Oracle has pledged to raise no more than USD 20 billion in additional debt and equity financing for its artificial intelligence (AI) infrastructure expansion through fiscal 2030, seeking to reassure investors concerned about the rising costs of the company’s data-centre ambitions.
The company, while making the announcement last week, also outlined a path towards stronger cash generation, forecasting that its capital expenditure activities would peak in the current fiscal year and next before declining.
Speaking at Oracle’s investor day in Las Vegas, Chief Financial Officer Hilary Maxson said the company expects free cash flow to turn positive in the second half of fiscal 2028 and exceed USD 30 billion in fiscal 2029.
Oracle also expects to stop relying on external financing for growth from fiscal 2029.
As per Maxson, Oracle’s net debt is forecast to fall below 3.5 times annual core earnings that year, as lower capital spending and stronger cash generation improve its financial position.
While investors are becoming concerned about American Big Tech companies committing to aggressive AI buildouts, Oracle faces a particularly troubling situation because, unlike cloud giants such as Microsoft and Amazon, it has a smaller revenue base and relies heavily on a single customer, OpenAI.
Its partnership with OpenAI has added hundreds of billions of dollars to Oracle’s cloud backlog.
However, fulfilling those contracts requires substantial investment in data centres and other AI infrastructure, forcing the company to depend heavily on debt and equity financing.
Oracle stated that vendor financing, hardware provided by customers, and customer pre-payments would help lower its upfront cash requirements until free cash flow becomes positive.
The company had previously announced plans to raise USD 40 billion through debt and equity in its current fiscal year.
The target also included a USD 20 billion stock sale completed in the first quarter.
Maxson said the additional funding requirement for the remainder of fiscal 2027 would be capped at USD 20 billion, split evenly between debt and equity.
“We’re announcing that this USD 20 billion, split evenly between debt and equity, is the final amount needed to fund the remainder of this cycle,” she said.
Oracle reported USD 125.34 billion in notes payable and other borrowings as of August 31, while cash and cash equivalents stood at approximately USD 36.37 billion.
Alongside its financing plans, Oracle provided an upbeat outlook for fiscal 2030.
It expects revenue to exceed USD 225 billion, with an adjusted operating margin above 40% and adjusted earnings of more than USD 21 per share.
The company also reaffirmed its fiscal 2027 forecast of revenue above USD 90 billion and adjusted earnings to USD 8.10 per share.
Oracle’s projections helped lift shares of other companies involved in AI cloud computing and infrastructure after weeks of selling pressure driven by concerns that the AI investment boom was becoming increasingly dependent on borrowing.
CoreWeave, Nebius, Applied Digital and Iren gained between 2.1% and 5% following Oracle’s announcement.
Oracle has also expanded its AI computing capacity, apart from diversifying the chips available through its cloud infrastructure business.
The company will, in the coming days, scale its deployment of Nvidia’s latest-generation Vera Rubin AI servers to “multi-gigawatt scale.”
Nvidia Chief Executive Jensen Huang has previously said that Vera Rubin systems could generate up to USD 40 billion in revenue per gigawatt for the chipmaker.
However, Oracle is also looking to reduce its dependence on a single chip supplier.
Even its Nvidia-based systems will use Oracle’s own Acceleron networking chips instead of some of Nvidia’s networking products.
Oracle also plans to offer Google’s tensor processing unit (TPU) chips for the first time, alongside processors from Intel-backed SambaNova Systems and Cerebras Systems.
The move reflects the different computing requirements of AI workloads. While Nvidia dominates the market for training AI models, the inference market, in which AI systems respond to user requests, has a wider range of competitors.
“Different AI workloads have different compute needs,” said Mahesh Thiagarajan, president of Oracle Cloud Infrastructure.
He said offering customers more chip options would allow them greater flexibility to select infrastructure based on performance, throughput and cost.
Cerebras Chief Executive Andrew Feldman described the Oracle agreement as a significant expansion of the company’s customer base, saying its systems were being deployed at scale in other companies’ data centres.