Oracle has entered into a strategic agreement with Bloom Energy to purchase up to 2.8 gigawatts of fuel cell systems to meet the escalating power demands of its AI data centers. This move addresses challenges such as slow grid approval processes and lengthy expansion timelines. By adopting an on-site, plug-and-play power generation model, Oracle significantly shortens the time required to deliver computing capacity. The companies have already signed an initial contract for 1.2 gigawatts of capacity. Oracle has also deepened its ties with Bloom Energy through warrants, securing a strategic capital relationship to ensure a continuous and stable energy supply for large-scale AI clusters serving key clients like OpenAI and xAI. Following the announcement, Bloom Energy's stock surged more than 14% in after-hours trading. The company’s shares have more than doubled this year due to soaring data center demand straining energy supplies. Oracle's stock rose nearly 2% after hours, following a 12.69% gain during the regular session that made it one of the top performers in the S&P 500 for the day. Despite this, Oracle's shares remain down about 21% year-to-date and have retreated more than 50% from their all-time high, as the market continues to monitor the sustainability of its AI monetization efforts.
The 2.8-gigawatt agreement underscores Oracle’s lack of confidence in the efficiency of traditional power grids. Surging demand for AI computing has exposed severe capacity shortages and approval delays in U.S. utility grids, with large data centers often facing wait times of five to seven years to connect. Oracle CEO Safra Catz has repeatedly emphasized in recent earnings calls that electricity, not chips, is the primary bottleneck restricting the expansion of Oracle Cloud Infrastructure. To bypass this obstacle, Oracle selected Bloom Energy’s solid oxide fuel cell technology, which can generate power on-site using natural gas or hydrogen. This off-grid, distributed energy strategy allows Oracle to bring a data center online in as little as 55 days, compared to a multi-year timeline for conventional construction. In addition to the initial 1.2-gigawatt order, Oracle secured warrants to purchase approximately 3.5 million shares of Bloom Energy, signaling a shift from mere customer to potential stakeholder in the energy supply chain.
Oracle has launched large-scale construction projects to build AI data centers for clients such as OpenAI and Elon Musk’s xAI. The company expects capital expenditures to reach $50 billion in the fiscal year ending in May. Driven by cloud services for AI companies, Oracle’s infrastructure business generated $4.9 billion in revenue in the quarter ended in February. Meanwhile, at the Customer Edge Summit, Oracle showcased how it is converting expensive computing and power resources into tangible AI products like Utilities Opower. This application uses generative AI to analyze vast amounts of electricity usage data, helping utility companies accurately forecast demand and encourage users to shift consumption to off-peak hours, thereby reducing grid stress. This creates a clear closed-loop logic: Oracle uses Bloom Energy’s power to generate computing capacity, then uses that capacity to create AI applications that optimize societal energy consumption. This business model—using efficient energy to produce computing power, which in turn optimizes energy distribution—was a key driver behind Oracle’s market-leading stock performance on Monday.
As power supply concerns are being addressed, demand in the compute rental market is experiencing explosive growth. CoreWeave, a key partner of NVIDIA, announced that it has expanded its AI cloud computing agreement with Meta to $21 billion, extending the partnership until 2032. The company has also secured multi-year collaborations with leading AI labs such as Anthropic. These long-term contracts signify CoreWeave’s evolution from a supplementary compute provider to a core player in AI infrastructure. By extending contract terms from one year to three and raising service prices, CoreWeave has successfully locked in long-term, predictable revenue. The capital market has responded positively, with several major investment banks, including Bank of America, Roth Capital, and D.A. Davidson, raising their price targets for the company, with one setting a high of $175. Analysts widely recognize CoreWeave as an essential benchmark in cloud services for the inference phase of large AI models. Industry observers note that as the focus of AI applications shifts from training to inference, requirements for stability and rapid deployment of computing infrastructure are intensifying. Oracle and CoreWeave’s strategies of securing long-term contracts and innovative power solutions are redefining the competitive landscape for next-generation AI cloud service providers.
Comments