Writer's 2026 enterprise AI survey shows 60% of companies planning layoffs for employees who cannot or will not adopt AI, while 92% are cultivating a class of AI super-users. The same survey shows why this is the wrong play. The 9-hour productivity delta is a training delta, not an identity delta. Layoffs reduce headcount without changing the underlying proficiency distribution. The path that compounds is internal development inside an existing workforce.
The data, in one block
Writer.com and Workplace Intelligence published their 2026 enterprise AI survey on April 7. 2,400 respondents, 1,200 C-suite executives and 1,200 employees, across the US, UK, Ireland, Benelux, France, and Germany. Companies from 100 to 10,000+ employees, roughly 30 industries. Field period December 17, 2025 to January 25, 2026.
Six numbers from that report should be sitting in front of every CEO right now.
- 92% of C-suite executives said they are actively cultivating a new class of "AI elite" employees.
- 87% said those AI super-users are at least 5x more productive than employees who are not embracing AI.
- Super-users save about 9 hours per week. Laggards save about 2.
- 60% of executives plan to lay off employees who cannot or will not use AI.
- 77% said employees who refuse to become AI-proficient will not be considered for promotions or leadership roles.
- 48% feel that AI adoption at their company has been "a massive disappointment."
The first five numbers point one direction. The sixth tells you that direction is not working.
Hours per week saved by AI super-users versus laggards inside the same companies. The C-suite is reading that delta as an identity difference between two kinds of employee. The data argues a different read. The super-user is the laggard with a measurement instrument, a development arc, and a manager who held them to the gain.
Source: Writer Enterprise AI Adoption Survey, April 7, 2026.The 9-hour delta is a training delta
The instinct most C-suites are running on right now is that the AI super-user is a different kind of person from the laggard. Younger. More technical. More comfortable with the language. The 92% identity-cultivation number tells you the cultural posture that is forming around that read.
The data points toward a different read. Writer's analysis frames the difference as structural. Companies need measurable outcomes, governance, business-team ownership, and change leadership to move workers along the AI proficiency curve. That makes the 9-hour delta look less like a fixed personality trait and more like a training shortfall. The interpretation here is mine. The underlying data is Writer's.
Inside one company, two employees with the same job title can sit on opposite sides of the 9-vs-2 delta after twelve months. The variable is environmental, not personal. The companies that gave both employees a measurement framework, a development arc, and a manager who held the cadence produced two super-users. The companies that gave neither produced two laggards.
Companies that did the development work get the super-users. Companies that did not are now staring at a productivity split inside their own workforce and concluding that the laggards are a personnel problem they can solve by replacing them.
That conclusion is what produces the 60% layoff number. It also helps explain the 48% disappointment number that already shows up in the same survey, up from 34% the year before.
Layoffs reduce capacity, not the proficiency curve
Here is the part most CEOs running the AI elite play have not done the math on yet.
Lay off the bottom 40% of your workforce by AI proficiency. You now have 60% of your previous headcount. The proficiency distribution inside that remaining 60% is unchanged. You have more L4s as a percentage. You also have fewer of them in absolute terms, because the ones who actually got there did so on top of institutional knowledge that goes back five, ten, twenty years inside your operation.
The L1 worker you laid off was the one who knew which supplier ships late in Q3. The L2 in the warehouse was the only person on the floor who could talk a frustrated customer down. The L1 in claims processing has handled every variation of one specific edge case for fifteen years.
You do not replace those people quickly on the open market in 2026. The hiring door is not closed, but it is narrow, expensive, and slow. One 2026 industry analysis estimates roughly 1.6 million open AI roles globally against about 518,000 qualified candidates, with time-to-fill on critical AI roles running into months. The exact numbers vary by role and market. The direction is consistent across the public estimates: anything you cut, you may not be able to replace quickly. You lose institutional memory, you lose customer relationships, and the proficiency distribution of what remains is the same distribution you started with at smaller scale.
That is capacity reduction, not capability creation. Writer's 48% disappointment number is what happens when you confuse one for the other.
May Habib said it on the record
The CEO of Writer, May Habib, gave the press release its load-bearing line.
"Layoffs are not a viable AI strategy."
She frames the right approach as human-agent collaboration and organizational redesign. The companies clearing real ROI in her data are the ones investing in the development of their existing workforce, not stratifying it into elites and casualties.
The headline numbers in Writer's report carry the room. The CEO's own read on what to do about them is quiet, and it is the part most coverage is going to skip.
What the right play actually requires
Three things. None are optional.
One. A measurement framework. You cannot move what you cannot place. Most companies talk about "AI training" without an instrument that puts a worker on a defined scale. The 7 Levels of AI Proficiency is the instrument we built for this. Seven defined stages from first exposure through full operational integration. Free self-assessment at assess.launchready.ai.
Two. A development arc. A defined timeline that takes an L1 worker to an L3 worker. 6 to 12 weeks. Not "go take a course." Not "watch this LinkedIn Learning module." A pre-measurement, a structured development sequence, and a post-measurement that shows movement. The companies producing super-users are running this loop. The companies producing laggards are not.
Three. A management system that holds the gain. Without operating cadence, training fades inside a quarter. The L3 you produced in week 12 is back to L2 by week 24 if no manager ever asks them to demonstrate the new behavior. This is the piece most workforce-development programs leave out and the reason most of them do not stick.
None of those three pieces is exotic. They are the same playbook elite manufacturing operations have used for decades on safety, quality, and lean. AI proficiency is the same kind of muscle. It responds to the same kind of training.
The CEO who reads this right
The Writer survey is going to be cited as cover for layoffs by every consulting firm running the "AI transformation" playbook. The number that will be in every deck is 60%.
The number you should hold against it is 48%. That is the share of executives who are sitting inside the AI elite play right now and calling it a massive disappointment. Read the report cover to cover and the picture is not "the brave 92% versus the laggard 8%." It is "everyone is doing the same thing and most of them think it is failing."
The CEO who reads this right does not start with who to lay off. They start with where their workforce sits on a measurable scale, what arc moves the median up two levels in 90 days, and which manager owns the cadence that holds it.
That is a different operating instinct than the one Writer documented. It is also the only one the data supports.
Related reading: The 7 Levels of AI Proficiency and The 80% Shortfall: What Companies Should Learn From Higher Ed.
Frequently Asked Questions
What is the Writer Enterprise AI survey?
Writer.com, in partnership with Workplace Intelligence, surveyed 2,400 respondents (1,200 C-suite executives and 1,200 employees) across the US, UK, Ireland, Benelux, France, and Germany between December 17, 2025 and January 25, 2026. Companies ranged from 100 to 10,000+ employees across about 30 industries. The survey was published April 7, 2026.
What did the survey find about AI layoffs?
60% of executives plan to lay off employees who cannot or will not adopt AI. 92% of C-suite leaders said they are actively cultivating a new class of "AI elite" employees. 77% said employees who refuse to become AI-proficient will not be considered for promotions or leadership roles.
Why is the AI elite plus layoff strategy the wrong play?
The Writer survey shows AI super-users save about 9 hours per week while laggards save about 2 hours. That delta is a training delta. The super-user is the laggard with measurement, a development arc, and managerial follow-through applied. Layoffs reduce headcount without changing the underlying proficiency distribution. Companies running the layoff strategy are skipping the development step and calling the result an identity problem.
What does the right play look like?
Three pieces. A measurement framework that places workers on a defined skill scale so movement can be shown. A development arc that takes a Level 1 worker to a Level 3 worker on a 6-to-12 week timeline. A management system that holds the gain so training does not fade inside a quarter. The 7 Levels of AI Proficiency framework gives the measurement instrument; the 7 Levels Engagement gives the arc. You can take a free self-assessment at assess.launchready.ai.
Find your AI Proficiency level
The free 7 Levels assessment places you across seven stages of AI capability. Under ten minutes. Research-backed scoring.