceptable outcomes with the least friction. The friction here is borne by the individuals making career decisions with incomplete information.
The Big Churn — Authentic Intelligence Briefings, May 2026The Office, Productivity, and a Long-Running Tension
The case for five-day physical attendance has always rested on something genuinely valuable but difficult to measure: the compound effect of unplanned human interaction. The difficulty is that this value is real but unevenly distributed across roles — and most mandates apply to everyone.
The research on RTO and firm performance is fairly consistent. Studies of S&P 500 companies found that mandatory RTO policies did not improve measurable performance — whether revenue or stock price — but did significantly increase turnover among senior and high-skilled employees: the people most costly to replace and most likely to have options elsewhere.
What shared physical environments do well: onboarding, relationship-building across teams, high-stakes collaborative work, the trust that accumulates through informal proximity. What a uniform five-day policy does less well: applying the same friction to work that involves none of those things, at a cost that falls hardest on caregivers, parents, and people who relocated during the years when flexibility was the explicit agreement.
The result, in many organisations, is a dynamic most people recognise. People are present, but not necessarily more productive. They build documentation and run alignment meetings and participate in coordination processes that exist partly to make work visible — because in the absence of clear output measurement, presence becomes its own proxy for contribution.
This is not a failure of individuals. It is what happens in any system when the measurement drifts from what it is supposed to measure. People respond to what is actually being measured.
What AI Is Actually Changing
The WEF Future of Jobs Report 2025 projects that 92 million positions will be displaced by automation by 2030, alongside 170 million new roles being created — a net positive, but one distributed unevenly across sectors, geographies, and career stages.
The displacement is not landing where most people initially assumed. It is not primarily hitting manual trades or service work. It is concentrated in the intermediate layer of knowledge work: clerical operations, administrative support, data processing, document synthesis, first-draft generation — roles that expanded significantly across the last twenty years as organisations grew their process infrastructure around human specialists doing repetitive cognitive tasks.
In practical terms, Cornell University research shows that companies adopting AI have already reduced junior hiring by approximately 13%. The Anthropic CEO projected in 2025 that roughly half of entry-level white-collar positions could be eliminated within five years. Amazon has publicly stated its expectation that generative AI and agents will handle more of what its corporate workforce currently does. The language of these announcements is notably flat — not triumphant, not apologetic, just operational.
A Shift in What Gets Rewarded
Both forces — RTO as friction-based restructuring, and AI eliminating the intermediate layer — are converging on the same outcome: a professional landscape that is sorting more sharply by what people can verifiably deliver in conditions that require real judgement.
In a slower-moving, more process-oriented era, professional careers were sustained by a combination of genuine expertise, institutional knowledge, and the ability to operate fluently within complex multi-stakeholder environments. Visibility mattered. Tenure mattered. The ability to navigate a specific corporate culture and its unwritten rules was itself a legitimate form of capital.
That model is not disappearing entirely. But it is under compression. The organisations that are growing are growing by doing more with smaller, more capable teams. The competitive advantage is shifting toward people who can navigate genuine complexity, build and ship real things, form trust with other humans, and make decisions when the information is incomplete and the stakes are real.
The careers that hold up tend to share a few characteristics: deep domain knowledge that cannot be assembled from public data alone, the ability to execute and be accountable for real outcomes, and the capacity to do the relational and collaborative work that requires actual human presence — not performative presence, but the kind that builds something.
These are not new virtues. They have always mattered. What has changed is that the scaffolding that once allowed people to build productive careers without them — elaborate process structures, credential accumulation as a substitute for demonstrated capability, coordination work as an end in itself — is losing its protective value faster than most organisations have acknowledged.
The Bigger Picture
The Big Churn is not a crisis in the dramatic sense. It is a correction — of a system that, for a long time, rewarded the appearance of work as generously as it rewarded work itself.
But the consequences of this correction do not stop at the individual or even the organisational level. When millions of white-collar positions are eliminated, restructured, or moved offshore to AI-adjacent processes, the payroll tax base that governments have relied on for decades starts to erode in ways that are structurally different from previous economic cycles. The revenue does not disappear — corporate earnings are rising — but it concentrates in ways that existing tax architectures were not designed to capture.
That is the fiscal dimension of the Big Churn. It is the part that governments have not yet priced in, and that deserves its own accounting.
What is your organisation actually measuring when it tracks who is in the office? And what would change if it measured outcomes instead?
The Interactive Data Report
Every number cited in this article is sourced and verified in the interactive report — including country-by-country sovereign exposure analysis, a fiscal simulation laboratory, and a full data provenance trail.