Transcription
Amazon is not cutting 30,000 jobs because they have too many managers. Whatever they may say, they're cutting 30,000 jobs because they need the money to buy GPUs. That's the story nobody wants to tell you. Not Andy Jesse on the earnings call. Not the business press dutifully repeating the culture narrative. Not the analysts who'd rather talk about AWS growth rates than the brutal arithmetic reshaping big tech from the inside out.
So here's what really happened. Amazon's quarterly free cash flow went negative. Negative $4.8 billion to be exact. At the exact moment their capital expenditure hit $125 billion, the highest of any company on Earth, with 75% of it going directly to AI infrastructure. They raised $12 billion in debt this year to fund data centers. And then they eliminated 10% of their white-collar workforce. That's the 30,000 jobs. This is not a layoff. This is a capital reallocation. Human headcount is being converted to compute capacity. Salaries are being transformed into silicon. And if you understand why Amazon is making this trade, you understand something very important about what's coming for the entire tech industry, because the numbers just don't add up.
Let's start with the contradiction that should make you suspicious of every official explanation you've heard. Amazon just posted one of its strongest quarters in years. Revenue hit $180 billion, up 13% year-over-year. AWS is growing at 20%, the fastest clip since 2022. Net income surged 38%. The stock jumped 10% after earnings. And by every conventional measure, this company is firing on all cylinders.
And then they fired the 30,000 people. Not warehouse workers, not seasonal fulfillment staff, corporate employees, engineers, product managers, the people who build things. And what will become the largest layoff in Amazon's 30-year history. The company is eliminating roughly 10% of the white-collar workforce. 14,000 in October, another 16,000 this week. And that was planned together.
CEO Andy Jasse wants you to believe this is about culture. "The announcement was not really driven financially," he said. And "it's not even AI-driven," he said. Instead, he said it was about culture. Too many layers, too much bureaucracy. Amazon, he continues to beat this drum, needs to operate like the world's largest startup. That is a compelling narrative. It's also carefully constructed to obscure what the balance sheet makes obvious.
Amazon's free cash flow tells a story that Andy Jasse, I think, would rather not discuss on earnings calls. In 2024, Amazon generated $38 billion in free cash flow. That sounds healthy until you understand what happened next. As of Q3 2025, quarterly free cash flow went negative. The trailing 12-month figure dropped 61% year-over-year. Free cash flow margin collapsed from 8.73% of sales to just 2.7%. What changed? Capex. Capital expenditure exploded. Amazon spent $83 billion on capex in 2024. In 2025, that number jumped to $125 billion, a 61% increase. CFO Brian Olsavski has already told investors to expect even higher spending in 2026. Amazon isn't stopping.
Roughly 75% of all of this money is going directly to AI infrastructure, GPUs, custom Tranium chips, data centers, and the power systems that run them. To put these numbers in perspective, Amazon is now spending more on infrastructure in a single year, one year, than the entire gross domestic product of the country of Morocco. They're adding 3.8 gigawatts of data center capacity annually, enough to power 3 million homes every single year. They're on track to double their entire computing power by 2027.
This is not a company with excess cash lying around. This is a company that went to the debt markets and raised $12 billion in bonds specifically to fund infrastructure investments. And oh, by the way, it's not a zero-interest rate environment anymore. So even if it's Amazon and it's a blue-chip bond, it's not free.
When Bank of America's credit strategists analyzed hyperscaler spending, they found that aggregate capex among the big five, that is Amazon, Microsoft, Google, Meta, and Oracle, now consumes, wait for it, 94% of operating cash flows after dividends and buybacks. That's approaching the theoretical maximum a company can spend without taking on really significant leverage.
So, what does this really mean? Amazon isn't cutting people because the culture is broken. Amazon is cutting people because they need the money. Look, let's do the math that Amazon's PR team hopes that you and I wouldn't do. 30,000 corporate employees eliminated. So, what does that actually save us? Amazon's median total comp for corporate roles runs around $217,000 per year. That includes base salaries, stock grants, and benefits. Assume the layoff pool skews a little bit below the median. Make it make the math. Call it $200,000 per head all in. 30,000 employees times $200,000 equals roughly $6 billion in annual savings.
Now $6 billion sounds like a rounding error against $125 billion in capital expenditure. But that is the wrong comparison. The relevant comparison is free cash flow. The actual money available to fund investments after you've paid for everything else. When your quarterly FCF is $4.8 billion, well, an extra $6 billion a year actually matters a lot. That $6 billion is roughly what Amazon spent building Project Kuiper, its massive AI computing platform packed with half a million custom Tranium 2 chips. It's enough to fund half of their annual investment in new data center campuses. It's the difference between issuing more debt and funding expansion internally.
The pragmatic engineer argued against this thesis when the first wave of layoffs hit in October. And his analysis pointed to Amazon's $93 billion in cash reserves and $32 billion in trailing free cash flow, concluding that the layoffs won't make a big difference to how much it can invest in data centers. I would argue that analysis used backward-looking metrics. The cash flow picture had already deteriorated even more. Quarterly FCF was negative. Capex guidance had increased by $7 billion. And Amazon had just raised another $12 billion in debt, something they wouldn't have had to do if they were swimming in excess cash.
The numbers don't lie. Amazon is under financial pressure. Not because the business is weak, but because the AI infrastructure race demands capital at a scale that even Amazon's profit engine struggles to generate.
So what do we make of the culture narrative that Jassy gave? Understanding why Jassy frames this as a culture problem rather than a financial reallocation reveals something really important about how CEOs communicate during times of technological transformation. There are three audiences that Jassy has to satisfy, and each one needs to hear, frankly, a different version of reality.
For employees, where cutting you to buy more GPUs is a really brutal message. It implies your work doesn't matter, that you're simply a line item to be optimized away. The culture framing is a lot more gentle. It suggests that the cuts are about organizational effectiveness and not your individual value. You weren't eliminated. You weren't eliminated because you cost too much. You were eliminated because there were too many layers between you and the customer. And to be fair to Andy Jassy, having worked at Amazon into the 2020s, he's not wrong. There are a lot of layers. It is possible that the culture narrative is true at the same time as the hard numbers forced his hand.
For investors, admitting financial pressure is dangerous. Amazon trades at a premium because the market believes AWS and AI will generate massive returns. If Jassy said, "We're cash-constrained and need to cut headcount to fund infrastructure," the stock would be in trouble. The culture narrative lets him position the cuts as proactive optimization rather than reactive cost-cutting.
For regulators and the public, the AI replacement story creates problems. If Amazon admits these layoffs are about freeing capital to build AI systems, they invite a lot of scrutiny, perhaps congressional hearings, perhaps labor activism, bad headlines. It's better to frame it as routine corporate restructuring. And I think that's part of why Jassy chose to call out explicitly that these layoffs were not about AI, as a productivity tool.
The tell in all of this is in Jassy's own contradictions. In June of 2025, just four months before that first round of the 30,000 were laid off, he sent a memo to employees explicitly warning that AI would mean Amazon needs "fewer people doing some of the jobs that are being done today." That memo is still on Amazon's public blog. But by the October earnings call, the message had shifted. "Not AI-driven." "Not right now, at least." The hedge in that sentence, "not right now, at least," reveals the tension. Jassy knows AI is part of this story. He just can't say it out loud when the analysts are listening.
And again, to be fair to Jassy, the culture story isn't fabricated. It's just not the whole truth. Did overhire during the pandemic. Between 2017 and 2022, the company's corporate headcount tripled. I was there during a lot of that time. When you grow that fast, you inevitably add layers. Middle managers accumulate. Decision-making slows down. I saw that happen. The ownership culture that defined Amazon's early years gets diluted across too many people, too many meetings, and too many documents. Jassy has been talking about this problem for over a year. And in September of 2024, he mandated a return to 5 days in the office as part of his goal to reintroduce startup discipline to Amazon. Every major organization, he said, had to increase its ratio of individual contributors versus managers by at least 15%. And he created a "no bureaucracy" email alias where employees could flag unnecessary processes. And that initiative, as much as it sounds silly, generated 1500 responses and led to apparently 450 changes. I appreciate that focus. As someone who saw decision-making slow down, who saw too many meetings, too many docs, I love that he's doing this. And so when Jassy says these layoffs are about culture, he's drawing on a real diagnosis. Amazon really did become bloated. The company does move slower than it used to. This is not the "Day One" thinking Jeff Bezos liked to talk about. It's very "Day Two."
But here's what the culture and narrative cannot explain. The timing. If the problem was pandemic overhiring, why didn't Amazon fix it in 2023 when everyone else did? They laid off 27,000 people that year. Is that enough? Apparently not. If the bureaucracy problem persisted, why wait until late 2025 to address it with the largest workforce reduction in company history? And why announce those cuts the same week you're telling investors the capital expenditure will hit $125 billion and be even higher in 2026? The culture problems are real, but they didn't suddenly become urgent in October of 2025. What became urgent was the cash, because free cash flow turned negative.
To fully understand what's happening at Amazon, you have to zoom out and see the competitive landscape driving this spending. The AI infrastructure buildout is the largest capital deployment in the history of technology, bar none. Bigger than the moon race, bigger than the atom bomb, anything you can think of, bigger than the pyramids. Goldman Sachs projects the top hyperscalers will spend $1.15 trillion on infrastructure between 2025 and 2027, more than double what they spent in the previous 3 years combined. For 2026 alone, aggregate capex among the big five is going to exceed a half a trillion dollars. It's estimated to be $600 billion. These are not discretionary investments. For the big five, these are existential. The companies that build the most advanced AI infrastructure first are going to capture the majority of enterprise AI spending for the next decade at least. The companies that fall behind are going to find themselves locked out of the most lucrative technology market ever created.
Microsoft has OpenAI, kind of. Google has Gemini. Amazon has Anthropic, kind of, and its own Nova models. Each of them is racing to build the compute capacity that will run the AI applications every business needs. And that is why nobody is pulling back despite the enormous expense. Meta just raised its capex guidance to $100 billion for 2026. Microsoft's capital intensity has hit 45% of revenue, historically unthinkable for a software company. Oracle, which barely registered in cloud at all 5 years ago, is now deploying tens of billions in AI infrastructure.
And Amazon's position in this race is particularly vulnerable. It remains the dominant cloud platform, but it has lost ground to Microsoft Azure and Google Cloud in AI-specific workloads. The perception that Amazon was slow to AI, late to partner with leading model providers, late to ship competitive AI services is very widespread and has weighed heavily on the stock. The $38 billion OpenAI infrastructure deal announced in January sort of helped to address that perception, but it also represents yet another massive capital commitment.
When you are in the infrastructure arms race and you need every dollar you can find, cutting $6 billion in annual headcount to flip yourself toward free cash flow positive, that's not just attractive, it becomes a necessity.
So what does this mean for the rest of us? The Amazon layoffs matter beyond Amazon for two big reasons. First, they reveal the true cost structure of the AI transition. Every hyperscaler is facing the same pressure. How do you fund a $100 billion-plus in annual infrastructure spending while maintaining profitability? The answer increasingly is that you fund it by cutting everywhere else. Research budgets, experimental projects, moonshot bets, and yes, people. This isn't about AI replacing workers in the sense that most people imagine. Robots are not doing the jobs that Jassy cut. Robots doing your job is something that people increasingly envision as the future. But that's not where we're actually seeing these cuts. The more immediate reality is that AI creates capital demands so enormous that companies have to shrink their human workforces simply to afford the infrastructure to play the game. You're not being replaced by an AI, really. You're being replaced by the need to buy GPUs.
Second, Amazon's cuts signal something broader about tech sector employment. The industry added workers at a very unsustainable rate during the pandemic. And the correction that started in 2023 is not over yet. Microsoft cut 15,000 people. Meta continues its year of efficiency. Intel eliminated 24,000 positions. I found data from Challenger that shows that over 1.1 million job cuts hit across the economy in 2025. And that's the highest number of job cuts since the pandemic itself.
So we've lived through these cycles up and down in tech. You know, I remember 2001. I remember 2008. What makes this cycle different is that the companies doing the cutting are not struggling. They're not losing money. They're not fighting for survival. These are some of the most profitable enterprises in human history. If they eased up on the AI capex, Amazon would be minting money, but their capital allocation priorities have shifted decisively from human labor to physical infrastructure. And that shift is, I think, structural rather than cyclical.
The workers who remain are going to be expected to do more with less on a long-term basis. Amazon employees report that managers now track AI tool usage via dashboards. Performance reviews increasingly factor in how effectively you leverage automation in your role. The implicit bargain is very clear: justify your existence by being more productive than the machines that are otherwise going to come for the role. Use the machines instead. Leverage yourself. And this comes back to one of the big things I like to talk about in these videos: How are you going to use AI as a mech suit to expand your scope, to be able to do more because you have so many more tools? I think it's the critical job skill for the next decade.
I don't want to be too pessimistic here. There is a version of this story that is more optimistic. Amazon is investing $125 billion in infrastructure that will power the next generation of AI services. Those services will generate new businesses, new jobs, and new value. And the short-term pain and layoffs is going to unlock long-term gains across the economy as AI unlocks productivity improvements. History tends to suggest that industrial revolutions, however painful in the moment, ultimately create much, much more prosperity than they destroy. The problem is we're not through the pain yet.
All we know for certain right now is that Amazon is making a very deliberate choice. They're choosing the GPUs over the people, data centers over headcount, infrastructure over organization. And they're making that choice not because their business is weak, but because the competitive dynamics of AI demand capital at a scale that forces hard trade-offs even among the most prosperous companies on Earth.
Andy Jassy is right to keep talking about culture. He's going to keep invoking the startup mentality and the ownership mindset and the need to move faster. He's right on all of those counts. Those things matter. They're overdue at Amazon, and they are some of the changes that Amazon needs to make to get back to "Day One" thinking and reinvent itself. I'm glad he's doing that. But I don't want to ignore the numbers, and I think they're not being talked about enough. Free cash flow going negative is unacceptable. That is a company that would have difficulty servicing its debt before long. Capex hitting $125 billion forces hard choices at the margins. And the hard choices were 30,000 people out the door. And a $12 billion bond offering to fund what cash flows can't cover.
This is the equation that's starting to reshape big tech. And I think we, we should just be honest about it. Human capital is at risk when it competes with compute capital. And if you think Amazon is alone in making this calculation, then you haven't been paying attention. Every major technology company is running the same arithmetic. The only question is who announces their layoffs next.
And for those of you who are impacted by the Amazon layoffs, you are going to make it. You are going to get through. I am an ex-Amazonian. There's a whole bunch of ex-Amazonians out there. Reach out to us. We're all alums. We'd love to be helpful. Best of luck out there.