Oracle (ORCL) is shelling out billions to effectively power the AI revolution.
However, inside the company, its employees are confronting a harder reality, where faster work does not automatically mean faster results.
Co-CEO Clay Magouyrk told staff that AI was accelerating coding without getting products into customers’ hands sooner, according to Business Insider. The bottlenecks simply moved elsewhere.
Investors are already restless. Oracle shares were down nearly 23% this year by Sept. 11, then lost another 1.8% the following week, closing at $147.61.
For employees navigating layoffs and an expensive infrastructure expansion, the stakes extend beyond learning another chatbot. Oracle’s internal rollout has attracted rapid adoption. But it also brought unexpected costs and exposed weaknesses in how work moves through the business.
Magouyrk’s message points to the next challenge, which involves turning individual productivity gains into measurable companywide results. That is where Oracle’s AI experiment gets considerably more complicated.
Oracle’s AI breakthrough exposes a bigger workplace problem
Oracle’s wake-up call is that faster employees do not automatically create a faster business. Magouyrk acknowledged that AI had accelerated coding without shortening the journey from development to customer delivery.
“When I think about where we were a year ago,” he told employees, “I don’t think we figured out how we make AI really that useful for ourselves,” according to Business Insider.
That’s a major admission from a company spending billions to support other businesses’ AI ambitions and involved in arguably the biggest tech buildout in history.
More Layoffs:
- Samsung cuts jobs as it shifts U.S. headquarters
- Another popular soda giant closes warehouse operation, cuts 184 jobs
- Meta layoffs take disturbing turn in new lawsuit
The internal adoption hurdle has since eased. CIO Jae Evans said Oracle reached 80% adoption within three months after rolling out ChatGPT Enterprise and Codex in April and May.
But adoption measures usage. It does not establish a return on investment.
“When you make the actual act of writing the code quicker, it doesn’t mean that suddenly everything is 1,000 times faster,” Magouyrk said.
Testing, validation, deployment, and release management now need to be redesigned to accommodate that faster output. For employees, the implication is significant: They’re producing more code, which can shift the workload onto colleagues responsible for checking and shipping it.
Oracle’s $664 billion backlog offers workers little protection
Oracle’s latest round of layoffs underscores the human cost of its AI expansion.
Despite a massive $664 billion backlog, as reported by TheStreet, Oracle began another round of cuts on Sept. 14. Employees received emails identifying that day as their last working day. U.S. severance offered four weeks of base pay plus one week per year of service, leaving newer workers with limited financial support. Oracle did not immediately comment.
California filings identify another 441 Bay Area job eliminations effective Nov. 13. The global headcount had already dropped from approximately 162,000 to 141,000 during fiscal 2026, a net reduction of 21,000 that includes departures beyond layoffs, the San Francisco Chronicle noted.
The financial pressure is substantial. Oracle reportedly carried roughly $129 billion in borrowings. Fiscal 2026 capital expenditure reached $55.7 billion, leaving free cash flow negative by $23.7 billion, as reported by CNBC.
Fiscal 2027 capital spending is projected at $90 billion to $95 billion. Backlog represents future contracted revenue; collecting it requires capacity that Oracle must finance and build first, as reported by TheStreet.
Meanwhile, Reuters reported that $18 billion in loans tied to its New Mexico data-center project reportedly traded at 89–91 cents on the dollar amid permitting and investor concerns. That project financing should not automatically be added to corporate debt.
For employees, the contradiction is painful: Oracle’s expanding order book coexists with shrinking teams. The AI opportunity promises years of growth, while affected workers face immediate income uncertainty.
Oracle’s cheap earnings multiple hides an expensive AI test
Oracle stock looks inexpensive, but its next leg higher depends a ton on execution.
Seeking Alpha’s data put shares at 18.13 times forward adjusted earnings, roughly 21% below the sector median and 18% below Oracle’s five-year average. Its forward PEG ratio of 0.60 strengthens the apparent bargain.
However, the earnings acceleration comes later. Analysts expect fiscal 2027 EPS of $8.14, up just 6.7%, on $90.46 billion in revenue. Fiscal 2028 EPS then jumps an estimated 35.1% to $11, bringing that year’s multiple down to 13.42.
That cheaper pricing setup depends on a substantial improvement that investors still need to see delivered.
The profitability snapshot exposes the tension. Oracle’s 34.27% operating margin comfortably exceeds its sector’s 8.02%. Yet capital expenditure equals 105.41% of revenue, while its levered free cash flow margin sits at negative 63.88%.
Meanwhile, gross margin is 63.96%, versus a five-year average of 73.03%, suggesting that investors should scrutinize the economics accompanying growth.
Magouyrk’s admission makes that scrutiny especially relevant, where internal AI adoption alone cannot establish financial returns.
Related: JPMorgan sends stark warning on $100 oil