Oracle is planning to slash up to 30,000 jobs to fund AI data centres

[rcjordan]

Oracle Layoffs: Tech giant to slash 30,000 jobs as banks pull out from financing AI data centres |

https://www.livemint.com/companies/news/oracle-layoffs-tech-giant-to-slash-30-000-jobs-as-banks-pull-out-from-financing-ai-data-centres-11769996619410.html

[ergophobe]

Scenario 2b - AI can't do the jobs of the 30,000 Oracle employees, but data center investments can suck up all the money that would have paid them.

In terms of your life as an Oracle employee this sucks.

But in terms of the AI Bubble Pop, this is probably a shot across the bow that will take some heat out of the system.

 - Oracle stock has been on a 6-month slide, so now people know you can't just add "AI" to every initiative and watch stocks rise

 - banks pulling out tells people that there isn't infinite money for exuberance

 - as data-center construction slows, that takes a lot of the heat out of the AI bubble

Something feels weird about this economy
Noah Smith
https://www.noahpinion.blog/p/something-feels-weird-about-this

TL;DR
 - traditional numbers look good: inflation, employment, growth

 - productivity is really high - if it continues at this rate, it will be the highest since WWII based on most recent numbers. Productivity boomed in 2023 and 2024. It slowed down in 2025, but is still high by recent standards

 - job growth is terrible.

The jobs one is interesting

 - prime-age employment is higher than it has been in a long time
 - prime-age workforce participation is higher than it has been in a long time
 - unemployment has bumped up, but it appears that is because more people are actually looking, rather than sitting on the sidelines (remember, "unemployment" is a measure of the number of people *looking* for jobs, not the number who do not have jobs).

So much for the "nobody wants to work these days" narrative.

BUT (yes, all-caps BUT), why?

Is it because AI has goosed productivity?

It appears that it has improved productivity in manufacturing due to better automation, though no effect has yet been seen due to white-collar workers sitting around chatting with Claude.

BUT, BUT... the main driver has been data center construction. This partly because data centers have high value and get built pretty fast, but it is also a function of the fact that once built, you have to fill them with high-value computers. The ratio of work hours to final value is very high, which pulls productivity numbers up for the entire economy.

While the money lasts. And based on this article, that moment may be over.

The good scenario is that this takes some heat out of the AI bubble, valuations start to slowly fall as Oracle's have been doing for six months. Companies that made better fiscal decisions than Oracle mostly hang  onto workers. All those data centers get put to work and we have a meaningful but no catastrophic fall in productivity as data center construction slows, but bit by bit that gets pulled back up and companies figure out how to implement AI.

The bad scenario is left as an exercise for the student. The teacher's manual doesn't give a solution, it just has a note that says, "See threads from last 12 months"

SIDE NOTE: the fact that bank money is drying up for data center construction could also be a ray of hope for, ahem, someone who is faced with a data center going up within spitting distance of her house.

[rcjordan]

" as banks pull out from financing AI data centres"

That's the part that caught Debbie's eye.  She believes this is the needle that'll pop the bubble.

[littleman]

I am finding it hard to wrap my head around what's happening with AI, it is simultaneously projected to devastate employment and collapse in a enormous bubble that will take down the economy.

I suppose it is a bit like the dot-com bust of the early century.  There was an initial over-hype and expectation, but now there is no denying that the internet has fundamentally transformed the way commerce works globally.  Everyone knew it would harbor massive change, but no one knew exactly how it would shake out and the money flowed to what were often dead ends.

[ergophobe]

>>the money flowed to what were often dead ends

In retrospect, people seem to think the big win from the internet bubble was the building out of way too much optical fiber capacity, while most of the money that flowed into actual websites went poof. When the collapse came, e-greeting cards (was this the stupid idea ever?) disappeared, but the fibre was still in the ground and that meant cheap bandwidth that made way for the internet that we know with streaming movies and always-on surveillance.

I suppose the analogy, and a thing I hear mentioned from time to time, is that the companies that build the data centers might go belly up, but the fallout will mean a lot of cheap compute that can be leveraged by survivors and startups.

The counter to *that* assertion is that the processors have short life cycles, unlike the fiber.

The counter to the counter is that 1) the data centers themselves are major infrastructure projects, especially since more and more they have to build their own power plant and 2) just because a processor is not good for cutting edge work doesn't mean that massive compute capacity can't be turned to things like running lighter but perfectly adequate models, even if training the new models will require the best chips.
>>  banks pull out

I had heard money was drying up, but that's the first I heard that it had truly dried up for Oracle. Previously, the fact that Oracle had been forced to bank financing rather than cash or being able to find direct investors, was taken as a sign that the bubble was deflating.

>>  wrap my head around

Agreed. I started a document putting down my thoughts. I think it grew to 15 pages over a couple of months (some repetition because I was rapid firing). But never really felt like I was more clear on it than when I started.

I like to use electricity as an analogy in my thinking, rather than the internet. And the two lessons I take from that are

 - implementation of a fundamental technology takes a long time, so I think job destruction will be slower than the maximalists say. Just because it is technically possible for an AI to do a given job 18 months from now doesn't mean that it is practically sensible.

 - the results are so far-reaching that even 10 years into the revolution, it's impossible to imagine what the result will be 100 years into it. Who in Edison's lab predicted spreadsheets, GPS and ChatGPT?

[Brad]

Related:

Iranian strikes on Amazon data centers highlights industry's vulnerability to physical disasters

https://www.washingtontimes.com/news/2026/mar/3/iranian-strikes-amazon-data-centers-highlights-industrys/

These were AWS data centers not AI but the vulnerabilities still apply. Sabatours with drones are hard to defend against.

[Brad]

>as banks pull out from financing AI data centres

Inside the secret meeting that led to the AI political resistance | The Verge

The Pro-Human Declaration has been signed by the American Federation of Teachers, the Congress of Christian Leaders, the Progressive Democrats of America, and Steve Bannon.

[ergophobe]

>  vulnerability

Day One of a war with NATO, Russia cuts every undersea cable they can get to

[Brad]

>  vulnerability

Day One of a war with NATO, Russia cuts every undersea cable they can get to

True that, and China has been watching the Russian experiments in cable cutting.

[grnidone]

Fernando de Leon, founder of Leon Capital Group, says he sees big problems in data center financing.

"... I see large technology companies, the largest companies on the planet, with $4 trillion market cap, saying, 'I don't want to own this asset. I don't want to have this on my balance sheet.' So I ask, Why? Why doesn't the largest company in the world want to own its own asset?" De Leon said.

...

De Leon surmises that what is inside these data centers, the technology of artificial intelligence, will quickly become obsolete. AI, after all, is designed to make everything more efficient, including itself. And the value of the centers is not the four walls, but what's inside. "

https://www.cnbc.com/amp/2025/12/17/billionaire-real-estate-developer-red-flag-data-centers.html