Lenny Rachitsky’s How Tech Workers Are Feeling in 2026 landed this week with a finding that deserves more attention than most of its coverage gave it. The workforce is not just stressed or optimistic — it is splitting into two distinct populations. One half describes feeling amplified, more capable than at any prior point in their career. The other uses words like chaos, layoffs, and dystopia. Burnout climbed from 44.7% to 55.7% year over year. Career optimism declined. Those two trends coexisting is not a contradiction — it is the shape of a workforce in active bifurcation.
This is Rachitsky’s second large-scale annual survey, which makes the bifurcation signal more credible: this is not noise in a single data set, it is a direction. The amplified cohort is growing more capable. The other is growing more burned out. The gap between them is widening.
Lenny’s framing is primarily a product and PM story, and it is accurate. But there is a technology leadership angle the data surfaces that the survey doesn’t fully develop: this split is happening inside individual organizations right now, and leaders who don’t actively manage the diverging employee experience will end up with a specific, predictable outcome — their AI-amplified engineers leave for better scope, and their AI-resistant engineers become a structural drag on adoption.
sequenceDiagram participant A as AI-Amplified Engineer participant E as Org Event participant R as AI-Resistant Engineer Note over E: AI tools adopted across team A->>E: Output climbs; scope narrows R->>E: Output gap opens; avoids tools Note over E: Headcount pressure; reviews begin A->>E: Options appear; evaluating alternatives R->>E: Anxiety rises; performance scrutinized Note over E: No active leadership response A-->>E: Leaves for role with better scope R-->>E: Stays; AI adoption stalls
For the working software engineer
If you are on the amplified side of this divide, the survey’s finding is useful context for what you are probably already feeling: the output gap between you and your teammates is real and widening. That is not comfortable. Engineers who are used to being part of a collaborative team are now outproducing it — and in many cases outpacing their manager’s ability to evaluate or review their work. The organizational challenge, not the technical challenge, is usually the binding constraint at this point. You need clearer ownership boundaries, faster decision loops, and leadership that understands what you are actually doing. If you are not getting that, you already know why recruiters are calling.
If you are on the resistant side, the honest framing is this: the gap is measurable, and organizations under financial pressure are measuring it. The window to cross it is not unlimited. The engineers who come through this transition intact are generally the ones who got structural support — someone who helped them engage with the tools on a real project with real stakes, not a training program running in parallel to actual work. If your organization is not offering that, ask for it directly.
For business owners and operators
The survey describes a workforce-wide phenomenon. The managerial version of this problem is smaller and more tractable: the ten to twenty engineers on your team right now are probably not all in the same place on the AI adoption curve. Some are amplified. Some are working around the tools. Some are quietly struggling with the gap.
The failure mode most organizations are living is passive management — letting the amplified engineers do their thing while hoping the rest catch up. That approach loses the amplified engineers first. They are the ones with options, and they will take a role with clearer scope and better tooling the moment one materializes. What remains is a team that is slower to adopt and increasingly stressed about a gap that is only growing.
Active management is not complicated, but it requires an explicit decision to treat this as a leadership problem rather than an HR problem. Name the bifurcation. Design different paths for the two cohorts. Make sure the AI-resistant engineers have structured exposure on real work — not a standalone training program that runs alongside the actual work and nobody finishes. The amplified engineers need bigger problems and faster decision authority. The resistant engineers need scaffolding, not supervision.
Budget the management time honestly. Moving a bifurcated workforce back toward a common capability level is a leadership project with a six-to-twelve month horizon. It does not resolve itself, and a one-time training investment will not move it.
My take
I led more than 30 developers across multiple teams at LERETA over five years — a modernization program that grew to a $20M board commitment and included the integration of a technology acquisition. What made that engagement work was a decision made early and sustained throughout: we did not try to pull the whole team at the same speed.
The teams had genuinely different appetites for the direction we were heading. Some engineers were driving the technical modernization. Others were slower, more anchored in the existing stack, more cautious about the new architecture’s durability. Managing the gap between them required something more deliberate than setting the direction and letting people sort themselves out.
What worked was giving the engineers who were leading the technical direction real ownership — not just more tickets, but clear technical authority on their area and a path for their work to matter at the level the board was watching. For the engineers who were behind, we built on-the-job paths: real projects, real accountability, working alongside someone who could help them understand the new direction in the context of actual deliverables. Not an abstract training track.
When LERETA also went through an acquisition and I helped integrate the acquired company’s developers into the existing team, the same design principle applied. You cannot manage a cultural gap at the team level by announcing a shared direction and walking away. You design the integration — who works with whom, what the first shared project is, where the new engineers get early wins. It requires explicit thought, not optimism.
Lenny’s survey surfaces a real split. What it doesn’t say is that the split is structurally manageable if you treat it as a leadership design problem. The organizations that do that work will hold their amplified engineers and bring the resistant cohort across. The ones that don’t will lose the best people first.
