Turning on their laptops early today, thousands of Oracle employees across the world got a terse e-mail telling them they had lost their jobs. The decision, it explained, was part of a “broader organizational change”.
Even going by large retrenchments in Big Tech of late, this was cold. Some employees who had spent decades in the company were told to go by the e-mail and then logged off from their systems within hours.
Oracle, notably, hasn’t even publicly confirmed the number of people laid off. Some news outlets estimate more than 10,000 are affected, with others reporting as many as 30,000 across the world.
By now, the mask has come off for Big Tech firms like Oracle. The retrenchments aren’t just about AI being so good that that it’s made people redundant. They have just been sacrificed to build more AI for a pay-off that has become increasingly costly.
Last month, Meta was reportedly readying a cut of 20 per cent or more of its staff to offset heavy AI spending. That saves the company US$6 billion, which it can add to its war chest to fight rivals like Anthropic, Google and OpenAI.
Why do these Big Tech firms seem desperate? Oracle had reported 22 per cent higher quarterly revenues in March, with quarterly net income totalling US$3.7 billion. Yet, the cloud and database company is carrying a debt of more than US$100 billion thanks to its AI ambitions.
The war in the Middle East this past month has fuelled fears of more uncertainty ahead, as Oracle keeps stacking up the risks of building out more infrastructure for AI.
It’s been in a somewhat “circular” US$1 trillion deal with chipmaker Nvidia and AI company OpenAI late last year.
Nvidia first agreed to invest US$100 billion in OpenAI in September 2025, which then inked a US$300 billion deal with Oracle to use its cloud computing infrastructure. And to build this, Oracle needs to buy US$40 billion of Nvidia chips.
The aim of all this? Be the first to get to artificial general intelligence (AGI), which theoretically could take on tasks like humans and self-learn through “common sense” over time.
Along the way, some humans would have to sacrificed, of course. For Oracle, cutting thousands of employees this round could free up cash flow of between US$8 billion and US$10 billion, according to some analysts.
Unfortunately, if you recall the huge sums of money in the US$1 trillion deal just months ago, the layoffs only bring back a small drop of cash in an ocean.
So, the question next is how long this carries on. Might the investments slow down at some point to more sustainable levels?
After all, more efficient models such as China’s DeepSeek are showing that throwing more Nvidia graphical processing units (GPUs) at a problem isn’t always the answer.
And at some point, questions on the pay-off will arrive, if they are not already being raised. For now, each time these Big Tech layoffs happen, the stock market rewards companies by lifting share prices.
What happens if these job cuts in the name of AI have a clearly negative impact for the long term? Or if AI doesn’t live up to its promises?
Consider the failings of late of some of the biggest AI proponents. Amazon Web Services (AWS) faced significant outages in recent months when its AI agent autonomously deleted and recreated parts of its cloud environment.
Look beyond AWS and you can understand how such incidents may not be one-off. If you consider how much code is AI generated today, a small amount of bad code in the codebase could snowball over time to create more problems. That technical debt needs paying off.
Then, there’s the story of McKinsey’s overly trusting AI. Earlier this month, the consultancy firm had its production database exposed by a “red team” of cybersecurity experts who exploited its internal chatbot in just two hours.
The security firm CodeWall found millions of conversations by employees and hundreds of thousands of files tied to its consulting work. Perhaps security and guardrails should have been tighter.
While previous technology hype cycles – from dot.coms to cloud computing – have involved enormous investments and risks, the scale of what’s in the AI boom today is unprecedented. What does the promised land look like?
Capital expenditures for AI infrastructure hit a staggering US$400 billion in 2025, with up to US$600 billion expected this year. Yet, the revenue from generative AI products and services was about US$50 billion last year.
That’s not to mention that almost all of the United States’ growth had been based on the AI boom last year. Consider, too, the devastating impact that data centres have on the environment. The costs are mounting but the reward for them is still unclear.
For those laid off, the toll is deeply personal. Cast aside in a cold calculus, their predicaments will stack up and rightly force questions about what the pay-off is.
The costs have been tremendous but what is at the end of the road? Sadly, few people can truly say they know.
