Oracle has cut nearly 21,000 jobs, roughly 21% of its workforce, while committing $70 billion in capital expenditure for fiscal year 2027 to build out AI service infrastructure. The company also plans to raise $40 billion through debt and equity to fund the buildout, according to Financial Express.
The Scale of the Bet
Combined, the $70 billion capex and $40 billion financing represent a $110 billion commitment to AI infrastructure, the largest capital allocation in Oracle’s history. The company has secured multi-year contracts for AI services that it says justify the spending level, though the conversion of order book into recognized revenue remains the core execution challenge.
The 21,000 job cuts are Oracle’s deepest headcount reduction in more than a decade. While the company has not disclosed which divisions absorbed the cuts, the restructuring follows a pattern seen across enterprise technology: legacy roles in sales, support, and on-premises infrastructure shrinking as capital shifts toward cloud and AI workloads.
Competing for Agent Workloads
Oracle’s pivot places it in direct competition with AWS, Microsoft Azure, and Google Cloud for the emerging market in AI agent infrastructure. Unlike the hyperscalers, Oracle’s historical strength lies in enterprise database and ERP customers, many of whom are evaluating where to run production agent workloads.
The bet is that these existing enterprise relationships convert into AI infrastructure contracts. Oracle has positioned its cloud regions and autonomous database products as optimized for agent workloads that require low-latency data access and transactional reliability. Whether that positioning holds against native cloud providers with larger developer ecosystems is an open question.
Execution Risk Is the Story
The $70 billion capex figure puts Oracle in the same spending tier as Microsoft and Google, companies with significantly larger revenue bases. Oracle reported approximately $53 billion in total revenue for FY2026. Spending $70 billion on capex in a single fiscal year while simultaneously cutting 21% of the workforce creates a narrow margin for error.
If multi-year AI contracts deliver as projected, the restructuring looks like a well-timed generational pivot. If enterprise AI adoption slows, if agent workloads consolidate on two or three hyperscaler platforms, or if the order book converts slower than expected, Oracle will have over-levered into a market it cannot dominate.
The market will get its first read on execution when Oracle reports Q1 FY2028 earnings later this year.