Jack Dorsey cuts 40% of Block's workforce. 4,000 people gone in a single announcement. The stock rises 24% the same day. He calls it a structural shift, not a crisis response. Oracle follows in late March 2026, cutting thousands more, framing it as broader organisational changes. Multiple companies, mass layoffs, and carefully worded explanations.

Before assuming AI is the villain behind every headline, it is worth asking a harder question: what is the data actually saying, and what is happening beneath these announcements?

This post breaks down the real numbers, the pattern underneath them, and what the shift means for the people it affects most freelancers navigating a changing market and founders deciding who to hire and what to build.

Also read: Anthropic's Project Glasswing: How AI Is Entering Cybersecurity and What It Means for Businesses

The Numbers Behind the Headlines

The scale of what is happening

The displacement is real. According to Challenger and Gray, 55,000 job cuts were directly attributed to AI in 2025, more than 12 times higher than just two years earlier. Entry-level job postings have fallen 15% year on year. The roles disappearing are concentrated in specific categories: coordination, first-level support, data processing, and administrative functions that follow repeatable rules.

Block cut headcount from 10,000 to under 6,000. Meta restructured recruiting, sales, and Reality Labs. IBM replaced hundreds of HR roles with AI chatbots while continuing to hire in technical areas. The pattern across these examples is the same: roles that move information between humans are being absorbed into software. Roles that require deciding what should happen are not.

The claims that AI is the sole driver should be taken carefully. Oxford Internet Institute researcher Fabian Stephany has documented that many companies are using AI as the public justification for corrections that were always inevitable. The over-hiring of 2020 to 2022 inflated tech headcounts at a scale that was never sustainable. Some of what is called AI displacement would have happened regardless of where model capabilities stood.

What is growing at the same time

In the same period that saw 55,000 AI-attributed cuts, Deloitte announced 50,000 new AI hires. OpenAI expanded its workforce toward 8,000 employees. Job descriptions mentioning AI skills have increased 400% over two years.

These two sets of numbers exist in the same economy at the same time. The story is not that jobs are disappearing. It is that specific types of jobs are disappearing while other types are being created rapidly. The contradiction in the headlines is real, but it resolves when you look at which categories are moving in which direction.

The Real Pattern Underneath the Cuts

The shift is not primarily about which companies are cutting. It is about which type of work is being absorbed into software.

McKinsey's analysis of 800 occupations identifies the key variable: vulnerability depends on whether a role is cognitive and routine, physical, or relational. Cognitive and routine work, the kind that involves applying consistent rules to varied inputs, is in the affected category. Work that involves deciding what the rules should be, or applying judgment that a system cannot replicate reliably, is not.

The pattern concentrates in work that is digitally mediated and rules-based: coordinating information between systems, summarising and reporting, routing tasks, handling first-level decisions that follow predictable logic. These functions are being absorbed into software at a quality level that was not achievable in 2022.

The shift from execution to orchestration

The change is best understood as a shift from doing the work to managing the systems that do the work. Previously, a person performed the task directly. Now the person manages the agent that performs the task and is responsible for the quality of its output.

This shift has a measurable economic consequence. PwC's Global AI Jobs Barometer finds that workers with advanced AI skills earn 56% more than peers in the same roles without those skills. The gap between someone who operates a system and someone who designs and manages it has become visible in compensation data. That is a relatively new development and it is accelerating.

Why This Is Happening Now

AI has crossed a genuine capability threshold

The primary structural cause is that AI has crossed a functional threshold for specific task categories. Summarising, routing, scheduling, first-level support, and structured output generation now work at production quality. This is not hype. It is a documented capability shift that happened primarily between 2022 and 2024, and the displacement it causes in those specific categories is permanent. Those roles will not return to humans.

The over-hiring correction

Many companies hired aggressively in 2020 to 2022 at a scale that was always unsustainable. Some of what is labelled AI displacement is a financial correction that would have happened regardless. This matters because conflating the two inflates how much AI is actually responsible and makes the picture harder to interpret clearly for anyone making decisions based on it.

Investor behaviour has structurally shifted

The market is now rewarding demonstrated leverage: better output from leaner teams. Block's stock rising 24% after a 40% workforce reduction is not an anomaly. It is a signal about what investors now value. Gartner predicts that by end of 2026, 20% of organisations will use AI to flatten their structures and eliminate more than half of current middle management positions. This incentive structure has changed permanently, and it will continue to influence hiring decisions across every public company.

What This Means for Freelancers

What the demand shift looks like

Entry-level postings are down 15% year on year. The roles shrinking fastest are task-execution roles: processing, coordinating, and passing work along without owning the outcome. The roles growing require a different posture: designing the system, briefing the AI tool, managing the output, and being accountable for the result.

BeGig sees this shift in real time across both sides of the market. On the supply side, average experience in the BeGig freelancer pool has risen from 2.3 years in 2023 to 4.1 years today. Senior engineers with 5 to 12 years at Google, Meta, and Amazon are joining at 3 times the rate of two years ago. They are not freelancing as a gap fill. They are choosing it as a deliberate structure because the market for senior, outcome-focused technical talent has strengthened significantly.

On the demand side, the gigs being posted have changed. Clients are not asking for task executors. They are asking for people who can scope a problem, build a workflow around it, and own the result.

What freelancers should actually do

PwC's data makes the most actionable point clearly: AI-fluent workers in the same roles earn 56% more. This is not about becoming an AI researcher. It is about understanding how to work alongside systems rather than waiting to be assigned tasks by them.

For a designer, this might mean integrating AI-assisted feedback loops into the creative review process. For a developer, it means being able to scope and build workflow automation, not just write application code. For a data analyst, it means operating the pipeline that produces the output, not just producing it manually.

The window for this reskilling is open. The 15% decline in entry-level postings suggests it is narrowing. A freelancer repositioning toward workflow design, AI system operation, or automation building is on the right side of the shift. One waiting for the market to stabilize is not.

What This Means for Operators and Founders

The risk most founders are not seeing right now

Most founders are making hiring decisions based on what their team needs today. Deloitte's own research shows that only 11% of organisations are actively using agentic AI in production right now. 35% have no formal strategy at all. That means the majority of founders are building team structures for a business environment that will look significantly different in 24 months, without a plan for how the team changes with it.

The risk is not moving too slowly on AI adoption. The risk is hiring for roles that quietly erode under you while you are focused on the product. A coordination role hired today may not exist in its current form in two years. A role that requires judgment, system design, or outcome ownership will not only survive but become more valuable.

The question that determines how you build

The hiring question has changed structurally. It is no longer how many people do I need. It is which decisions in this business genuinely require a human, and what can be structured into a reliable system that handles the rest.

At BeGig Studio, the first question in every scoping call is not how many roles can this replace. It is where does the judgment actually live in this workflow. That question determines the system boundary, which determines what gets built, what stays human, and what the team looks like in two years. Founders who have answered that question are building fundamentally different cost structures. Those who have not are still hiring the way they did in 2022.

Conclusion

The layoff narrative of 2026 is partially true and significantly inflated. What is genuinely happening is a repricing of work. Tasks that never required a human decision are being absorbed into systems. What remains becomes more valuable, not less.

The shift is not AI versus humans. It is whether you understand your work well enough to design it rather than just perform it. For freelancers, that means moving from doing to designing. For founders, it means building for the team you will need in two years, not the one that made sense in 2022.

The window for that repositioning is open. The data suggests it is narrowing.

Frequently Asked Questions

Are tech companies really cutting jobs because of AI in 2026?

Partially. AI is genuinely responsible for displacing roles that involve repetitive, rules-based cognitive work such as coordination, summarising, and first-level decision-making. However, the 2020 to 2022 over-hiring correction is also a significant driver, and some companies are using AI as the public framing for cuts that were financially inevitable regardless. Challenger and Gray attribute 55,000 job cuts directly to AI in 2025, but researchers like Oxford's Fabian Stephany note that conflating both causes overstates AI's actual responsibility.

Which types of roles are most at risk from AI automation right now?

Roles that are cognitive, routine, and digitally mediated carry the highest displacement risk. This includes data entry and processing, first-level customer support, basic content generation, scheduling and coordination, and administrative functions that follow predictable rules. McKinsey's analysis of 800 occupations identifies these as the most vulnerable categories. Roles requiring judgment, relationship management, creative decision-making, or system design are in the lower-risk category and in many cases are seeing increased demand.

Is AI creating more jobs than it is eliminating overall?

The World Economic Forum's Future of Jobs Report 2025 projects 92 million jobs displaced by 2030 and 170 million new roles created, a net gain of 78 million. However, this net positive conceals a significant problem: the jobs being created and the jobs being destroyed require different skills and are not accessible to the same people without significant reskilling. The headline number is less important than the skill gap it obscures.

What should freelancers do specifically to stay relevant in this shift?

PwC data shows AI-fluent workers in the same roles earn 56% more than peers without AI skills. The practical move is to identify where AI can augment your existing expertise and build that capability deliberately. For developers, this means workflow automation and system orchestration. For designers, AI-integrated creative workflows. For analysts, pipeline operation and output management. The goal is not to become an AI specialist but to be the person who understands your domain well enough to design and operate the system that does the work.

How should founders think about hiring and team structure differently because of AI?

The question has shifted from how many people to which decisions genuinely require a human. Deloitte research shows only 11% of organisations are actively using agentic AI in production, meaning most founders are making structural decisions without a clear picture of what their team needs to look like in 24 months. The practical starting point is mapping where judgment actually lives in your key workflows. What requires a human decision? What can be structured into a reliable system? Those answers determine what you hire for and what you build.