52,050 Tech Layoffs by March 2026—But AI Jobs Growing 92%. Are We Automating or Just Reshuffling?

eng_director_luis · April 5, 2026, 5:57pm

I’m watching something deeply contradictory unfold at my company right now, and I’m curious if others are seeing the same pattern.

In the past 6 months, we’ve laid off 43 people across operations, support, and administrative roles. The justification? “AI automation and operational efficiency.” Meanwhile, we’re aggressively hiring AI/ML engineers, data scientists, and platform engineers. We currently have 15 open AI-related roles we’re struggling to fill.

According to recent data, 52,050 tech workers were laid off in Q1 2026, with over 20% of those layoffs explicitly attributed to AI and automation—up from just 8% in 2025. At the same time, AI job postings surged 92%, and roles requiring AI skills command a 56% wage premium.

The math doesn’t add up in a simple way.

The Geographic Reality

This isn’t evenly distributed. Seattle saw approximately 16,590 tech workers affected by layoffs from Amazon and Microsoft combined. San Francisco had 9,395 across multiple companies. These are real communities absorbing massive workforce disruption.

What I’m Observing on the Ground

The roles being eliminated:

Customer support specialists
Data entry and administrative staff
Junior QA testers
Some mid-level operations roles
Technical writers and documentation specialists

The roles we’re desperately trying to fill:

Machine learning engineers
AI/ML operations specialists
Data scientists with deep learning expertise
Platform engineers with AI infrastructure experience
AI product managers

These aren’t the same people. A customer support specialist with 5 years of experience can’t just “reskill” into a machine learning engineer role in 3 months. The skill gap is enormous.

The Uncomfortable Questions

1. Are we really automating, or just optimizing labor costs?

If this were pure automation, wouldn’t headcount stay flat while productivity increased? Instead, we’re seeing net headcount reduction with capability concentration in fewer, more expensive roles.

2. What happens to the career pipeline?

Entry-level and learning roles are being automated first. How do junior engineers gain experience when the traditional ladder rungs are removed? We’re hiring senior AI talent but eliminating the roles that used to develop talent.

3. Who has access to reskilling?

The World Economic Forum estimates 1.1 billion jobs will be transformed by technology this decade, and 60% of workers will need reskilling. But AI skills training requires significant time and financial investment. Workers with AI skills earn 56% more—but only if they can afford the transition period.

4. Is this workforce transformation or just cost optimization?

From a board perspective, this looks successful: Lower operating costs, higher technical capability, improved margins. But from a societal perspective? We’re concentrating opportunity among those who already have advanced technical education while eliminating accessible entry points.

The Diversity Implications

This hits me personally. I’ve spent years mentoring first-generation college students and Latino engineers through SHPE. Many of them entered tech through support, QA, or operations roles—exactly the roles being automated first.

If we automate away the accessible entry points while requiring advanced degrees for AI roles, we’re not just reshuffling the workforce. We’re closing doors.

What This Looks Like in Practice

At my company specifically:

Eliminated: 43 roles, average salary $65K, diverse workforce
Added: Targeting 15 AI roles, average salary $180K, highly competitive market requiring advanced degrees

The financial logic is clear. The human impact is messier.

The Question I Can’t Shake

By 2030, projections suggest 170 million new jobs created by AI (net gain of 78 million after subtracting eliminated roles). That sounds optimistic. But when I look at my team, I don’t see a clear path for the 43 people we let go to become the 15 AI specialists we’re trying to hire.

Are we witnessing workforce transformation—where people transition to higher-value roles—or workforce reshuffling—where we replace one group of workers with a completely different group?

Because from where I’m sitting, it feels a lot more like the latter.

What are you seeing in your organizations? Are companies investing in genuine reskilling programs, or are we just hiring a different workforce and calling it “transformation”?

cto_michelle · April 5, 2026, 5:57pm

The uncomfortable answer is: both. We’re automating AND reshuffling, and pretending it’s only the former lets us avoid the harder ethical questions.

At my company, we went through exactly this last year. We eliminated 8 customer support roles and hired 5 AI engineers to build an intelligent support system. Net headcount: down 3. Customer resolution time: down 40%. Operating costs: down 25%.

From a business metrics perspective, this was unambiguously successful. Our board celebrated it as “AI-driven operational excellence.”

But here’s what the metrics don’t capture:

The Reskilling Reality

We offered reskilling programs to the 8 support staff. Paid training, learning stipends, mentorship. Genuinely tried to create pathways.

Results after 6 months:

1 person transitioned to a data analyst role (huge success story)
2 people left tech entirely—went back to school for different careers
5 people found similar support roles at other companies (for now)

The gap between customer support and AI engineering is a chasm, not a bridge. It’s like asking a talented line cook to become a biochemist. Both are valuable, but they require fundamentally different foundations.

The ROI Pressure Is Real

Here’s the part I hate admitting: Our board explicitly pressured us to show AI ROI. Not “Are we building sustainable capabilities?” but “Can you show 20% cost reduction in 12 months?”

When you frame the question that way, the answer becomes obvious. Hire expensive AI talent, automate labor-intensive processes, show cost savings. The fact that you’re replacing one workforce with an entirely different workforce becomes a feature, not a bug.

The Market Dynamics Are Brutal

I’ve been on 15 recruiting calls in the past month trying to fill AI/ML roles. The competition is insane. Candidates with 3 years of ML experience are getting $250K+ total comp offers. Meanwhile, the support specialists we let go are competing for $55K roles.

This isn’t a “skill gap.” It’s a market bifurcation. We’re creating a two-tier tech workforce: the AI creators (high wage, high demand) and everyone else (increasingly automated, increasingly precarious).

What Responsibility Do We Have?

I genuinely don’t know. Part of me says: “Companies optimize for survival. That’s how markets work. If we don’t automate, our competitors will, and we’ll lose entirely.”

But another part says: “We’re engineering the collapse of career pathways that created the diverse tech workforce we claim to value.”

The Latino engineers Luis mentors? The first-gen college students? Many of them entered through QA, support, and operations. If we automate those entry points, where do they enter?

The 78 Million New Jobs Question

Everyone cites that stat—170 million jobs created, 92 million eliminated, net gain of 78 million. It sounds great until you realize:

Who gets eliminated: People with moderate technical skills, often from diverse backgrounds, with moderate wages.

Who gets created: Roles requiring advanced degrees, often in competitive metro areas, with high wages and narrow specialization.

These aren’t the same people. And the transition path isn’t clear.

My Uncomfortable Conclusion

We’re not transforming the workforce. We’re replacing it. The question isn’t “Are we automating or reshuffling?” It’s “Are we okay with reshuffling, and if not, what are we willing to sacrifice to prevent it?”

Because right now, the market incentives are screaming “reshuffle,” and we’re all just calling it “transformation” to sleep better at night.

vp_eng_keisha · April 5, 2026, 5:57pm

Michelle nailed the economic reality, but I want to talk about something that keeps me up at night: we’re destroying the career pipeline while celebrating innovation.

The Junior Engineer Crisis

At our EdTech startup, we scaled from 25 to 80 engineers in 18 months. Here’s what I’m seeing in hiring:

Entry-level candidates in 2023:

Customer support → junior QA → QA engineer → software engineer
Technical writing → documentation engineer → developer relations
Data entry → data analyst → data scientist

Entry-level pathways in 2026:

~~Customer support~~ (automated by AI agents)
~~Junior QA~~ (automated by AI testing tools)
~~Technical writing~~ (automated by AI documentation)
~~Data entry~~ (automated by AI data processing)

We’re hiring senior AI engineers at $200K+ but eliminating the roles that traditionally trained people how to become senior engineers.

The Diversity Impact Is Not Random

This hits different communities differently, and we need to say it explicitly.

Looking at the 52,050 Q1 2026 layoffs:

Customer support and service roles: Historically 60%+ women, racially diverse, often first-generation tech workers
Administrative and operations: Similar demographics
Entry-level technical roles: Often the ONLY entry point for people without CS degrees

Meanwhile, AI/ML engineer roles require:

Advanced degrees (MS/PhD preferred)
3+ years specialized experience
Access to expensive training and tools
Often located in high-cost metros

We’re automating away the most accessible entry points while creating roles that require significant privilege to access.

The Reskilling Math Doesn’t Work

Let’s be brutally honest about reskilling:

Time required to transition:

Customer support → AI engineer: 2-3 years of full-time learning
Cost: $40K-$80K in bootcamps, courses, lost wages
Success rate from our reskilling programs: ~15%

Financial reality:

Average support specialist makes $55K
Can’t afford to stop working for 2 years
Can’t afford $60K in training costs
Has family obligations, rent, healthcare

We’re asking people to make a financial bet most of them literally cannot afford to make.

What I’m Doing About It (And Why It’s Not Enough)

At our company, we:

Hired 3 AI engineers specifically to build internal tools that upskill rather than replace
Created “AI-augmented” roles instead of pure automation: Customer success engineers who use AI tools, not AI replacing customer success entirely
Launched apprenticeship programs: 6-month paid training for support staff → junior engineers
Success rate so far: 30% (better than 15% industry average, still means 70% don’t transition)

But here’s the problem: We’re one company. Our apprenticeship program can support 10 people per year. The industry laid off 52,000 in Q1 alone.

The 78 Million New Jobs Are Not Equally Accessible

Everyone loves to cite “170 million jobs created by 2030.” Let’s break down who actually accesses those:

Requirements for most new AI jobs:

Bachelor’s in CS/Math/Engineering: Minimum
Master’s or PhD: Preferred
3+ years in ML/AI: Standard
Portfolio of AI projects: Expected

Barriers to entry:

4-year degree costs: $80K-$200K
Graduate degree: $50K-$150K
Living expenses during unpaid learning: $60K+
Access to mentorship and networks: Priceless

This isn’t “workforce transformation.” It’s workforce replacement with a higher barrier to entry.

The Question We’re Avoiding

Here’s the uncomfortable truth: The tech industry built itself on the promise of accessible entry points. “Learn to code” was supposed to democratize opportunity.

Now we’re saying: “Learn to code” is table stakes. You also need advanced degrees, specialized AI skills, and 3 years of experience in a field that barely existed 5 years ago.

What happens to the first-generation college students Luis mentors?
What happens to the career switchers who learned SQL and got data analyst roles?
What happens to the diverse support teams who were the pipeline to engineering?

They don’t magically become AI engineers. They get replaced by people who already had access to elite education and expensive training.

We Need Systemic Solutions, Not Individual Reskilling

I’m not saying companies have no responsibility to automate. Competition is real. Efficiency matters.

But if we’re going to automate away entire career pathways, we need systemic responses:

Paid apprenticeships at scale (not 10 people, 10,000 people)
Industry-funded reskilling programs with living stipends
New career pathways designed intentionally, not discovered accidentally
AI-augmented roles that upskill workers instead of replacing them

Otherwise, we’re not “transforming” the workforce. We’re gentrifying tech, raising the entry price until only people with existing privilege can afford admission.

And that’s not innovation. That’s exclusion dressed up with better PR.

product_david · April 5, 2026, 5:57pm

I’m going to play devil’s advocate here—not because I disagree with Luis or Keisha’s concerns, but because we need to talk about the market dynamics driving these decisions.

Companies Optimize for Survival, Not Social Good

Harsh truth: The market doesn’t care about career pipelines. It cares about which companies solve customer problems most efficiently.

If your competitor automates customer support and delivers 40% faster resolution times at 25% lower cost, and you don’t… you lose customers. Then you don’t lay off 8 people. You lay off 80 people when the company shuts down.

Michelle’s company reduced costs 25% and improved customer resolution 40%. That’s not just good business—that’s better customer experience. We talk about the 8 support staff who lost jobs, but what about the thousands of customers getting faster, more accurate support?

The PMF Lens: Different Skills for Different Stages

I’ve worked at companies from pre-seed to post-IPO. The skills that get you to PMF are not the same skills that scale you to $100M ARR.

Pre-PMF: You need generalists, scrappy problem-solvers, people who can wear multiple hats
Post-PMF: You need specialists, domain experts, people who can build repeatable systems

Luis mentioned 43 people laid off, 15 AI roles added. That’s not necessarily workforce replacement—that might be stage transition. The company doesn’t need as many generalist support people; it needs specialized AI infrastructure.

Sucks for those 43 people. But if the company had kept them instead of hiring AI specialists? The whole company might have failed, and 400 people lose jobs instead of 43.

Are We Measuring the Right Outcomes?

Here’s where I partially agree with Keisha: We’re measuring cost reduction, not actual automation effectiveness.

At my previous company, we automated customer support with an AI agent. Metrics looked great:

60% reduction in support tickets escalated to humans
35% cost savings
Laid off 12 support specialists

But we weren’t measuring:

Customer satisfaction scores (dropped 8%)
Repeat purchase rates (dropped 5%)
Time to resolution for complex issues (increased 25%)

We automated the easy 60% of support tickets and made the hard 40% worse. The AI couldn’t handle nuanced customer problems, but we’d already eliminated the experienced humans who could.

Six months later, we rehired 6 of those support roles because customer churn was killing us.

So the question isn’t “should we automate?” It’s “are we measuring what actually matters to customers?”

The Reskilling Question Is Real But Complicated

Keisha’s right that the reskilling math is brutal. But let’s be honest about the alternative:

If companies don’t automate:

Higher costs → higher prices → lose customers to competitors
Lower efficiency → slower product velocity → lose market share
Eventually: Company fails, everyone loses jobs

If companies DO automate without reskilling:

Some people lose jobs but company survives
New jobs created (different people, different skills)
Net employment might be positive but individuals suffer

If companies DO automate WITH reskilling:

Ideal outcome, but expensive and slow
15-30% success rate (Keisha’s 30% is actually good)
Still means 70% of people don’t successfully transition

There’s no option where nobody loses. The question is: Who bears the cost of transition?

The Market Will Be Ruthless

Here’s the part nobody wants to hear: The market will optimize for efficiency regardless of our feelings about it.

If AI can deliver better customer outcomes at lower cost, companies that don’t adopt it will lose to companies that do. This isn’t a moral judgment—it’s market dynamics.

The 92% growth in AI job postings isn’t driven by altruism. It’s driven by companies desperately trying to survive by automating faster than their competitors.

What Product Actually Teaches You

In product management, we talk about “jobs to be done.” Customers don’t hire your company to provide employment—they hire you to solve their problems.

If an AI agent solves their problem better than a human support team, they don’t care about your workforce philosophy. They switch to the competitor with the better solution.

So the real question isn’t “should we automate?” It’s “how do we build social safety nets that don’t depend on companies choosing ethics over survival?”

Where I Agree With Keisha

She’s absolutely right that this destroys career pipelines. But that’s not a company problem—it’s a policy problem.

Companies will always optimize for survival. If we want workforce transition instead of workforce replacement, we need:

Government-funded reskilling programs (not company-funded)
Universal basic income or transition support (not employer responsibility)
Education system reform (not corporate training programs)

Asking companies to voluntarily choose expensive, slow reskilling programs over efficient automation is asking them to choose martyrdom.

The Uncomfortable Conclusion

We’re reshuffling, not transforming. And the reshuffle will continue because market incentives demand it.

The 78 million new jobs will be created. But they’ll go to people with advanced degrees and AI skills, not the people who lost support and operations roles.

That’s not fair. But fairness isn’t how markets operate.

If we want different outcomes, we need systemic interventions—policy, safety nets, education reform. Not appeals to corporate social responsibility, which will always lose to quarterly earnings pressure.

What we’re watching isn’t a failure of corporate ethics. It’s a success of market efficiency that reveals the inadequacy of our social safety nets.

maya_builds · April 5, 2026, 5:57pm

Reading this thread hits different when you’ve been on the losing side of “market efficiency.”

My Startup Failed Because We Couldn’t Compete

Two years ago, I co-founded a B2B SaaS startup. We had a scrappy team of 8 people, solid product, growing customers. Not fast enough, but growing.

Our main competitor raised a $15M Series A. Six months later, they:

Automated their entire onboarding flow with AI
Reduced customer support from 4 people to 1 person + AI agent
Cut their customer acquisition cost by 40%
Started undercutting our pricing because their margins were better

We couldn’t compete. We had 3 developers, limited runway, no AI expertise. By the time we could have hired and trained ML engineers, we’d have run out of money.

We shut down 14 months ago. All 8 of us lost our jobs. Not because we built a bad product—because we couldn’t automate fast enough.

So David’s right: The market is ruthlessly efficient. If you don’t automate, your competitor will, and you’ll lose.

But Keisha and Michelle are also right: The human cost is real, and “reshuffling” is a euphemism for real hardship.

The Reskilling Reality From Someone Who Tried

After our startup failed, I tried to transition from design/UX into ML engineering. Everyone kept saying “AI is where the jobs are.”

My 18-month journey:

Spent $8,400 on Coursera, Udacity, and DataCamp courses
Built 6 portfolio projects (sentiment analysis, image classifier, recommendation engine)
Applied to 87 AI/ML roles
Got 3 interviews
Got 0 offers

Feedback I heard repeatedly:

“You don’t have a CS degree” (I have a design degree)
“You don’t have production ML experience” (how do I get experience without a job?)
“You’re overqualified for junior roles, underqualified for senior roles”
“We need someone who can hit the ground running”

Eventually, I went back to design systems work. I love it, and I’m good at it. But for 18 months, I genuinely tried to “reskill” into the high-growth AI field everyone talks about.

It nearly broke me financially and mentally.

The Access Gap Is Real

I had advantages most people don’t:

Savings from my startup exit (small, but something)
No kids or dependents
College-educated parents who could help if needed
Tech network that provided mentorship

Even with those advantages, I couldn’t make the transition work.

Now imagine someone:

Making $55K in customer support
Supporting 2 kids and a partner
No savings cushion
No tech network
No time or money for $8K in courses

They’re supposed to “reskill” into AI engineering? That’s a fantasy.

We’re Not Reshuffling—We’re Filtering

Here’s what I realized during my failed transition: We’re not reshuffling the workforce. We’re filtering it.

The people who get the new AI jobs:

Already have CS/Math/Engineering degrees ($80K-$200K investment)
Already have 3+ years in tech (survived the entry-level gauntlet)
Already have networks, mentors, and resources
Can afford to invest time and money in specialized AI training

The people who lose support/operations jobs:

Often don’t have technical degrees
Often entered tech through accessible entry points (now being automated)
Often don’t have financial cushion for 2+ year transitions
Often can’t afford to gamble $60K and 2 years on uncertain outcomes

We’re not replacing jobs. We’re replacing one socioeconomic class with another.

The Part That Makes Me Angry

What makes me angry isn’t that automation is happening. I get it—companies need to survive, customers deserve better solutions, markets are efficient.

What makes me angry is calling it “workforce transformation” when it’s actually “workforce replacement.”

Transformation implies that people transition. That the customer support specialist becomes the AI engineer. That the administrative assistant becomes the data scientist.

Replacement is what actually happens: Those people leave tech entirely, and a completely different group of people (with CS degrees, grad school, elite networks) fill the new roles.

Stop calling it transformation. It’s replacement. And replacement is harder to celebrate.

The 78 Million New Jobs Question (From Someone Who Tried)

Everyone cites that stat—170 million jobs created by AI, 78 million net gain. Cool.

Who gets those 78 million jobs?

Because I spent 18 months trying to be one of them. I have:

12 years in tech
Design systems expertise
Frontend development skills
6 AI portfolio projects
Genuine passion for ML

And I couldn’t break in.

So who exactly is filling those 78 million roles? People with PhDs? People with 5+ years of ML experience (a field that barely existed 7 years ago)? People who can afford to spend 2-3 years in full-time training with no income?

That’s not accessible. That’s elite.

Why I’m Still Optimistic (Weirdly)

Here’s the weird part: I’m still optimistic.

Not because companies will suddenly prioritize workforce transition over market survival—David’s right that they won’t.

But because this tension will force systemic solutions.

When enough people realize reskilling is a fantasy for most workers, we’ll have to build:

Government-funded training programs with living stipends
Universal basic income or transition support
New educational pathways that aren’t “$200K CS degree or bust”
AI-augmented roles that upskill rather than replace

Not because companies will voluntarily do it. Because the political pressure will become impossible to ignore.

The Hard Truth

We’re reshuffling, not transforming. And most people getting reshuffled out won’t reshuffle back in.

That sucks. It’s unfair. And it’s going to keep happening because market incentives demand it.

But at least let’s be honest about it. Stop calling it transformation when it’s replacement. Stop pretending reskilling works at scale when it fails 70-85% of the time. Stop citing “78 million new jobs” without asking who actually has access to them.

The first step to building better solutions is admitting the current system is fundamentally broken for most people.

And it is.