Oracle is expanding its partnership with OpenAI and integrating Google chips into its infrastructure

Oracle announced an expansion of its partnership with OpenAI, under which the ChatGPT chatbot and the Codex code-writing tool will be integrated into the company’s enterprise applications. At the same time, the cloud giant expanded its partnership with Nvidia and signed agreements to use chips from Google, Cerebras, and SambaNova in its networks, in an effort to reduce its dependence on the largest processor manufacturer, according to Yahoo Finance. Oracle’s stock rose nearly 4% during trading on October 27.

At the Oracle AI World conference, the company announced an expansion of its partnership with OpenAI, for which Oracle is building a network of data centers in the U.S. as part of a five-year, $300 billion contract. The ChatGPT and Codex AI tools will be integrated into Oracle’s enterprise applications, and customers will gain access to ChatGPT’s models to enhance cybersecurity. The hyperscaler and AI developer will also launch Oracle AI Security Mesh: enterprise customers will be able to use OpenAI models for cybersecurity testing.

At the same time, Oracle announced that it will expand its partnership with Nvidia and deploy Vera Rubin systems on a “multi-gigawatt scale,” according to Reuters. The parties did not disclose the value of the agreement. Earlier, the chipmaker’s CEO, Jensen Huang, said that Vera Rubin systems are capable of generating up to $40 billion in revenue per gigawatt of capacity for the chipmaker itself.

In addition to expanding its partnership with Nvidia, Oracle is diversifying its infrastructure: even in systems based on Nvidia chips, the company will phase out some of Nvidia's networking solutions in favor of its own Acceleron chips.

In addition, Oracle announced that, for the first time, it will begin offering Google’s tensor processing units (TPUs) to customers. The company also plans to offer products from SambaNova Systems—which is backed by Intel—and Cerebras Systems. Both companies manufacture chips for inference—processing user queries using pre-trained AI systems. The OpenAI Jalapeño AI chip will also be available on Oracle’s cloud infrastructure.

“Different AI tasks place different demands on computing power. This choice allows customers to select the right infrastructure based on the performance, throughput, and cost-effectiveness of each AI workload,” explained Mahesh Thiagarajan, president of Oracle Cloud Infrastructure.

Investors reacted positively to the news of the hyperscaler’s new partnerships: Oracle’s stock rose 3.6% to $130.6 during trading on October 27. However, over the past 12 months, the company has lost just over half of its market capitalization: the market is concerned about Oracle’s high debt and its heavy reliance on a single company.

Oracle did not disclose how it intends to finance chip purchases under the new agreements. During its earnings call for the previous quarter, the company stated that customers are increasingly paying for equipment themselves and maintained its spending forecast for the current fiscal year at approximately $90–95 billion.

As of the end of August, Oracle’s debt stood at approximately $125 billion. In the previous fiscal year, the company raised $43 billion through bond issuances and plans to raise another $40 billion or so through debt and equity financing in the current fiscal year. The company has already raised half of that amount—$20 billion—through the sale of shares. Oracle does not plan to issue new bonds until the end of 2026.

Wall Street has a positive outlook on Oracle stock: 36 out of 45 analysts tracking the cloud computing company’s stock recommend buying it, according to MarketWatch. Another eight recommend holding it, and one recommends selling it. The analysts’ average price target is $241.1—which is 84.7% higher than the closing price on October 27.

This article was AI-translated and verified by a human editor