Oracle Cuts Jobs While Expanding Data Centers, OpenAI Slows Frontier AI Development: Is the AI Spending Boom Losing Momentum?

TradingKey - Anthropic calls for controlling the pace of model capability growth, OpenAI has slowed down certain frontier AI projects, and Elon Musk also supports related calls. Oracle (ORCL) initiated a new round of layoffs during the same period, but continues to increase AI data center investment.

However, judging from Anthropic's long-term computing power agreements and the capital expenditure plans of Microsoft (MSFT), Amazon (AMZN), Alphabet (GOOGL), and Meta (META), major tech companies' AI infrastructure investments continue to expand.

With R&D pacing and safety requirements tightening on one side, and computing power and data center investments continuing to ramp up on the other, is the AI spending boom about to cool down?

Anthropic and OpenAI Adjust Frontier AI Development Pace

On September 12, Anthropic CEO Dario Amodei published "We Must Pace the Frontier," advocating for controlling the pace of capability advancements in frontier models to allow time for safety testing and risk assessment.

Amodei proposed that third-party evaluation teams should be granted access levels close to those of internal employees, frontier AI companies should establish shared safety standards, and governments worldwide should strengthen coordination. In addition to these proposals, Anthropic has committed to introducing an external evaluation mechanism.

OpenAI made similar statements. CEO Sam Altman supports unified frontier AI safety requirements and independent evaluations. President Greg Brockman subsequently disclosed that the company has slowed down some frontier projects and introduced safety monitoring as early as the model training phase.

Previously, OpenAI had paused some projects and temporarily reassigned about 25% of its engineers to troubleshoot and fix vulnerabilities in its internal infrastructure and security architecture; the company also used Astra to inspect internal systems until Astra no longer detected new top-priority vulnerabilities.

Elon Musk publicly supported Amodei's position. However, his xAI has not publicly announced any plans to adjust its model training, compute procurement, or infrastructure construction.

Notably, tightening at the safety level has not affected Anthropic's long-term compute stockpiling. According to data compiled by The Information, cloud service, chip, and data center agreements signed by Anthropic over the past 11 months have an estimated contract-term value of up to approximately $517 billion, corresponding to at least 14.8 gigawatts of contracted and planned capacity.

On the financial side, according to a September 14 report by the Financial Times, Anthropic's annualized revenue as of the end of July this year exceeded $65 billion, compared to approximately $9 billion by the end of 2025. The Financial Times also reported that the company expects its adjusted operating profit, excluding stock-based compensation, to remain positive for a second consecutive quarter.

According to a September 11 report by Reuters, Nvidia (NVDA) is considering participating in Anthropic's potential IPO with up to $10 billion as an anchor investor, with relevant plans still under discussion.

Oracle Cuts Jobs as Data Center Investment Continues to Increase

Meanwhile, according to a report by Business Insider, Oracle launched a new round of layoffs on September 14, though the exact number of employees affected has not been disclosed. As of the end of May 2026, the company had approximately 141,000 global employees; affected by layoffs and natural attrition, the workforce decreased by about 21,000 within a year, a decline of roughly 13%.

Financial results show that Oracle's capital expenditures reached approximately $28.5 billion in the first quarter of fiscal 2027, compared with $8.5 billion in the same period last year, while free cash flow was recorded at negative $5 billion. The company expects full-year capital expenditures to reach $90 billion to $95 billion, primarily directed toward expanding cloud computing and AI data centers.

In the first fiscal quarter, Oracle's cloud infrastructure revenue increased 121% year-over-year to $7.4 billion, remaining performance obligations rose to $664 billion, and new AI cloud contracts exceeded $30 billion.

Laying off workers while expanding data centers, Oracle's strategy is to use the cost savings from layoffs to support massive capital expenditures, while accelerating the revenue conversion of AI contracts to cover depreciation, financing, and operating costs.

Capital Expenditures of Four Major Tech Companies Remain High

Microsoft's capital expenditures for fiscal year 2026 (ending at the end of June) were approximately $115.9 billion, compared with about $64.6 billion in the previous fiscal year. During the same period, fourth-quarter revenue from Azure and other cloud services grew 43% year-over-year.

Amazon raised its capital expenditure plan for calendar year 2026 to $220 billion, targeting businesses such as AWS, AI chips, logistics, robotics, and satellites. In the second quarter, AWS revenue grew 37% year-over-year to $42.2 billion.

Alphabet raised its 2026 capital expenditure guidance to $195 billion to $205 billion, primarily targeting servers, data centers, and network infrastructure.

Meta expects its 2026 capital expenditures to be $130 billion to $145 billion, with the lower bound of its guidance raised from $125 billion to $130 billion.

As of September 15, none of the four companies had lowered their latest capital expenditure guidance. Against the backdrop of growing skepticism over AI returns, tech giants have nonetheless chosen to double down.

Is the AI Spending Boom Losing Momentum?

To determine whether AI investment is cooling off, the key is to look at whether cloud providers are cutting capital expenditures and whether orders for GPUs, HBM, and data centers are weakening. Currently, none of the four major tech companies have lowered their capital expenditure guidance due to the recent safety debate.

OpenAI's adjustments are focused on certain high-risk frontier projects and internal safety rectifications, while Oracle continues to expand its data center investments even as it cuts headcount. Judging by these indicators, there are currently no signs of a broad-based retrenchment.

AI spending by major cloud providers continues to expand, but standards for capital deployment are rising. Safety assurance, financing costs, project returns, and the pace of revenue conversion will become key evaluation priorities in the next phase.

Moving forward, several signals warrant close monitoring: whether cloud providers' capital expenditure guidance shifts, the training pace of OpenAI's frontier projects, the revenue conversion progress of Oracle's AI contracts, and changes in orders for Nvidia GPUs and HBM.