Oracle's AI Backlog Soars While Cash Runway Shrinks

Oracle reported a significant surge in contracted demand for its AI cloud services during its first fiscal quarter of 2027, booking more than $30 billion in new agreements. While this figure suggests robust market interest, it does not represent immediate cash in hand. Instead, these commitments sit in the company’s backlog, waiting for the physical infrastructure to be built and activated before they can be converted into recognized revenue.

The financial reality of this expansion is visible in Oracle’s cash flow. The company spent nearly $28.5 billion on capital expenditures during the quarter, exceeding the cash generated by its operations. This resulted in a free cash flow deficit of $5.4 billion, a sharp deterioration from the $362 million deficit seen in the same period last year. Investors are watching this gap closely, as it highlights the heavy upfront cost of serving AI workloads.

Bookings Differ From Actual Sales

Understanding Oracle’s financial position requires distinguishing between three key metrics: bookings, remaining performance obligations, and recognized revenue. Bookings reflect new contractual commitments, while remaining performance obligations represent the total value of services Oracle has agreed to provide but has not yet delivered. Recognized revenue, on the other hand, is only recorded when those services are actually performed.

The $30 billion in new AI contracts added to the backlog rather than flowing directly into quarterly sales. This creates a large pool of future obligations that can be many times larger than current revenue. According to reports from Reuters and GN auto tech/cloud: cloud infrastructure, this backlog now stands at $664 billion. However, until Oracle builds the necessary data centers and deploys the hardware, these numbers remain a measure of potential rather than realized profit.

Infrastructure Bottlenecks Limit Revenue Growth

Oracle is actively working to close the gap between demand and supply. During the quarter, the company brought 850 megawatts of additional data center capacity online and delivered over 300,000 GPUs to customers. These efforts are already showing results, with cloud infrastructure revenue increasing by 121 percent year-over-year. This growth demonstrates that the new capacity is being used, but it is not yet enough to satisfy the entire backlog.

The constraint is physical capacity, not just digital contracts. Oracle states that demand for AI training and inference continues to outpace its ability to supply it. This means a portion of the contracted work remains dependent on the completion of new servers, networking equipment, and cooling systems. Until these components are operational, the company cannot recognize the associated revenue, creating a lag between signing contracts and earning income.

High Costs Drive Negative Cash Flow

The negative free cash flow is a direct result of the capital required to build this infrastructure. Oracle’s operating cash flow for the quarter was $23.1 billion, but its capital expenditure reached $28.5 billion. This means the company spent more cash on long-term assets than it generated from its day-to-day operations. While this is not an accounting loss, it indicates a period of intense investment that consumes significant liquidity.

A notable portion of the operating cash flow, approximately $11.4 billion, came from customer prepayments. These upfront payments provide a temporary boost to cash flow but do not offset the massive ongoing costs of construction and hardware procurement. The trade-off is clear: Oracle is betting its current cash reserves on the long-term viability of the AI market, hoping that future revenue from the $664 billion backlog will justify the current cash burn.