Half of product managers are at risk. The question is which half. Nikhyl Singhal, founder of The Skip and former VP at Meta and Google, lays out exactly why the PM role is fracturing - and what separates the people who will thrive from those who won't. This isn't another "AI will automate your job" hot take. It's a precise diagnosis of a profession at a crossroads.

In this episode with Lenny Rachitsky, Singhal introduces frameworks that cut through the noise: the builder vs information-mover divide, the psychological barriers preventing reinvention, the concept of "smiling exhaustion" spreading through the PM community, and his prediction that companies will shed 30,000 roles and rehire 8,000 - all AI-first. He also explains what judgment actually means for PMs today, and why your company logo is becoming the least useful thing on your CV.

🎥 Watch the full episode here:
📆 Published: 19 April 2026
🕒 Estimated Reading Time: 9 mins. Time saved: 60+ mins! 🔥

Key insights from the full article:

Not feeling this one? Try these instead:**

Singhal's central claim is blunt: product management is splitting into two roles, and only one of them has a secure future.

Builders are PMs who are close to the product - forming opinions on what to create, getting hands-on with prototypes, shipping things directly. Information-movers are PMs whose primary value is communication: translating between teams, managing stakeholders, running processes. In a world where AI can handle much of the coordination layer, that second role is shrinking fast.

This isn't a new tension in PM. But AI has made it existential. When a single engineer with Claude or Cursor can prototype in hours what used to require a PM-led sprint, the coordination work simply diminishes. The PMs who remain essential are those who bring something AI can't - genuine product intuition, taste, and the ability to make hard calls on what matters.

The uncomfortable truth: many PMs who consider themselves builders are actually sophisticated information-movers who happen to write good PRDs. The test isn't your job title or the quality of your documentation. It's whether the product would be meaningfully worse without your specific judgment.

Key Takeaways:

Singhal introduces a term that will resonate with anyone who's attended a PM conference in the last two years: "smiling exhaustion". It's the experience of publicly agreeing that AI changes everything while privately struggling to figure out what that means for you, personally.

The pressure is real. Senior PMs who've spent a decade building expertise in roadmapping, stakeholder alignment, and product strategy are being told those skills are worth less. Simultaneously, they're expected to be enthusiastic about AI tools, rebuild their intuition, and project confidence. The smile is for the audience. The exhaustion is for the hotel room.

This psychological dimension is, Singhal argues, the actual bottleneck. Most PMs aren't short on information about what's changing - they're short on the emotional capacity to act on it. Reinvention requires admitting that your current identity isn't sufficient, and that's genuinely hard when your current identity has served you well.

His prescription isn't motivational. It's structural: find your "moments of joy" with AI - the specific interactions where the technology actually excites rather than threatens you. Build from there. Reinvention that starts from genuine enthusiasm compounds. Reinvention that starts from fear stalls.

Key Takeaways:

Singhal's most striking prediction: tech companies will collectively shed approximately 30,000 product and adjacent roles in the near term, then rehire around 8,000 - all optimised for AI-first work. That's not a rounding error. It's a structural reduction of nearly 75%.

This isn't science fiction. The pattern is already visible in how companies are thinking about headcount. Teams that required eight PMs to manage information flow, run sprint ceremonies, write specs, and coordinate with engineering are discovering they can operate with two or three PMs who use AI tools natively. The leverage ratio has shifted dramatically.

What does "AI-first" actually mean in this context? Singhal is specific: it's not about using ChatGPT occasionally or having AI on your CV. It's about having rebuilt your intuition in an environment where AI is a genuine collaborator, not a novelty. The 8,000 roles being created aren't easier versions of the 30,000 being cut - they require demonstrably higher-leverage individuals who've done the reinvention work.

The implication for career strategy is significant. Waiting to see how this plays out is itself a choice - one that leaves you competing for a shrinking number of traditional PM roles while others are positioning for the AI-first ones.

Key Takeaways:

Singhal pushes back on the vague invocation of "judgment" as the PM skill that AI can't replace. Judgment only protects you if it's the right kind of judgment - and most PMs are optimising for the wrong kind.

The judgment that's losing value: process judgment. Knowing how to run a discovery sprint. Knowing when to write a PRD vs a brief. Knowing how to sequence a roadmap politically. These are important skills, but they're learnable by AI and increasingly baked into tools.

The judgment that's gaining value: product judgment. Knowing which problem is worth solving. Knowing what a good solution actually feels like to the user. Knowing when to kill a feature that's technically working but missing the point. Knowing when the data is right but the conclusion is wrong. This is harder to replicate because it requires genuine domain understanding, taste, and the courage to back a call that can't be fully justified with data.

For Singhal, the test of whether you have this second type of judgment is simple and uncomfortable: can you point to decisions where you were right when others were wrong, with evidence? Not decisions where you managed a process well, but decisions where your specific view on what mattered produced a better outcome. If the honest answer is no, that's the gap to close.

Key Takeaways:

For a decade, the hiring signal in PM was your employer history. A Google or Meta on your CV opened doors automatically. Singhal's argument is that this credential inflation is deflating fast, and the PMs most exposed are those who've built an identity around where they've worked rather than what they've built.

Why is this happening? Because AI has created a new signal that's harder to fake: demonstrated building capability. Can you show a prototype you shipped? Can you point to a product decision that had a measurable outcome? Can you demonstrate that your intuition works in the current environment, not just in the pre-AI playbook? These questions are increasingly what separates candidates - and they're questions that a Google logo can't answer.

This creates an uncomfortable career decision for experienced PMs at prestigious companies. The "harvest" strategy - staying in a well-paid, lower-intensity senior role and maximising compensation - is rational in the short term. But it comes with a cost: your intuition about how great products get built is calcifying around assumptions that are becoming obsolete.

Singhal's view is that taking a "step backward" to a more hands-on, lower-seniority role at an AI-native company isn't actually backward. It's tuition. You're paying with status to upgrade your mental model of how products work in the current environment. The PMs who'll be most valuable in three years are likely the ones making uncomfortable moves now.

Key Takeaways:

The middle of the PM spectrum is disappearing. Singhal's framework points to four distinct roles that will remain viable as the AI transformation plays out - each requiring a different combination of skills and tolerating a different set of trade-offs.

1. The AI-Native Builder - Close to the product, genuinely excited about building with AI tools, operating with high autonomy. Typically found at AI-native companies or technical product roles. Requires rebuilt intuition, comfort with ambiguity, and genuine product obsession.

2. The AI Transformation Lead - Senior PM at a traditional company whose job is to bring AI thinking into an existing product and organisation. This is a management and change role as much as a product role. Requires credibility, organisational navigation skills, and enough AI literacy to set direction credibly.

3. The Specialist - Deep domain expertise in a specific area (fintech, health, legal, etc.) combined with enough product craft to translate that expertise into product decisions. AI can't replicate genuine domain understanding; this is the defensible specialist path.

4. The Technical Orchestrator - PMs who've built enough technical depth to work fluidly with AI systems, agentic tools, and complex infrastructure. More technical than most PMs, but not engineers. The emerging category as AI products themselves become more complex.

The roles not on this list: the generalist PM coordinator who sits between engineering and stakeholders without a distinct building, domain, or technical edge. That role is the one at risk.

Key Takeaways:

Singhal's framework is most useful not as a prediction about the market, but as a prompt for honest self-assessment. The question isn't "will product management survive?" It's "which PM am I, and is that PM in demand in two years?"

His suggested starting point: write down the last three product decisions you made where your specific view changed the outcome. Not processes you managed well. Not teams you coordinated. Decisions - where you had a perspective, backed it, and the product was better for it. If you struggle to fill three, that's diagnostic information about where your value currently sits.

From there, the path is specific to which of the four archetypes you're closest to and which constraints you're operating under. Someone with high financial obligations and deep domain expertise takes a different path than a mid-career PM with high risk tolerance and genuine building obsession. The framework isn't prescriptive - it's a lens for making your own call more clearly.

The one thing Singhal is categorical about: waiting to see how it plays out is the worst option. The PMs who'll be well-positioned in two years are making uncomfortable bets now.

🔗 Links Referenced:

If this one resonated, these five go deeper on related themes - each covers a different angle on the same underlying question: what does it take to build great products in the AI era?

That’s a wrap.

As always, the journey doesn't end here!

Please share and let us know what you liked or want changing! 🚀👋

Alastair 🍽️.