Standfirst
Oracle reduced its workforce by roughly 21,000 employees while preparing to spend about $70 billion expanding AI infrastructure. Across Silicon Valley, companies are making similar decisions. The layoffs and the spending spree look like separate stories. They are not. Together they reveal a fundamental shift in how the technology industry creates growth.
The Same Decision, Twice
When Oracle disclosed that its workforce had fallen from 162,000 to 141,000 employees over the previous year, the headline was familiar enough. Another technology company had announced a major round of job cuts.
The second announcement received far less attention.
Oracle also told investors it expected to spend around $70 billion on capital expenditure during the coming fiscal year, largely to expand the cloud infrastructure needed to support artificial intelligence. In its annual filing, the company acknowledged that broader deployment of AI technologies had contributed to workforce reductions alongside restructuring, acquisitions and other operational changes.
Most people saw two separate stories.
One was about layoffs.
The other was about AI investment.
In reality, Oracle was announcing a single strategic decision.
The company was shifting where it believed future growth would come from. Instead of directing its next wave of investment primarily toward expanding payroll, it was committing unprecedented sums to data centres, advanced processors, networking equipment and the electrical infrastructure required to run increasingly powerful AI models.
Oracle is simply the clearest example of a pattern that has quietly spread across the technology industry.
One Strategy, Not Two
Over the past two years, the world’s largest technology companies have followed remarkably similar playbooks.
Meta cut more than 20,000 jobs during its “Year of Efficiency” while substantially increasing AI investment. Microsoft has continued workforce reductions even as it accelerates spending on AI infrastructure. Alphabet has raised its 2026 capital expenditure guidance to between $175 billion and $185 billion, far above what analysts had expected, citing investment in AI computing capacity, data centres and networking. Amazon has likewise committed enormous resources to expanding the infrastructure behind its AI ambitions.
Taken individually, these decisions look like ordinary corporate restructuring.
Taken together, they point to something much larger.
Wall Street now expects the largest hyperscale technology companies to invest well over $700 billion in AI infrastructure during 2026, a figure that has continued to rise as companies increase their spending plans. Investors have become less interested in whether AI matters and more interested in whether these unprecedented investments will generate acceptable returns.
The layoffs have dominated headlines because they are immediate and visible.
The balance sheets tell the more important story.
Technology companies are changing what they believe creates competitive advantage.
When Software Became Infrastructure
For most of the internet era, software was celebrated because it scaled without requiring enormous physical investment.
Once an application had been built, serving another million users cost relatively little. The primary constraint on growth was talent. Companies competed aggressively to hire engineers because engineering talent was the scarce resource from which almost every competitive advantage flowed.
Hiring became strategy.
Payroll became investment.
Artificial intelligence changes that equation.
Unlike conventional software, frontier AI systems depend on vast amounts of physical infrastructure. Training and operating them requires advanced AI chips connected through specialised networking, supported by enormous amounts of memory, storage, cooling capacity and electricity. Researchers estimate that the computational cost of training frontier models has been rising at an extraordinary pace, with future frontier systems likely to require investments measured in the hundreds of millions, and eventually billions, of dollars.
The technology industry has a name for this collection of resources.
It calls it compute.
That word understates what has changed.
Compute is becoming the essential production input for artificial intelligence in much the same way that electricity became the essential production input for twentieth-century manufacturing. Like electricity before it, compute is becoming a general-purpose input that quietly determines which companies can scale, compete and innovate. Those with greater access to it can build larger models, serve more customers and improve products faster than rivals with weaker infrastructure.
The industry’s fundamental scarcity is shifting.
For decades, the defining question in boardrooms was simple:
Can we hire enough talented people?
Increasingly, another question is taking its place:
Can we secure enough computation before our competitors do?
Redirecting Capital
None of this means people have suddenly become less important.
The opposite is true.
The world’s leading AI researchers remain among the highest-paid employees in the industry because genuinely scarce expertise commands an even greater premium.
What has changed is where the next dollar of investment is going.
Imagine a software company that wants to increase output by 20%.
A decade ago, the obvious response would have been to hire more engineers.
Today, executives increasingly believe they can achieve the same objective by equipping existing teams with more capable AI systems and greater computational resources. If AI allows each engineer to become substantially more productive, then expanding infrastructure may produce higher long-term returns than expanding headcount.
Whether that expectation proves fully correct remains uncertain.
But investment decisions are based on expectations about the future, not perfect evidence from the present.
That is why layoffs and AI spending should not be viewed as separate developments.
They are two visible consequences of the same change in capital allocation.
The Market’s Biggest Bet
The scale of this shift explains why investors have become both enthusiastic and uneasy.
The largest technology companies are making one of the biggest capital allocation decisions in the history of the software industry before the full economic return has been demonstrated. The debate on Wall Street is no longer whether artificial intelligence will matter. It is whether today’s spending will ultimately justify its cost.
History offers reasons for both optimism and caution.
The telecommunications boom of the late 1990s ended with widespread accusations that companies had overbuilt fibre-optic networks. Many investors lost fortunes.
Yet the infrastructure itself did not become obsolete.
It became the foundation on which the modern internet was built.
That history is one reason executives remain willing to spend extraordinary sums today.
The question is no longer whether computation will become essential.
The question is whether today’s winners are building the right infrastructure at the right price.
Compute Becomes Strategic
The consequences of this shift are already extending far beyond corporate strategy.
When software companies competed primarily for engineers, the race was largely confined to labour markets. Universities produced graduates, recruiters competed for talent and companies built cultures designed to attract the best people.
The AI era depends on a different supply chain.
Advanced semiconductors, high-capacity data centres, reliable electricity, specialised networking equipment and large-scale financing have become strategic assets in their own right. Building a frontier AI system now requires resources that extend far beyond the software industry.
The United States has tightened export controls on advanced AI chips. Countries across Europe, the Middle East and Asia are competing to attract data centres through tax incentives, energy policy and streamlined permitting. Electric utilities are planning for demand growth driven not by factories or housing developments, but by clusters of AI servers operating around the clock.
Artificial intelligence is becoming an infrastructure business.
Silicon Valley is no longer simply building smarter software. It is building the infrastructure that will power the next generation of AI.
That makes this moment fundamentally different from every previous software revolution.
A successful social media platform could be built with talented engineers, venture capital and a compelling product. Frontier AI increasingly requires industrial-scale investment, resilient supply chains and access to energy measured in gigawatts rather than kilowatts.
The competitive landscape is changing accordingly.
Technology companies are no longer competing only with better software.
They are competing with stronger infrastructure.
Not Every Layoff Is an AI Story
It would be a mistake to attribute every technology layoff to artificial intelligence.
Many companies are still correcting for aggressive hiring during the pandemic. Others have reduced headcount in response to slower revenue growth, higher interest rates or changing strategic priorities. In many cases, restructuring reflects ordinary business cycles rather than technological disruption.
Recognising those factors strengthens rather than weakens the central argument.
The evidence does not suggest that AI has already replaced tens of thousands of workers. It suggests something more fundamental: as companies prepare for a future in which computational capacity becomes a decisive competitive advantage, a growing share of new investment is flowing into infrastructure rather than payroll.
The structural signal lies not in the layoffs themselves.
It lies in where the money is going.
The New Measure of Scale
For most of the software industry’s history, growth was measured by people.
Expanding companies hired larger engineering teams, opened new offices and built increasingly specialised organisations. Investors often interpreted rising headcount as evidence that demand was growing and new products were on the way.
Artificial intelligence is changing that relationship.
Future winners will still need extraordinary people.
But they will also require extraordinary computational capacity.
Access to advanced chips, high-performance networking, data centres and reliable energy may become just as important as attracting world-class engineers. Competitive advantage is increasingly being defined by a company’s ability to finance, build and operate the physical systems that make advanced AI possible.
That is a different model of scale.
It also explains why investors have become increasingly focused on capital expenditure. The question is no longer simply whether companies are investing enough in AI. It is whether they are investing wisely enough to generate durable returns from an infrastructure base that will cost hundreds of billions of dollars to build and maintain.
The market’s attention has quietly shifted from software margins to infrastructure economics.
That may prove to be one of the defining financial transitions of the AI era.
What Winning Looks Like Now
Public debate has largely centred on a familiar question:
Will artificial intelligence replace human workers?
It is an important question.
It may not be the first one businesses need to answer.
The first transformation is already taking place inside corporate balance sheets.
For decades, the software industry’s default response to growth was straightforward.
Hire more people.
Today, the first instinct is increasingly different.
Build more compute.
That does not mean people have become less valuable.
It means the next dollar of investment is increasingly being directed somewhere else.
For more than three decades, software companies competed to attract the world’s best engineers.
The next decade may reward something different.
For the first time in the history of the software industry, the companies shaping the future may not be those that employ the most people. They may be those that build, own or can reliably access the greatest amount of computation.
