Defining Oracle’s AI Infrastructure Bet
Oracle’s AI infrastructure bet is the company’s plan to pour tens of billions of dollars into Oracle Cloud Infrastructure (OCI) capacity so it can run large-scale AI workloads for enterprises under long-term contracts, while proving that the heavy capital spending can convert into durable, profitable cloud revenue instead of becoming a debt burden. This bet is now central to the company’s story. Oracle has signaled capital expenditures of USD 50 billion (approx. RM230 billion) against an expected USD 67 billion (approx. RM308.2 billion) in fiscal 2026 revenue, alongside a remaining performance obligation that stood at USD 553 billion (approx. RM2.544 trillion) at the end of Q3. With Q4 earnings arriving on June 10, investors are focused on whether OCI’s economics can scale fast enough to support this backlog without overwhelming the balance sheet.
Q4 AI Contract Wins Show Demand Is Real
Oracle’s latest numbers show that demand for Oracle AI infrastructure is not experimental but contractual and large. In Q4, the company signed USD 67 billion (approx. RM308.2 billion) in AI infrastructure contracts, with four customers each committing more than USD 8 billion (approx. RM36.8 billion). Total Q4 revenue reached USD 19.2 billion (approx. RM88.3 billion), up 21% in US dollars, driven by 93% growth in cloud infrastructure revenue. Remaining performance obligations climbed to USD 638 billion (approx. RM2.935 trillion), up 363% year over year and USD 85 billion (approx. RM391 billion) sequentially, with nearly half scheduled to be recognized within 36 months. According to Oracle CFO Hilary Maxson, this long-term backlog provides “exceptional visibility into future revenue growth,” signaling that enterprises are locking in multi-year AI infrastructure commitments rather than short-lived pilot projects.
Capex, Debt, and the Cost of Building AI Capacity
Behind the AI contract wins sits an infrastructure build-out that is both rapid and expensive. Oracle has guided to USD 50 billion (approx. RM230 billion) in capital expenditures for fiscal 2026, and Mizuho analysis suggests the company may need to spend at least USD 80 billion (approx. RM368 billion) over the next three years before free cash flow turns positive in 2029. To fund the next leg of expansion, Oracle plans to raise approximately USD 40 billion (approx. RM184 billion) in debt and equity in fiscal 2027, including a previously announced USD 20 billion (approx. RM92 billion) at-the-market equity issuance, while stating it does not plan to raise additional debt in calendar 2026. The key question for investors is whether borrowing required to fund server capacity will remain below USD 100 billion (approx. RM460 billion) and still support acceptable returns.
Cloud Infrastructure Economics: Utilization vs. Free Cash Flow
Operationally, Oracle’s cloud infrastructure economics look tight but promising. The company delivered more than 1.2 gigawatts of data center capacity in fiscal 2026, with its Abilene, Texas facility providing 42% of that and more capacity due within months. Q1 fiscal 2027 delivery is expected to approach 1 gigawatt, roughly matching the prior four quarters combined. This new supply is being absorbed quickly: Oracle reports a global GPU utilization rate of 97.5%. In Q4, 35,000 GPUs came up for renewal and 49% of customers renewed for 92% of those GPUs, with most remaining units resold within the quarter. These figures suggest the company is not overbuilding. However, Mizuho expects free cash flow to remain under pressure until late in the decade, which keeps the spotlight on how efficiently each dollar of capex converts into high-margin, recurring cloud revenue.
Can AI Infrastructure Growth Deliver Sustainable Returns?
Whether Oracle’s AI infrastructure gamble pays off depends on converting contract wins into profitable, scalable services. Management has raised fiscal 2027 revenue guidance to USD 90 billion (approx. RM414 billion) and now targets 34% revenue growth for that year, roughly double the expected pace in fiscal 2026. Q1 fiscal 2027 cloud revenue is guided to grow 58% to 64% in US dollars. If that momentum holds while Oracle limits borrowing and net capex in fiscal 2027 to around USD 70 billion (approx. RM322 billion) after customer prepayments, the current backlog can act as a bridge to a larger, cash-generative business. The risk is that capital intensity outpaces returns, turning multi-year AI infrastructure contracts into a balance-sheet strain rather than a profit engine. The upcoming Q4 earnings call will be a major test of that thesis.
