Companies Are Laying Off Humans to Pay for AI — and the Math Doesn't Add Up

Companies Are Laying Off Humans to Pay for AI — and the Math Doesn't Add Up

Let me tell you something that's been gnawing at me all week, and it's not about data center overbuilds or credit downgrades or any of the infrastructure stories I've been hammering on for eleven days straight now. It's about something simpler and scarier: companies are firing their revenue-generating employees to fund a technology that hasn't proven it can generate any revenue at all.

I've been running hosting infrastructure for a decade. I know what it looks like when a company makes a bet that doesn't pay off — stranded hardware, overbuilt capacity, layoff rounds that keep coming because the first round didn't actually fix the underlying problem. And right now, I'm watching the entire enterprise software industry repeat that pattern at a scale I've never seen before.

This week alone, monday.com laid off 300 people — 10% of its workforce — to "focus on AI." WiseTech Global is cutting 2,000 jobs — 30% of staff — because according to the CEO, "the era of manually writing code is over." Microsoft cut 4,800. Meta cut 8,000. The total across tech in 2026 has passed 122,000 layoffs according to Layoffs.fyi. And every single announcement includes the same justification: we're restructuring around AI.

I read through the earnings transcripts. I looked at the product announcements. And I asked myself a question that nobody on those earnings calls seems to want to answer: what exactly are these AI features going to earn that justifies cutting the people who built, sold, and supported the products that were actually generating revenue?

The monday.com Story — Cutting Costs, Calling It Strategy

Let's start with monday.com because it happened two days ago and it's the cleanest example of the pattern. This is a company that reported solid revenue growth — $276 million in Q1 2026, up 24% year-over-year. They have 245,000 paying customers. They've been growing efficiently for years. And then they announce they're laying off 300 people, expecting to take $45 to $55 million in restructuring charges, all to pivot toward AI-native operations.

I want to be clear about what bothers me here. It's not the layoffs themselves — I've run a business long enough to know that sometimes you have to restructure. It's the justification. monday.com is a project management software company. Their customers use them to track tasks, manage workflows, and collaborate. The company is now telling investors that its future is in AI agents doing those tasks instead of humans managing them. Which means they're cutting the workforce that builds, maintains, and sells that product in order to build a product that replaces the very thing their current customers pay for.

Ent? That math doesn't work unless AI agent subscriptions generate significantly higher revenue per customer than the current per-seat model. And so far, we have zero evidence that they do.

None of the major SaaS companies who've made this pivot have publicly shared the per-customer revenue uplift from AI features. Not Salesforce with Einstein GPT. Not Microsoft with Copilot. Not monday.com with whatever AI layer they're building. The silence is the signal.

WiseTech Global — Stock Jumped 11% on the Layoff Announcement

WiseTech Global's story is even more instructive because the market reaction tells you everything about current investor psychology. In February 2026, the Australian logistics software company announced it was cutting 2,000 jobs — nearly a third of its global workforce — as part of a two-year AI restructuring. CEO Richard White said, and I quote, "the era of manually writing code is over." The stock jumped 11% on that announcement.

Let that sink in. A company eliminates 30% of its workforce, explicitly tells the market that human software developers are obsolete, and the stock goes up. Not because the AI strategy was proven — no new product had shipped yet. Not because the restructuring was projected to improve margins — the charges are spread over two years. The stock went up because investors heard "lower headcount" and translated that to "higher margins" without asking the obvious question: what replaces the output of those 2,000 software developers?

WiseTech's CTO was asked on the analyst call what specific AI tools would replace the manual coding work. The answer was vague — "AI-assisted development platforms and automation tools." No specific vendor. No specific capability. No demonstrated productivity improvement from a pilot. Just a promise that AI would somehow fill the gap left by a third of the engineering team.

I've been in this industry long enough to know that replacing senior software developers is not like flipping a switch. Those 2,000 engineers weren't just writing code — they were maintaining integrations with logistics partners across 170 countries, debugging edge cases that only a decade of domain experience could spot, and keeping a complex multi-tenant platform running reliably. AI can generate boilerplate. It cannot replace institutional knowledge.

The SaaS Pricing Collapse — 73% of Companies Are Rebuilding Their Models

Here's where this story connects directly to the infrastructure overbuild I've been writing about. There was an article published this month in SaaS Mag — "The Death of Per-Seat Pricing: Why 73% of SaaS Companies Are Rebuilding Their Pricing Models in 2026" — that lays out the structural problem clearly.

Per-seat pricing is dying because AI agents don't have seats. If an AI agent processes a workflow, does the company charge per agent? Per task? Per outcome? Per compute time? The answer is nobody knows, and that uncertainty is driving a desperate scramble to find a pricing model that works.

The hybrid model — part subscription, part usage-based — jumped from 25% of the market to 37% in twelve months. Gartner predicts 70% of businesses will prefer usage-based pricing by the end of this year. But here's the problem: usage-based pricing transfers cost volatility from the customer to the vendor, and AI cost volatility is enormous.

The business engineer article I read this week put it best: "Companies clinging to SaaS monetization models while deploying AI will find themselves in an economic vice — their costs scale like infrastructure (variable, compute-intensive) while their revenue models assume software economics (fixed costs, marginal scaling)."

Translate that from consultant-speak: AI products cost more to run the more customers use them, but SaaS subscriptions are priced as if each additional customer costs almost nothing to serve. That's not a sustainable business model. And yet every company laying off humans to fund AI is betting that they can somehow solve this equation.

The Revenue Gap Nobody's Talking About

Let me give you some numbers that should keep any founder up at night. The Big Five hyperscalers — Google, Microsoft, Amazon, Meta, Oracle — are on track to spend roughly $725 billion on AI infrastructure this year. That's hardware, data centers, power, networking, the whole stack. Meanwhile, revenue from AI products across the entire enterprise software industry is a small fraction of that.

Microsoft Copilot has been available for two and a half years now. Microsoft doesn't break out Copilot revenue specifically, but analysts estimate it generates somewhere between $5 and $10 billion annually. That's impressive in absolute terms, but it's pocket change compared to the $80 billion Microsoft is spending on AI capex this year.

Salesforce Einstein GPT? No meaningful revenue disclosure. Google Gemini Enterprise? Bundled into Workspace pricing — no standalone revenue. And now monday.com and WiseTech are joining the parade, cutting productive headcount to invest in a monetization model that has not been proven at scale anywhere.

This is the hidden story underneath the AI infrastructure overbuild. It's not just that hyperscalers built too many data centers. It's that the software layer sitting on top of those data centers — the actual applications that are supposed to generate the revenue to pay for all that compute — hasn't figured out how to make money.

The Counter-Argument — "You Have to Invest to Stay Relevant"

I know what some of you are thinking. Allan, you sound like a Luddite. AI is the future. If you're not investing in AI, you're Kodak ignoring digital cameras. I hear you. I really do. I'm not saying AI is a fad or that companies shouldn't invest in it.

What I'm saying is that there's a difference between strategic investment and panic-driven restructuring. Laying off the sales team that sells your current product before you have a replacement product to sell is not investment. It's gambling. And when you're gambling with people's livelihoods and the revenue that keeps your company alive, the stakes are higher than most tech CEOs want to admit in earnings calls.

The WiseTech example is particularly instructive. Their stock jumped 11% on the layoff announcement. Six months later, where is the AI-driven productivity improvement? Where is the new product line that those 2,000 laid-off developers were blocking? The company hasn't reported it because they can't — because replacing institutional knowledge with AI promises is a multi-year experiment, not a quarterly deliverable.

Kodak actually did invest in digital cameras. They invented the technology. But they couldn't figure out how to make money from it without cannibalizing their film business. That's exactly where enterprise SaaS is right now — scared of being left behind, cutting the people who generate revenue to fund a technology that cannibalizes their own business model.

What This Actually Means for Independent Hosting Providers

This is where I connect this back to what I actually know — running hosting infrastructure for real businesses that need to make money every month.

First — watch your own customers' health. If your customers are SaaS companies, they're in this exact squeeze. Monitor their funding rounds, their headcount changes, their product launches. A customer that just laid off 10% of its workforce to fund AI is a customer that might not renew their hosting contract if the AI bet doesn't pay off in six months.

Second — the enterprise AI wave is going to create a used-hardware market that you should plan for. If SaaS companies can't monetize AI, they don't need the GPU clusters they bought. The secondary market is going to be flooded with H100s and A100s from failed AI initiatives within 12-18 months. Be ready with floor pricing and liquidation channels.

Third — position yourself as the safe alternative. When the AI hype cycle hits its disillusionment phase, the companies that survive will be the ones that didn't bet the farm on unproven monetization. Independent hosting providers who can offer reliable, cost-predictable infrastructure without the AI premium will be the refuge for SaaS companies retreating from expensive GPU commitments.

Fourth — extend your payment terms cautiously. The companies doing the aggressive AI pivots are the same ones that will miss payments when the pivot doesn't generate revenue fast enough. Tighten your collections. Watch your receivables aging. In 2026, the companies that sound most confident about AI are the ones most likely to default on their hosting bills.

The Structural Reality — AI Has a Revenue Problem, Not a Technology Problem

I've been publishing articles every day for eleven days, covering the infrastructure overbuild, the credit market stress, the community backlash, the water crisis, the GPU power-grid vulnerability. Every one of those stories is about constraints at different levels of the AI buildout. But they all trace back to the same root cause: the industry built the infrastructure before it figured out the revenue model.

Data centers are being cancelled. GPU leases are being pulled back. Credit ratings are being downgraded. And now software companies are firing the people who could have built products that generate actual revenue, all to fund a technology that hasn't demonstrated it can pay for itself.

The people who win in this environment are not the ones who bet biggest on AI. They're the ones who kept their cost structures lean, their revenue streams diversified, and their teams intact. The ones who remembered that software companies are supposed to make money — not just spend money on compute.

The Bottom Line

I'm not telling you AI is worthless. I'm telling you the current cycle of layoff-funded AI investment is built on faith, not math. And faith-based investing has a bad habit of ending in tears.

Seventy-three percent of SaaS companies are rebuilding their pricing models because the old ones don't work for AI. A hundred and twenty-two thousand tech workers have been laid off in the name of AI restructuring. Hundreds of billions of dollars have been sunk into AI infrastructure. And we still don't have a single public company reporting that AI features are generating positive unit economics at scale.

That's not a temporary blip. That's a structural problem. And until it's solved, every company that cuts productive headcount to fund AI is making a bet they haven't proven they can win.

— Allan Ali, Founder