Oracle Plans Thousands of Job Cuts in Face of AI Cash Crunch

Welcome to this deep dive. If you're joining us today, you are stepping right into the middle of

honestly one of the most jarring high stakes moments in modern business history.

Well, absolutely. It's a defining moment for the decade, really.

Yeah. And we're setting the scene here. It is the spring of 2026. And the tech world is grappling

with this massive, almost impossible paradox. Right. We're focusing today on Oracle Corporation.

And you know, they're writing this unprecedented high explosive cloud demand record

start prices executives taking victory lap. I think a lot of victory. Exactly. But at the exact same

time, they are planning to cut thousands of jobs practically overnight, which is I mean,

what's fascinating here is the sheer irony embedded in that strategy. Oracle is sprinting at this

breakneck speed to commoditize and sell the exact same automation technology that is currently

forcing these incredibly painful efficiency measures inside its own walls. It's a wild. It

really is. It's a story of a self-inflicted cash crunch, you know, born entirely from their

ambition to rule the next decade of computing. Okay. Let's unpack this because we have a massive stack

of source material to get through today. We do. We're pulling from this deeply reported nearly 5,000

word investigative piece that just dropped. It details. Oracle's high stakes AI pivot. And we're

also looking at their financial earnings reports, right? Right. The latest earnings plus some

a pretty panicked analyst commentary from Wall Street. Highly panicked in some quarters. Yeah,

very. And we're cross referencing all of that with firsthand on the ground insider accounts,

people who are actually walking the halls of Oracle right now, which really grounds it in reality.

Exactly. So our mission today is to truly understand the brutal math and honestly the human cost

of this 2026 AI infrastructure race. We want to know what Oracle's massive gamble means for

the future of work. For you, for us, for everyone, because to really understand the scale of

Oracle's ambition here, we first have to look at the immediate crisis unfolding inside their

campuses right now. Right. In early March, 2026. Yeah. I mean, the environment inside the company

has shifted at a speed that is, well, giving people whiplash. Totally. Just to give some

context, as of mid 2025, Oracle had roughly 162,000 employees, a massive sprawling global workforce.

Right. And the source material details how these kind of ambient low level whispers turned into

confirmed structural anxiety just a few weeks ago. An anxiety almost undersells it. This isn't

your standard corporate anxiety about, you know, missing a quarterly sales target. No, not at all.

This is the anxiety of obsolescence. Oracle is planning to eliminate thousands of jobs across

multiple divisions. And what makes this so jarring is that these aren't standard rolling reductions.

Right. They aren't just trimming the bottom 5% of underperformers. Exactly. They are surgically

targeting roles that they believe AI will soon render entirely obsolete, which is terrifying if

you're in one of those roles. And the financial scale of this restructuring is, I mean, it's genuinely

staggering. It is. The sources show the company has set aside up to $2.1 billion. Billion

dollars would be just a massive number. Yeah. $2.1 billion, specifically for severance and

related costs for the fiscal year 2026. And they've already recognized about $826 million of that.

Right. And we need to contextualize that $2.1 billion figure because in corporate accounting,

that is a massive red flag regarding the sheer volume of human capital being injected.

How so? Like, break that down for us. Well, a standard severance package and enterprise tech

might cost a company anywhere from say 30 to $80,000, depending on 10 year and location. Okay.

So when you are provisioning over $2 billion, specifically to pay people to leave,

you are completely reshaping the foundational structure of the company.

Oh, wow. So it's not a trim. It's an amputation. Exactly. It indicates a pivot so sharp

that the existing workforce is no longer viewed as an asset to be retrained.

There are liability to be cleared out to free up cash. You know, when I was reading through

the source material, I couldn't help but think of it like a snake eating its own tail.

That's a great way to put it. You're building this ultimate automation engine, right?

But just to afford the parts to build it, you actually have to feed your own workforce into it.

Yeah. It's a profound tension. And that tension is precisely what the investigative report labels

the AI paradox. The AI paradox. Right. You have this technology driving absolutely explosive,

record-breaking demand for cloud services. The theoretical wealth is astronomical.

But simultaneously building the infrastructure to capture that wealth is demanding such

ruthless internal efficiency that they're hollowing themselves out.

Because Chairman Larry Ellison basically bet the company's entire future on this, didn't he?

He did. He bet everything on becoming the premier cloud provider for AI workloads.

And that required an aggressive infrastructure build out.

Which results in this massive cash crunch?

Exactly. A localized cash crunch inside one of the wealthiest companies on earth.

It's crazy. And you know, if you're listening to this right now and you've ever felt the pressure

of adopting a new tool at work, it's like something that makes you faster but also makes your

specific skills a little less necessary. That pit and your stomach.

Right. If you felt that, you are experiencing a microcosm of what Oracle is going through

right now at a macroeconomic scale. It's the anxiety of the 21st century knowledge worker

amplified to the corporate scale. Absolutely. But okay, let's pivot here a second.

Because why is Oracle a company famously known for legacy databases, right?

The boring stuff. Yeah, the boring stuff. Why are they the ones making this massive infrastructure play?

To understand their current financial strain, we really have to trace the pivot that saved

them from obsolescence in the first place. We do. We have to look at their history

because it's vital context. So the sources remind us. Oracle was founded back in 1977 by Larry

Ellison, Bob Meiner, and Ed Oats. Originally called software development laboratories.

Interestingly enough. Right. And they completely dominated the 80s and 90s with the first commercial

sequel relational database. Which was revolutionary at the time. It was the nervous system of the fortune 500.

Yeah, it replaced physical ledgers. And they grew to over $50 billion in revenue through the

2000s, buying up companies like PeopleSoft, Seabull, Sun Microsystems, massive acquisitions.

But then the mid 2010s arrived. And that was an existential threat for them, right?

A huge one. AWS, Microsoft, Azure, Google Cloud. They started aggressively siphoning off Oracle's

traditional customers. Because nobody wanted a massive physical server in their basement anymore.

Exactly. Yeah. Amazon made computing a utility. And Oracle was losing ground fast.

So what did they do? In 2016, they pivoted hard. They launched Oracle Cloud Infrastructure or OCI.

OCI, right? And this was designed as a second generation cloud built entirely from scratch for

heavy duty enterprise workloads. And the text emphasizes two things about OCI, bare metal

performance and low latency networking. Yes, those are the key terms. But I have to push back

here for a second. Ask a dumb question. Go for it. For a long time, Oracle was viewed as this boring

legacy tech company for like municipal banks and governments. How did building a cloud for them

suddenly make Oracle the cool kid in the AI hardware space? It's a great question. And the

answer is that the boring legacy trust is exactly their superpower now. Really? Yeah. Let's break

down that bare metal term in a standard cloud like early AWS. Amazon puts a layer of software

called a hypervisor between you and the physical hardware. Okay. It lets them chop up one server

and rent it to 10 different startups. Very efficient. Sure. Make sense for them. But that middleman

software adds a tiny microscopic delay to every computation. It adds latency. So it's like having

a translator in the room during a conversation. It just slows things down slightly. That's a perfect

analogy. Now, for hosting a website, you don't care. But Oracle was building a cloud for banks

running high frequency trading. They needed zero delay. Oh, okay. So OCI removed the translator.

They offer bare metal services where the software talks directly to the hardware. No middleman.

Exactly. And secure enough for highly regulated industries. So how does that connect to AI? Well,

jump forward to the 2024 2025 inflection point. Generative AI explodes. And suddenly,

data centers are the new oil fields as the text calls them. Because training these huge AI models

takes petawatts of power, right? Power and massive parallel GPU clusters. Generative AI doesn't

think sequentially. It thinks in parallel. Right. Tens of thousands of GPUs talking to each other

in microseconds. If you have that hypervisor translator in the middle, the delay compounds across

billions of calculations. Oh, wow. So it was a ton of time and money. Exactly. So OCI's low latency

bare metal networking built for boring banks was perfectly almost accidentally primed for the AI

boom. That is fascinating. It gave them the perfect product. It really did. Which brings us to the

astonishing sales numbers they've recently posted. Because we need to look at the actual receipts

of this AI gold rush. The numbers are mind boggling. Yeah. The sources outline how Larry Ellison

personally courted the biggest AI players. We're talking multi-billion dollar commitments.

Right. Massive landmark deals. There's the open AI deal. Reportedly a $30 billion cloud services

agreement plus a $300 billion framework for future compute. 300 billion. I know. And similar

packs with XAI met a tick-tock Nvidia AMD. The biggest names in tech. Yeah. And if you look at

their Q3 fiscal year 2026 results, which just ended in February 2026, their total revenue hit

17.2 billion dollars up 22%. Tron hours. Cloud revenue was 8.9 billion dollars up 44%. Infrastructure

as a service or IS hit 4.9 billion dollars up 84%. And then there's the AI specific number. Yes.

AI infrastructure within IS was up an incredible 243%. 243% that is explosive. They delivered over 400

megawatts of capacity in just that quarter, keeping gross margins on AI capacity at 32%. But here's

where we have to look deeper. The real story is the remaining performance obligations or RPO.

Yes. The RPO. The sources say that exploded to $553 billion up 325% year over year. Right.

And here's where it gets really interesting that $300 billion framework with open AI. That's not

just a contract. That is the GDP of a small country dedicated just to computing power. It is. But

we really need to break down what an RPO actually is for you listening. Please do because it's tricky.

Remaining performance obligations are basically future contracted revenue. It's a legally binding

promise from a client saying we will buy this much compute from you over the next 5 to 10 years.

So it's a promise. Exactly. All street initially love these massive contracted numbers. The stock

ran up. But an RPO is just a promise on paper. Right. To fulfill that promise to turn that paper

into actual cash, Oracle has to actually build the physical infrastructure first. They have to

build the data center. Yes. If we connect this to the bigger picture, this immense backlog of

demand is exactly what triggers the financial prices we discuss at the start. Right. So Oracle has

$553 billion in contracted future revenue. But they are broke enough to fire thousands of people.

Because realizing that revenue requires moving from digital promises to physical concrete,

steel and electricity, which brings us to the liquidity trap and the physical bottlenecks.

Because the FY 2026 capital expenditure, the CapEx guidance, is an astonishing $50 billion.

$50 billion in one year. That's triple their earlier levels. Triple. And to afford that,

Oracle announced plans to raise $45 to $50 billion through debt and equity. Which is wild when

you consider they already carry over $100 billion in existing debt. Over $100 billion in debt already.

And because of this CapEx, analysts project negative free cash flow to persist until around 2030.

Yeah, negative free cash flow for years. Because of depreciation, construction timelines,

power procurement, it's a massive drain. It's like, okay, think of it this way. It's like winning

a massive catering contract for the Olympics, right? Okay. I like this analogy. But then you realize

you have to build five commercial kitchens from scratch, buy all the ovens and pay for all the

electricity before you can serve a single meal and actually get paid. That is exactly what it is.

And building those kitchens or data centers involves incredibly harsh physical realities.

What kind of realities? Well, the source is detailed that Oracle has secured over 10 gigawatts

of power for the next three years. Major projects in Texas, New Mexico, Wisconsin, Michigan.

Right. But the physical constraints are brutal. There are global supply chain issues

with high voltage transformers, GPUs, specialized cooling equipment.

You can't just order the stuff on Amazon Prime. No, there's a multi-year wait list for transformers

alone. And then there's the Avaline Texas expansion. Oh, yeah, the text mentions that. It was recently

scrapped, right? Completely scrapped. Uh-huh. Due to stalled financing and scope negotiations.

What does that actually mean? It means the local power grid couldn't handle it. The utility likely

said you want a gigawater power. You need to build us a new substation, which costs hundreds of

millions of dollars upfront. And banks are growing very cautious about financing these projects.

Data centers cause billions require years of permitting and rely on massive utility upgrades.

Plus, you need what 80 to 90% utilization just to justify the profit margin.

Exactly. If the servers are idle, you're bleeding money. And contrast Oracle's balance sheet

with the hyperscalers like AWS, Google, and Microsoft. It's a different game for them.

Right. Because Amazon has retail, Google has search ads, Microsoft has Windows. They have

deep pockets to absorb this capital intensive warfare. But Oracle doesn't have a consumer cash

cow like that. They rely on enterprise software. Right. So when the banks get nervous about your

$100 billion debt and your $50 billion cap ex bill, you have to find cash elsewhere.

You do fast. And tragically, the easiest place to find cash is in the payroll of the people

who built the company, which brings us to the human cost automation inside the automator.

Yeah. The scope of these layoffs is sobering. The investment bank TD Cowan initially floated

an estimate of 20,000 to 30,000 layoffs, which is 12 to 18% of their entire workforce.

Just to free up 8 to 10 billion in cash flow, Bloomberg later crystallized it to simply thousands.

But it's the targeted roles that are so fascinating. This is the crux of it. They are just cutting

middle management. Right. They are targeting routine database operations, basic cloud provisioning

administrative functions roles that Oracle explicitly expects to be quote, backfilled with AI.

They are using tools like their own AI agent studio and embedded intelligence in their ERP and

HCM suites to replace their own workers. They're testing their products on their own org chart.

It's brutal. And the human element of this is just it's playing out in real time. The text mentions

reddit's employees, awful call subreddit. Yeah. That part was striking. People are tracking real time

headcount drops, like watching the company directory fall from 166,300 in a matter of hours.

Just refreshing the page and watching your colleagues disappear. Exactly. And it's hitting everywhere.

Redwood City, Austin, India, Romania, Ireland. No one is immune because the data center math is

fundamentally changing the labor model. How so? Think about it. It takes tens of thousands of skilled

trace people to build a gigawatt scale data center, pouring concrete, laying wire, installing cooling.

A massive jobs program. Right. But once it's built, it only takes maybe a thousand permanent

workers to run it. The software does the rest. Wow. You know, I have a question about that. There's

a quote in the source material from CEO, Safra Katz and Larry Ellison. Okay. They talk about AI

enabling the company to build more software and less time with fewer people. Standard corporate

messaging. Right. But I have to ask you, is this just ruthless corporate cost cutting

disguised as technological progress? Or is it truly an inevitable evolution of how tech companies

must operate? This raises an important question. And it's one about the psychological toll on

white collar professionals. Yeah. These workers joined Oracle's first ability. They aren't

to start up risk takers. They joined a 40-year-old giant. Right. And now they're facing the exact

same disruption they once helped inflict on legacy industries, like manufacturing or retail.

It's deeply ironic. Oracle is building the infrastructure that powers the AI models that are now

capable of coding and doing customer support. Effectively rendering their own employees redundant.

To answer your question. It's both. It's cost cutting because they desperately need the cash.

But it's also inevitable. If they don't do it, Microsoft or Google will and Oracle will lose

the margins. Wow. So Oracle situation isn't happening in a vacuum. Not at all. It is just the

loudest most concentrated example of a tectonic shift happening across the entire economy right now

in 2026. The ripple effect is massive. Let's look at the industry-wide stats from the piece.

Over 70 companies have already announced roughly 40,000 layoffs in 2026 alone.

And they are specifically citing AI efficiencies as the reason. Right. And at the same time,

the hyperscalers collectively plan nearly $700 billion in data center spending this year alone.

$700 billion. It's an unprecedented reallocation of capital.

TD Cowen calls this the battle between the productivity flywheel and the short-term cash flow vortex.

I love that phrasing. It's great. The future outlook is technically great, right? Oracle's

FY 2026 revenue guidance is $67 billion. With projections of high teens to low 20% revenue growth

through 2027 and beyond. But the risks in that cash flow vortex are huge execution delays,

power shortages. Regulatory pushback on energy is becoming a huge factor.

Local governments are realizing these data centers drink entire lakes of water for cooling.

Oh wow. Yeah. And then there are the geopolitical tensions. Right. The supply chains,

Nvidia and TSMC. All of this relies on advanced GPUs manufactured primarily by TSMC in Taiwan.

If that supply chain is disrupted by geopolitical conflict, that $553 billion RPO means nothing.

You can't build the servers without the chips. So what does this all mean? How does Oracle actually

survive this? Will they become a top tier hyperscaler alongside AWS and Azure? Or is this an AI bubble

waiting to pop if energy costs spiral? Well, industry veterans often draw historical

carolals here. Look at the fiber optic boom of the late 90s. Okay. The dot com era. Right.

Telecom companies took on massive debt to lay millions of miles of fiber optic cable. They overbuilt

many with bankrupt. Right. It was a huge bubble. But the fiber stayed in the ground.

And that dark fiber eventually enabled the smartphone revolution and the modern cloud

in the 2010s. We saw a similar thing with the smartphone bill out itself.

So you're saying the $700 billion CapEx boom is basically laying the dark fiber for the AI era.

Exactly. Massive capital deployment today for productivity gains tomorrow.

The infrastructure will survive even if some of the companies don't.

But what about the workers? That's where labor economists are sending the alarm.

New jobs will absolutely emerge. Things like AI ethics, prompt engineering, model governance.

Sure. There will be new roles. But the transition is uneven. You can't just take a database

tuner who just got laid off and make them an AI ethicist tomorrow. Mid-career engineers will

struggle the most to adapt. Yeah. That makes sense. People caught in the middle.

Well, to summarize this deep dive. The story of Oracle in 2026 is a microcosm of creative

destruction on steroids. It truly is. Larry Ellison is risking the legacy of a four-year-old

Titan to build the electricity of the future. And it's being funded by the harsh recalibration

of his current workforce. And as we always say, knowledge is most valuable when applied.

For you listening, remember that while data centers hum with electricity,

the human recalibration we are seeing at Oracle will linger. It's going to spread across the entire

industry. It really is. And I want to leave you with one final provocative thought.

We just spent this time talking about how Oracle is replacing its human cloud

provisioners and database tuners with the very AI those employees helped build.

Right. The automated being automated. Exactly. But as hyperscalers pour 700 billion dollars

into massive data centers this year requiring thousands of physical construction workers,

how long will it be until the AI itself is tasked with designing, optimizing, and fully managing

the next generation of physical infrastructure? It's the logical next step.

If the automator is currently being automated in software, what happens when the infrastructure

learns to physically build itself? Fascinating and terrifying question.

Thank you so much for joining us for this deep dive. We'll catch you next time.