Oracle shares plunged 12% at Thursday's open, marking their largest single-day drop since last December. The earnings report released after the previous day's close showed its cloud infrastructure business (OCI) growth accelerating further, AI-related orders hitting new highs, and remaining performance obligations (RPO) surpassing $638 billion. However, following the report, the stock price turned from gains to losses, falling sharply in after-hours trading, a stark contrast to the positive data mentioned.
According to a research report from Bernstein published on June 11, Oracle's cloud business performance this quarter was mixed: overall cloud revenue was slightly below expectations, primarily dragged down by extended procurement cycles for SaaS (Software as a Service) business. However, OCI (Oracle Cloud Infrastructure), the core focal point, exceeded expectations, with IaaS revenue growing 93% year-over-year to $5.8 billion.
Management also updated its fiscal 2027 guidance: reaffirming a revenue target of $90 billion and raising the non-GAAP earnings per share guidance to $8.05. Regarding capital expenditure, the company disclosed that net cash capital expenditure for fiscal 2027 would be approximately $70 billion. When including customer prepayments and supplier payment timing differences, total capital expenditure is projected to reach between $90 billion and $95 billion.
The market's concern is not demand, but the intensity of investment. Just as investors were still digesting the new round of AI computing arms race among OpenAI, Anthropic, and major cloud providers, Oracle announced a far higher-than-expected capital expenditure guidance, significantly exceeding prior market consensus. In contrast, Oracle's total revenue for fiscal 2025 was $57.4 billion. A capital expenditure of this magnitude is almost unprecedented for a company primarily focused on enterprise software and cloud services.
This has created a core divergence in the capital markets: Is Oracle locking in the growth dividends of AI for the next decade in advance, or is it embarking on a high-leverage, high-risk expansion path? This is precisely the root cause of the sharp stock price volatility following the earnings report.
First Concern: The $90 Billion Capex – Is It Uncontrolled Expansion?
The biggest point of contention in the earnings report stems from the capital expenditure guidance. Oracle expects total capital expenditure for fiscal 2027 to reach $90 to $95 billion, representing explosive growth from previous levels. This figure is approaching the investment scale of some hyperscale cloud providers.
For many investors, this implies two risks. The first is cash flow pressure. Building AI data centers requires massive investment in GPUs, networking equipment, and power infrastructure, and the payback period for such assets often spans several years. If future AI demand growth slows, the massive investment could quickly turn into an asset burden.
The second is financing pressure. The company also disclosed that it will arrange approximately $40 billion in new financing over the next year, including a previously authorized $20 billion equity financing plan. This further intensifies market concerns about equity dilution and balance sheet deterioration.
From the valuation framework of a traditional software company, this aggressive expansion path is clearly unsettling.
Second Concern: Could an AI Bubble Recur?
The market's deeper concern stems from the entire AI industry. Over the past year, the biggest debate in the tech sector has consistently revolved around a core question: Is AI demand genuine growth, or is it being pulled forward by capital markets?
Giants like Microsoft, Amazon, Google, and Meta are building data centers at an unprecedented pace, and Oracle is now joining this race. If future demand for AI model training slows and enterprise customers cut back on computing power procurement, the large-scale data centers being built today face the risk of declining utilization.
This also explains why some investors view Oracle's massive capital expenditure as a "signal of the AI cycle peak." In their view, the company is betting on sustained demand for the coming years at current valuations, and any slowdown in demand could amplify the investment risk.
The Overlooked Signal: The $638 Billion RPO
However, Bernstein analysts believe the market's focus may be misplaced.
Compared to the $90 billion capital expenditure, a figure more worthy of attention is the $638 billion RPO. RPO, or remaining performance obligations, refers to the backlog of signed but not yet recognized revenue. As of the fourth fiscal quarter, Oracle's RPO reached $638 billion, with $85 billion added in the single quarter, a staggering 363% year-over-year increase.
More crucially, the new orders primarily came from large AI projects. The company revealed that just this quarter, four customers signed long-term contracts exceeding $8 billion. This means Oracle is not building data centers first and then looking for customers; rather, it is expanding capacity based on a large volume of orders already secured.
For the infrastructure industry, this sequence is critical. Investors see capital expenditure, while management sees locked-in future revenue.
AI Data Centers: Not Vacancy Risk, but Capacity Bottlenecks
Based on information disclosed by management, Oracle's current problem is not insufficient demand, but insufficient supply.
To date, the company has approximately 1.2GW of data center capacity and plans to add nearly 1GW next quarter. Simultaneously, the company has secured resources for over 10GW of future capacity and stated that the vast majority of this capacity is already reserved by customers.
This means Oracle's investment logic is closer to manufacturing capacity expansion rather than the traffic bets common in the internet era. When customers have already signed contracts and prepayments are in place, new capital expenditure essentially serves to fulfill future revenue.
In fact, the company disclosed that approximately $20 to $25 billion of capital expenditure would be directly covered by customer prepayments, further reducing its own funding pressure.
The Underestimated Growth Curve: Sovereign AI Cloud
Beyond AI training clusters, Bernstein also highlighted an easily overlooked opportunity—sovereign AI cloud.
As governments worldwide increasingly prioritize data security and AI autonomy, more countries are beginning to build localized AI infrastructure. Compared to Microsoft, Amazon, and Google, Oracle has long been deeply entrenched in the government and enterprise database market, giving it a unique first-mover advantage in the sovereign cloud space.
Analysts believe sovereign AI cloud could become a new growth engine following OCI in the coming years. Notably, this type of business typically employs a customer-owned hardware or prepayment model, which is not only more capital-efficient but may also offer better profit margins than traditional cloud business.
Comments