A little bit off topic: but I couldn't even start to read the article because "I reached my article limit" out of I site I never visited before... What are they using to determine how many articles I've read?
Opening in a private window solved the issue, however I'm pretty sure I don't regularly read anything on this site (maybe never was an overstatement?).
I can't read the article but that won't stop me from commenting..
This year alone something like 400B was spent on investing in chips, datacenters, electricity buildouts. That's 400B that could have otherwise been invested in people.
While i don't doubt that people will find a few solid business cases for LLMs, i am on team-bubble. I don't think this investment will add 400B worth of value and I very much doubt that this 400B is any good for future growth or long-term aspirations of AGI. Investing 400B into people and (tech) manufacturing would be a solid long-term bet with benefits.
Matter of time until markets reckon with AI investment crowding out non-AI investment (cf. the massive oversubscription of Meta's latest bond offering). Must suck to be a small-cap firm squeezed by tariffs raising costs, unemployment lowering demand, and AI investment raising your own non-AI cost of borrowing.
That's an interesting point that I haven't considered before: that the narrative of AI replacing jobs plus the widespread cheating in school using LLMs is making students less engaged and new graduates less employable, becoming a self-fulfilling prophecy for AI.
I think the article is missing two points: if the latest layoffs aren't related to AI, then this doesn't mean AI won't have or has an impact on head count.
And investment and experiments by definition include the risk of failing. In almost everything lies a survivorship bias and no one talks about the 100+ car makers that went into goldrush mode 100+ years ago. This is life. Netflix vs Blockbuster - already forgotten?
Also the "fail rate" - so what part is failing and why? What's with the 5%? If we have a look at exponential functions this might be a really good deal, if the 5% can account for the losses. After all, benefits compound over time.
I witnessed first hand in FAANG some quota hires and I believe that now that no one gets paid for contrived and artificial business advantages, we are back to a more merits based evaluation of workers.
But AI should not be written off as fancy something with no impact. That's the wrong take. Whether it will be a springboard to new jobs that compensate for losses or replacements - I am not yet sure, but tent to be in the former group. ML engineers take care of ML - something new that takes care of something new.
If Amazon’s layoffs aren’t AI driven, and AWS is making more money than it’s spending on AI infra… how is Amazon evidence of AI spending replacing jobs? This is an interesting topic, but this particular article left me pretty unsatisfied, it feels like juxtaposing a bunch of barely related numbers with some very popular talking points, but no new information? It hints at “AI washing” without doing any digging at all, and cites the MIT study without noting that it’s getting a ton of legitimate criticism. Is AI really a sideshow or an excuse for layoffs that big companies were hoping to do anyway, is it taking the blame for low confidence in the economy?
This does not reflect my reality. I have worked on half a dozen projects where we would have hired consultants to get the job done but have used coding agents to document, understand, migrate a project, create backend services, dashboards, and other things with budgets ranging from a few thousand euros to 200,000 euros. No consultants hired, nothing spent.
Over the last 20 years of tech, the giants have taken the smartest folks out there and put golden handcuffs on them. You could hire up all the smart folks and put them to work, or leave them out there and have them compete with you. With the launch of cloud providers and (expensive) dynamic scaling the problem only got worse. Think about hiring in the pandemic. Every one at home, with a stimulus check and nothing to do. Rather than a flurry of new software you got mass hiring.
But now we are in a capital intensive hardware cycle. Where in order to compete you need to have lots of $$$$ as well as software know how. It does not matter that there are smart people out there, without hefty backing they wont get very far.
I suspect that software is about to enter its "punk" era. We have software for small businesses that will help with accounting, HR, customer service, and cloud providers are starting to see some interesting competition. Much like the old punk poster showing you 3 chords and telling you to start a band it is entirely possible to find three friends and start a business that makes 1-10 mill a year with little effort and lower costs. The moment you stop thinking "unicorn" and start thinking "sustainable" the economics shift radically.
As a recently laid of senior engineer, to the extent that my job was replaced, it was replaced with offshore junior devs who'd already been working with the company for over year with a pretty rough level of productivity by man-hour, though maybe taking 3x the time to get things done is worth it if they're cheap enough. Which is to say I see my layoff as as cost cutting backed by a premise that there is no value in retaining senior level talent, to try and keep operating in the black, not because AI was materially producing a lot of benefits. (Because to the extent it was, I was the one reaping them compared to the offshore folks and less experienced onshore ones.)
>> though maybe taking 3x the time to get things done is worth it if they're cheap enough.
Started doing automation about 6 years ago. We had several offshore teams. The one caveat was that the company was vying for several government contracts. One of the requirements was the ability to pass a low level security clearance and on yeah, you cannot, under any circumstances off shore any of the work.
Therefore, they needed an onshore team to handle the government work.
Well, when things got slow, our team was essentially the clean up team. We had several hockey players on our team so we nicknamed our team "Zamboni". A large chunk of what we were doing was fixing the off shore teams coding and things would break so regularly two of devs became full-time off shore support.
The company was literally paying our off-shore team to build an automation. Then when it constantly break, they would pull us in to fix it.
Absolutely maddening they couldn't understand they were paying triple or quadruple for work we could do once properly instead of two or three times before we got it.
Because in any country with poor worker protections, the outcome is layoffs regardless.
AI succeeds? Layoffs of unneeded roles.
AI fails? Layoffs to cut expenditure to make up for the written-off expenditure.
In the UK, if AI makes a well-established employee redundant, the employee is entitled to redundancy pay. And if the company fucks up and overspends on chasing a ridiculous Macguffin they can't just fire people without making them formally redundant.
The damage that is going to be done in the USA if the AI bubble bursts is going to be generational.
Slightly tangential to the article: a lot of the "AI layoffs" are really just old fashioned layoffs, but with exciting press releases meant to reassure investors.
AMZN for example overhired in various functions because it expected demand that never materialised.
Admitting that is bad for the share price, but writing some woo about "AI and agility" will convince at least some investors to keep the faith.
The article seems to hinge on the core assumption that revenue is much less than spending:
> Those expenditures may be approaching $1 trillion for 2025, while AI revenue—which would be used to pay for the use of AI infrastructure to run the software—will not exceed $30 billion this year
While it's clear that the author is summing up the spending from the big players, it's not clear to me that their math is right for revenue. Yes, OpenAI, Anthropic, Thinking Machines, SSI, etc. have pretty limited direct revenue (including zero!).
But this comparison assumes no revenue growth for other top computing users. Some companies are certainly saving money on some tasks and increasing revenue, particularly in fields like customer support. See the confusing figure in section 5 of https://hai.stanford.edu/ai-index/2025-ai-index-report/econo... .
That chart is by number of respondents and not weighted by revenue. Like the MIT study, it would not be surprising that "just pipe this to an LLM" isn't enough for most fields or companies. But a few have likely made material improvements.
10% of respondents saying they've seen a >10% revenue gain could be substantial, if they're bigger firms with high leverage in computing.
Edit to add: the comparison also makes a classic "GDP vs market cap" style mistake. Capital expenditure has multiple years of useful life. Revenue is annual. You'd want to compare depreciation vs revenue.
AI processing hardware deprecates (and depreciates) at a much faster rate than conventional CPUs, as much as 50% per year. Consider the billions being dumped into compute at that rate of depreciation and explain to me:
1. How will tangible assets generate profit net of near term capex requirements and interest on debt?
2. Why wouldn't payroll shrink as a result of the increased AI capabilities emerging from the capex spend?
3. If AI lives up to the hype & given recent news that public backstops are being requested, why shouldn't the US quasi-nationalize cash strapped players and distribute equity to every American?
4. As NVDA and AAPL local models and local compute eat into utility and base automation business, how do edge players maintain profitability without pricing capabilities well beyond the affordability of SMBs and individuals?
So, jobs are being replaced by AI spending (not by AI), and AI spending is increasing because of AI, which means AI is replacing the jobs. Did I get wrong?
This article gets the phenomenon right but the causation wrong: it's not "AI spending vs. AI replacing jobs". both are happening simultaneously, and they're causally linked.
The spending-revenue gap is real. Hyperscalers are projected to spend $300-550B on AI infrastructure in 2025[1] while generative AI revenue won't exceed $30-40B [2]. Amazon's capex jumped from $48B in 2023 to $84B in 2024 to a projected $100B+ in 2025[3], that's capital intensity doubling from historical norms of 11-16% to over 22% [4].
But here's what the article misses: this isn't financial desperation. When Amazon's CEO announces 14,000 layoffs and explicitly states that AI will enable "fewer people doing some jobs"[5], he's revealing the strategic logic — show me the incentives and I'll show you the outcome. Companies aren't cutting jobs despite AI spending; they're cutting jobs because they know AI spending will pay off.
To be clear, the article treats the spending-revenue gap as evidence of irrationality. But infrastructure buildouts always precede revenue: railroads looked insane before they transformed commerce, electricity grids consumed massive capital before delivering returns, the internet required enormous infrastructure investment before creating trillion-dollar companies.
What's different now is companies are pulling the future forward. If we take this article at face value which I can appreciate is a BIG “if” then AI is already automating 25% of tasks and delivering 10-55% productivity gains[6] so they're not waiting for AI to replace jobs organically. They're cutting headcount now to fund the infrastructure that will make those cuts permanent.
More broadly, this is rational capital reallocation in a winner-take-all race. Companies that don't build AI infrastructure won't gradually decline, they'll lose competitive positioning entirely. That's why Meta is using off-balance-sheet financing for a $27B data center[7], why Oracle is borrowing $25B annually despite already carrying 450% debt-to-equity [8]. They're all-in because the alternative is obsolescence.
The real story isn't "spending causes cuts" it's that AI infrastructure commoditizes human expertise, the complement to compute infrastructure. Companies are trading labor costs for compute infrastructure because they've correctly identified compute as the new moat. The job cuts aren't the price of spending on AI; they're the business model shift that AI enables.
The article is right that we're not seeing mass AI job replacement yet. But the job cuts are happening in anticipation of replacement, not as an unfortunate side effect of spending. That's not desperation just business strategy.
If AI really generates the value it claims to cutting jobs is short sighted. If existing human knowledge is commoditized, then we should be able to invest in generating new knowledge, and creating new kinds of products that were not even possible before.
61 comments
[ 3.5 ms ] story [ 69.3 ms ] threadOpening in a private window solved the issue, however I'm pretty sure I don't regularly read anything on this site (maybe never was an overstatement?).
This year alone something like 400B was spent on investing in chips, datacenters, electricity buildouts. That's 400B that could have otherwise been invested in people.
While i don't doubt that people will find a few solid business cases for LLMs, i am on team-bubble. I don't think this investment will add 400B worth of value and I very much doubt that this 400B is any good for future growth or long-term aspirations of AGI. Investing 400B into people and (tech) manufacturing would be a solid long-term bet with benefits.
And investment and experiments by definition include the risk of failing. In almost everything lies a survivorship bias and no one talks about the 100+ car makers that went into goldrush mode 100+ years ago. This is life. Netflix vs Blockbuster - already forgotten?
Also the "fail rate" - so what part is failing and why? What's with the 5%? If we have a look at exponential functions this might be a really good deal, if the 5% can account for the losses. After all, benefits compound over time.
I witnessed first hand in FAANG some quota hires and I believe that now that no one gets paid for contrived and artificial business advantages, we are back to a more merits based evaluation of workers.
But AI should not be written off as fancy something with no impact. That's the wrong take. Whether it will be a springboard to new jobs that compensate for losses or replacements - I am not yet sure, but tent to be in the former group. ML engineers take care of ML - something new that takes care of something new.
We will see.
https://www.futuriom.com/articles/news/why-we-dont-believe-m...
https://www.cnbc.com/2025/11/04/white-collar-layoffs-ai-cost...
Given all those articles with AI generated images I bet that some artists lost their jobs
Over the last 20 years of tech, the giants have taken the smartest folks out there and put golden handcuffs on them. You could hire up all the smart folks and put them to work, or leave them out there and have them compete with you. With the launch of cloud providers and (expensive) dynamic scaling the problem only got worse. Think about hiring in the pandemic. Every one at home, with a stimulus check and nothing to do. Rather than a flurry of new software you got mass hiring.
But now we are in a capital intensive hardware cycle. Where in order to compete you need to have lots of $$$$ as well as software know how. It does not matter that there are smart people out there, without hefty backing they wont get very far.
I suspect that software is about to enter its "punk" era. We have software for small businesses that will help with accounting, HR, customer service, and cloud providers are starting to see some interesting competition. Much like the old punk poster showing you 3 chords and telling you to start a band it is entirely possible to find three friends and start a business that makes 1-10 mill a year with little effort and lower costs. The moment you stop thinking "unicorn" and start thinking "sustainable" the economics shift radically.
Started doing automation about 6 years ago. We had several offshore teams. The one caveat was that the company was vying for several government contracts. One of the requirements was the ability to pass a low level security clearance and on yeah, you cannot, under any circumstances off shore any of the work.
Therefore, they needed an onshore team to handle the government work.
Well, when things got slow, our team was essentially the clean up team. We had several hockey players on our team so we nicknamed our team "Zamboni". A large chunk of what we were doing was fixing the off shore teams coding and things would break so regularly two of devs became full-time off shore support.
The company was literally paying our off-shore team to build an automation. Then when it constantly break, they would pull us in to fix it.
Absolutely maddening they couldn't understand they were paying triple or quadruple for work we could do once properly instead of two or three times before we got it.
Because in any country with poor worker protections, the outcome is layoffs regardless.
AI succeeds? Layoffs of unneeded roles.
AI fails? Layoffs to cut expenditure to make up for the written-off expenditure.
In the UK, if AI makes a well-established employee redundant, the employee is entitled to redundancy pay. And if the company fucks up and overspends on chasing a ridiculous Macguffin they can't just fire people without making them formally redundant.
The damage that is going to be done in the USA if the AI bubble bursts is going to be generational.
They are however pretty well trained in getting people to quit by making their job/life hell, avoiding all redundancy processes and cost
AMZN for example overhired in various functions because it expected demand that never materialised. Admitting that is bad for the share price, but writing some woo about "AI and agility" will convince at least some investors to keep the faith.
Not a single mention of jobs being moved to India including tens of billions of investments in new offices.
Backing links from a quick search earlier this week when an obviously Indian HNer tried to deny this was really happening:
Microsoft announces US $3bn investment over two years in India https://news.microsoft.com/en-in/microsoft-announces-us-3bn-... (Jan 2025)
Google announces $15B investment in AI hub in India https://apnews.com/article/google-artificial-intelligence-vi... (3 weeks ago)
[Indian] ex-Accenture CTO named Google Cloud’s Chief Product https://www.hindustantimes.com/trending/us/who-is-karthik-na... (last week) (a lot of people speculate they named an Indian Accenture guy to move as much as possible to India)
Big Tech giants defy US-India trade tensions, record strongest 12-month headcount growth in India in 3 years https://www.moneycontrol.com/technology/big-tech-giants-defy... (September)
https://www.reuters.com/world/india/openai-launch-first-indi...
https://www.anthropic.com/news/expanding-global-operations-t...
and so on. Something very crazy is going on.
NOTE: I am not American and this doesn't affect me directly.
> Those expenditures may be approaching $1 trillion for 2025, while AI revenue—which would be used to pay for the use of AI infrastructure to run the software—will not exceed $30 billion this year
While it's clear that the author is summing up the spending from the big players, it's not clear to me that their math is right for revenue. Yes, OpenAI, Anthropic, Thinking Machines, SSI, etc. have pretty limited direct revenue (including zero!).
But this comparison assumes no revenue growth for other top computing users. Some companies are certainly saving money on some tasks and increasing revenue, particularly in fields like customer support. See the confusing figure in section 5 of https://hai.stanford.edu/ai-index/2025-ai-index-report/econo... .
That chart is by number of respondents and not weighted by revenue. Like the MIT study, it would not be surprising that "just pipe this to an LLM" isn't enough for most fields or companies. But a few have likely made material improvements.
10% of respondents saying they've seen a >10% revenue gain could be substantial, if they're bigger firms with high leverage in computing.
Edit to add: the comparison also makes a classic "GDP vs market cap" style mistake. Capital expenditure has multiple years of useful life. Revenue is annual. You'd want to compare depreciation vs revenue.
1. How will tangible assets generate profit net of near term capex requirements and interest on debt?
2. Why wouldn't payroll shrink as a result of the increased AI capabilities emerging from the capex spend?
3. If AI lives up to the hype & given recent news that public backstops are being requested, why shouldn't the US quasi-nationalize cash strapped players and distribute equity to every American?
4. As NVDA and AAPL local models and local compute eat into utility and base automation business, how do edge players maintain profitability without pricing capabilities well beyond the affordability of SMBs and individuals?
The spending-revenue gap is real. Hyperscalers are projected to spend $300-550B on AI infrastructure in 2025[1] while generative AI revenue won't exceed $30-40B [2]. Amazon's capex jumped from $48B in 2023 to $84B in 2024 to a projected $100B+ in 2025[3], that's capital intensity doubling from historical norms of 11-16% to over 22% [4].
But here's what the article misses: this isn't financial desperation. When Amazon's CEO announces 14,000 layoffs and explicitly states that AI will enable "fewer people doing some jobs"[5], he's revealing the strategic logic — show me the incentives and I'll show you the outcome. Companies aren't cutting jobs despite AI spending; they're cutting jobs because they know AI spending will pay off.
To be clear, the article treats the spending-revenue gap as evidence of irrationality. But infrastructure buildouts always precede revenue: railroads looked insane before they transformed commerce, electricity grids consumed massive capital before delivering returns, the internet required enormous infrastructure investment before creating trillion-dollar companies.
What's different now is companies are pulling the future forward. If we take this article at face value which I can appreciate is a BIG “if” then AI is already automating 25% of tasks and delivering 10-55% productivity gains[6] so they're not waiting for AI to replace jobs organically. They're cutting headcount now to fund the infrastructure that will make those cuts permanent.
More broadly, this is rational capital reallocation in a winner-take-all race. Companies that don't build AI infrastructure won't gradually decline, they'll lose competitive positioning entirely. That's why Meta is using off-balance-sheet financing for a $27B data center[7], why Oracle is borrowing $25B annually despite already carrying 450% debt-to-equity [8]. They're all-in because the alternative is obsolescence.
The real story isn't "spending causes cuts" it's that AI infrastructure commoditizes human expertise, the complement to compute infrastructure. Companies are trading labor costs for compute infrastructure because they've correctly identified compute as the new moat. The job cuts aren't the price of spending on AI; they're the business model shift that AI enables.
The article is right that we're not seeing mass AI job replacement yet. But the job cuts are happening in anticipation of replacement, not as an unfortunate side effect of spending. That's not desperation just business strategy.
-- 1.(Morgan Stanley: https://www.datacenterdynamics.com/en/news/morgan-stanley-hy...) 2. (Grand View Research: https://www.grandviewresearch.com/industry-analysis/generati...) 3. (CNBC: https://www.cnbc.com/2025/02/06/amazon-expects-to-spend-100-...) 4. (Cerno Capital: https://cernocapital.com/accounting-for-ai-financial-account...) 5. (CNBC: https://www.cnbc.com/2025/10&...
Likewise many tasks I used to do with Java, .NET, Node.js, have been replaced by low code/no code tools with agentic orchestration.
Thinking that AI isn't replacing jobs is wishful thinking.
I'd be interested in hearing more about the particulars of this.