Oracle Fired 30,000 People to Fund AI — The Stargate Bet and What It Reveals About Enterprise AI Strategy

Oracle just cut 30,000 people (18% of its workforce) and funnelled the savings straight into AI data centres. TD Cowen estimates the layoffs freed up $8 to $10 billion in annual cash flow, and Oracle's capex has already surged 162% to $55.7 billion. This is not about AI replacing a few admin roles. This is a Fortune 50 company reorganising its entire balance sheet around an AI bet.

The Stargate project, a joint venture with OpenAI and SoftBank, has a $500 billion price tag. Oracle's share of that includes a $300 billion cloud compute deal with OpenAI alone. The lesson for anyone building with no-code and AI tools: the companies supplying the infrastructure are betting the house. That tells you something about where the puck is going.

What actually happened at Oracle?

On March 31, 2026, Oracle laid off between 20,000 and 30,000 employees. The company's fiscal 2026 10-K, filed with the SEC on June 23, later confirmed headcount fell from 162,000 to roughly 141,000, a 13% decline. The company booked $1.8 billion in restructuring costs to make it happen.

Oracle didn't target failing divisions. These were healthy operations whose only sin was that they didn't map directly to AI infrastructure.

TD Cowen analysts had previewed this back in January, writing that cuts of this scale could generate $8 to $10 billion in incremental free cash flow. The cash would be redirected toward Oracle's AI capital expenditure, which had already ballooned 162% to $55.7 billion in the prior fiscal year.

A 162% capex surge in a single year, at a company Oracle's size, is the kind of thing you'd see during a wartime mobilisation. That's basically what this is. The enemy isn't a nation state; it's the cost of GPU clusters, power contracts, and data centre concrete.

What's the Stargate project?

Stargate is a joint venture between SoftBank, Oracle, and OpenAI, announced in January 2025 and formally structured as a new company. The stated ambition: spend up to $500 billion building AI infrastructure in the United States by 2029.

OpenAI signed a $300 billion cloud computing deal with Oracle in September 2025, a five-year contract starting in 2027 to secure compute capacity for training and inference. Separately, the Stargate partners announced five additional data centre sites in early 2026, adding 4.5 gigawatts of capacity. For context, 4.5 GW is roughly the output of four nuclear reactors.

Oracle's role: the compute landlord. The company builds and operates the infrastructure; OpenAI and others rent it. The $300 billion deal with OpenAI alone is one of the largest cloud contracts in history.

What this means for no-code and AI builders

First, the infrastructure bet is real. Oracle is a 47-year-old enterprise company with $53 billion in annual revenue, reorganising around the belief that AI compute demand will be the defining market of the next decade. When a company that conservative makes a move this aggressive, it confirms the direction of travel in a way no analyst report can.

Second, the talent displacement is already happening. For years, the no-code conversation has included a vague hand-wave about "AI replacing jobs." Oracle just made it concrete. 30,000 roles, gone in a quarter, at a single company.

Third, the tools you use are downstream of all this. When you spin up a Bolt.new app or build a Replit agent, you're consuming compute that ultimately traces back to infrastructure bets like Stargate. The pricing, availability, and capability of AI-assisted no-code tools over the next three years will be shaped directly by whether Oracle's $500 billion gamble pays off.

The takeaway

Oracle's 30,000 layoffs represent a capital reallocation from people to silicon, executed at the scale of a Fortune 50 balance sheet. The Stargate project is the reason. If you build with no-code tools, AI agents, or vibe-coded apps, you are operating in the world Oracle is betting $500 billion on. That world is arriving faster and more violently than most people expected.