Oracle’s $455B Cloud Shock: Nvidia, AMD, and Broadcom Are Big Winners

During its earnings release on Monday, Oracle (NYSE:ORCL) just dropped a bombshell that caught even the most bullish analysts off guard. The company’s remaining performance obligations (RPO) for its cloud business exploded 359% year-over-year to $455 billion, compared to just $138 billion in the prior quarter. This jaw-dropping surge sent Oracle stock soaring nearly 36% in Wednesday’s trading, marking its best single-day gain in decades. Think of RPO as Oracle’s locked-in revenue pipeline – contracts already signed and sealed, just waiting to be delivered. This isn’t wishful thinking or ambitious projections; it’s money that is already committed by customers. CEO Safra Catz said that Oracle expects to sign several more multi-billion-dollar customers in the near term, pushing its RPO above the half-trillion-dollar mark within months. Here’s a quick look at how the big AI silicon players will benefit from Oracle’s big push. How Oracle Stock Surges 3x To $900?

Image by Pete Linforth from Pixabay

Big Infrastructure Demand

For Oracle, converting these obligations into revenue will depend on how fast it can scale its infrastructure. That requires electricity, regulatory approvals, and high-performance GPUs, which remain in chronic shortage worldwide. The company also raised its capital expenditure forecast to $35 billion for fiscal 2026 – a 65% jump that underscores the urgency of its AI build out. No wonder the blowout guidance sent shares of Nvidia, Broadcom, and AMD higher by between 2% to 10%, as these chip vendors form the backbone of almost every AI deployment. Every incremental Oracle contract should guarantee more demand for their semiconductors. Moreover, unlike cloud rivals such as  Google, AWS, or Microsoft, Oracle isn’t developing its own custom AI chips. It has relied instead on GPUs from Nvidia (NASDAQ:NVDA) and AMD (NASDAQ:AMD).

Nvidia – Top Tier Chips, Deep Ties

Oracle has a strong partnership with AI compute standard bearer Nvidia, which supplies the majority of GPUs used in Oracle Cloud Infrastructure currently. Oracle is unique among hyperscalers for offering so called “bare metal” instances equipped with Nvidia GPUs. This gives customers exclusive, direct access to top-tier physical chips A10, A100, H100, H200, and the flagship Blackwell chips, without the virtualization overhead. This allows enterprises to maximize performance for model training and inference. Oracle’s AI supercomputers deploy these GPUs in dense, liquid-cooled data centers designed for energy efficiency and throughput. This close collaboration enables Oracle to support some of the world’s most advanced AI applications at scale.  Nvidia GPUs have been the workhorses of Oracle’s AI infrastructure, and the partnership could remain crucial as Oracle executes on its massive AI contracts in the coming years. Nvidia Stock To Fall 50% As AI Cycle Turns?

AMD – Expanding Options

Oracle hasn’t been putting all its eggs in Nvidia’s basket either. In June, Oracle and AMD announced that the AMD Instinct MI355X GPUs will be available on OCI, offering more than 2x better price-performance than the prior generation. Oracle planned to deploy up to 131,072 MI355X GPUs enabling enterprises to build, train, and inference massive AI models at competitive economics. This partnership ensures that AMD captures a meaningful share of Oracle’s AI spend. The MI355X, built on AMD’s new CDNA 4 architecture and TSMC’s 3nm process, represents AMD’s biggest leap yet in AI accelerators. For Oracle, the move diversifies its hardware options, reduces single-supplier risk, and should crucially boost leverage in negotiations with Nvidia for GPUs, keeping its overall spending in check.

Broadcom – Inference And Custom Chips 

Oracle’s revenue outlook is apparently increasingly tied to AI inferencing, the process of running trained models at scale for real-time user queries. Buy or Sell AVGO Stock at $360? As AI shifts from development into mainstream adoption, inference workloads are expected to far outweigh training needs, creating a recurring, utility-like demand stream. This is precisely where Broadcom and its custom application specific integrated circuits (ASICs) could emerge as a major winner. While GPUs dominate training, inference at hyperscale is likely to lean toward custom chips that can deliver better cost and energy efficiency. To be sure, Oracle still primarily uses GPUs for AI workloads, but we believe that ASICs will almost certainly play a larger role going forward. This dynamic helps explain why Broadcom stock surged nearly 10% on Wednesday – outpacing Nvidia and AMD, which rose 2% to 4% – as investors bet Broadcom’s custom chips will capture a share of Oracle’s workload.