ExpertsFMWhy half of product managers are in trouble | Nikhyl Singhal (Meta, Google)
Lenny's Podcast: Product | Career | Growth · April 19, 2026
5 Selected Lessons:
prediction · 1 framework · 1 reasoning_pattern · 1 heuristic · 1 priority · 1
What to prioritize to thrive in the next two years, per Nikhyl
Source
Lenny's Podcast: Product | Career | Growth — Why half of product managers are in trouble | Nikhyl Singhal (Meta, Google) (April 19, 2026) · at 1:03:37 ▸ · at 1:07:03 ▸
When this applies
when anyone facing an industry-wide disruption wants an ordered action plan rather than abstract reassurance
The takeaway
In a period of disruption, sequence self-motivation (joy) before capability-building (obsolete tasks) before intensity (pace) before status sacrifice (ego) before strategic patience (long-term focus).
Why it works
Each priority builds on the last: finding joy sustains motivation, obsoleting low-value tasks frees capacity, increasing pace matches the new competitive tempo, swallowing ego enables access to current-skill-building roles, and long-term focus ('skip' thinking) ensures the sacrifices pay off later.
One level up
Durable adaptation to disruption requires an ordered stack of psychological, behavioral, and strategic shifts — not a single tactic.
Where it stops holding
This stops working when external constraints (health, caregiving, financial need) make it impossible to find the 'reserve' of time/energy this plan assumes.
Detail
hierarchy:
- 1. Find your moment of joy with new AI tools
- 2. Adopt the engineering mindset: obsolete your own low-value recurring tasks
- 3. Increase your pace and find reserve energy ('fire in the belly')
- 4. Swallow your ego — be willing to take smaller roles to stay current
- 5. Stay long-term focused ('skip' thinking) rather than optimizing only the next move
rationale: This is not a job. This is the next two years requires a lot of fire in the belly, a lot of agency.
In their words
You've got to find that reserve. You've got to make time. You've got to maybe disappoint others in a way that you haven't in the past in order to create time for you to stay current, find that joy, obsolete yourself from the things that are not worth it.
Judgment as the core remaining product skill
Source
Lenny's Podcast: Product | Career | Growth — Why half of product managers are in trouble | Nikhyl Singhal (Meta, Google) (April 19, 2026) · at 11:49 ▸ · at 30:54 ▸
When this applies
when automation collapses execution cost in any field, forcing a shift from 'doing the work' to 'deciding what work is worth doing'
The takeaway
When execution becomes cheap and abundant, competitive advantage shifts to evaluation and selection, not production.
Why it works
As the cost of generating and testing changes drops 10-100x, the bottleneck moves from production capacity to evaluating which changes are good, sustainable, and differentiated — a skill that cannot be automated as easily as execution.
One level up
Scarcity moves up the value chain as automation eats the layer below it; stay ahead by occupying the next scarce layer.
Where it stops holding
This stops working when the volume of changes is still low enough that execution remains the bottleneck, or when judgment itself becomes automatable by more advanced models.
Detail
procedure:
- Identify which product tasks are mechanical/repeatable (candidates for obsoleting via software/agents)
- Build or adopt tools to automate those mechanical tasks
- Redirect freed time to evaluating whether a proposed change is good/bad, sustainable, and differentiated
- Evaluate post-hoc whether the change met its success criteria
benchmarks: {"anticipated_volume_increase":"10 to 100 times more product changes presented than before"}
rationale: Judgment, not production, becomes the 'tip of the spear' skill once AI commoditizes execution.
In their words
So I think on the company side, I think that, and then maybe I'll slant it more towards product for a second. So, product leaders will increasingly get paid and be asked to drive judgment and then be the tip of the sphere on trying to essentially obsolete everything else through software, through AI, through agents.
Evaluating a teammate's leverage claim: shipping vs. systems work
Source
Lenny's Podcast: Product | Career | Growth — Why half of product managers are in trouble | Nikhyl Singhal (Meta, Google) (April 19, 2026) · at 30:54 ▸
When this applies
when a coordinator-type role in any fast-executing team is tempted to 'do the work' directly rather than build the systems that make the whole team more effective
The takeaway
True leverage in a 'hands-on' role comes from building systems that scale everyone's output, not from personally joining the production line.
Why it works
If a PM joins a 50-person engineering team as an additional shipper, they add marginal, low-differentiated output ('a cheap knockoff of an engineer'); but if they build the tools/systems that let the 50 engineers be tracked, coordinated, and evaluated for judgment, they multiply the whole team's output instead of adding to it linearly.
One level up
Ask of any 'should I get hands-on' instinct: am I adding to a queue of similar workers, or am I building the infrastructure that makes the queue work better?
Where it stops holding
This stops working when the team is small enough that direct individual contribution genuinely outweighs the overhead of building coordination tooling.
Detail
approach: Distinguish 'being the 51st engineer' (low leverage) from building internal tools to manage/judge the output of the other 50 (high leverage).
signals:
- team already has sufficient raw execution capacity
- coordination/judgment overhead (status reports, backlogs, standups) is the actual bottleneck
rationale: CPOs observed in Nikhyl's community are building efficiency tools (automated product reviews/standups), not personally shipping more features.
In their words
If on the other hand, the thing that you're building is ways to stay on top of what the 50 people are doing
AI race won't self-regulate (game theory of pace)
Source
Lenny's Podcast: Product | Career | Growth — Why half of product managers are in trouble | Nikhyl Singhal (Meta, Google) (April 19, 2026) · at 46:42 ▸
When this applies
when multiple competing actors each privately wish a collective race would slow but none can unilaterally do so without ceding advantage to rivals
The takeaway
In a competitive race with no enforcement mechanism, even unanimous private desire to decelerate will not change collective behavior.
Why it works
Because any single actor who slows down cedes competitive advantage to rivals who don't, even universal private preference for slowing down cannot produce an actual slowdown absent enforceable coordination.
One level up
Don't plan around a hoped-for collective de-escalation in any competitive dynamic lacking a coordination mechanism — plan around the race continuing.
Where it stops holding
This stops working when an external enforcement mechanism (regulation, binding agreement) actually constrains all competitors simultaneously.
Detail
rule: Even if every competitor agrees in principle to slow down, none will do so unilaterally because whoever doesn't slow down gains a large advantage.
threshold: Applies to any race dynamic where defection yields outsized gains and no binding coordination exists.
rationale: Used to justify the advice that individuals should adapt now rather than wait for the pace to ease.
In their words
They all said, yep. Which says a lot about just like they're all like freaked out a bit about how fast everything's going, how much is changing. And then there's the reality of, but that is not going to happen because the game theory of it all is it doesn't work.
Mass shedding and AI-first rehiring cycle
Source
Lenny's Podcast: Product | Career | Growth — Why half of product managers are in trouble | Nikhyl Singhal (Meta, Google) (April 19, 2026) · at 19:12 ▸
When this applies
when any organization realizes it overhired relative to output and a new capability (technology, automation) makes smaller, differently-skilled teams viable
The takeaway
Expect organizations to cut broadly then rehire narrowly for a new skill profile, rather than simply shrinking — the net change is a composition shift, not just a size shift.
Why it works
Companies that doubled headcount during a growth era didn't get proportional output, and AI tooling now lets a much smaller, differently-skilled team deliver equivalent or greater output, prompting simultaneous layoffs and new hiring.
One level up
Headcount grows during boom times far beyond what's operationally required; technological step-changes trigger a 'judgment day' reckoning of that excess.
Where it stops holding
This stops working when the overhiring premise is false (the company's growth was headcount-efficient) or when the new technology doesn't yet substitute for the cut roles.
Detail
approach: Estimate true operational need (e.g., ~9% of staff to 'keep the lights on') vs actual headcount; predict large-scale layoffs paired with smaller AI-first rehiring.
signals:
- rapid headcount doubling over 5 years without proportional output gains
- new tooling that changes required skill profile
- public companies under investor scrutiny
rationale: Companies hire for optionality/expansion during good times, creating fat that gets cut once judgment and efficiency tools reduce the need for scale-via-headcount.
In their words
You might see a company shed 30,000 and hire 8,000. But the 8,000 people they're going to hire are going to all be AI first.