Oracle (ORCL -2.72%) shocked investors recently when the company projected that its cloud infrastructure revenue would skyrocket to $144 billion in the next five years. On top of that, this revenue growth was basically locked in through non-cancellable contracts.
While Oracle's stock skyrocketed on the news, I don't think it's going to be the biggest winner of its cloud computing push. The company already carries a heavy debt load and isn't currently generating any free cash flow. Meanwhile, it'll have to spend an enormous amount of money on capital expenditures (capex) in the next few years to capture this revenue.
It's also worth noting that the economics of this revenue aren't certain, and the big three cloud computing providers -- Amazon (AMZN 0.78%), Microsoft (MSFT 0.88%), and Alphabet (GOOGL 0.28%) (GOOG 0.21%) -- basically decided not to pursue this opportunity. OpenAI will presumably be one of Oracle's big customers in this buildout. Close partner Microsoft decided earlier to tap the brakes instead of committing to spending so much money on new artificial intelligence (AI) infrastructure for it. Meanwhile, the big three cloud computing companies are also letting Oracle build some data centers for them, instead of building them themselves.
As such, I think there are two other companies much better positioned to benefit from Oracle's spending spree than Oracle itself: Nvidia (NVDA 0.27%) and Broadcom (AVGO -0.48%).
Image source: Getty Images.
Nvidia
Nvidia is the dominant player when it comes to AI infrastructure, as its graphics processing units (GPUs) are the most widely used chips for powering AI workloads. The company held a remarkable 94% market share in the GPU space in the second quarter, while its data center revenue surged 56% to $41.1 billion. This growth also wasn't only powered by GPUs, as its networking equipment sales nearly doubled to $7.3 billion.
The company has created a wide moat in the GPU market through both its CUDA software platform and NVLink interconnect system. Nvidia created CUDA as a way to let developers program its chips for tasks other than their original purpose, which was to speed up graphics rendering in video games.
While uptake in other markets was slow, Nvidia pushed the free program into universities and research labs that were doing early work on AI. This led to developers being taught to program GPUs through CUDA and tons of code and tools being written on top of it. Meanwhile, the company's NVLink system basically connects its GPUs to allow them to act like a cohesive unit, making them more powerful.
As such, whenever there are any big AI data center projects in the works, Nvidia is sure to be one of the biggest beneficiaries, given its grip on this market. Oracle has a close relationship with Nvidia, making it a clear winner from its future spending.
Broadcom
Nvidia isn't the only chipmaker set to benefit from Oracle's future data center spending. Despite it having the best GPUs on the market, customers don't want to rely solely on Nvidia for their AI processing power. They'd like to keep the company honest when it comes to pricing and diversify their supply chain. They'd also like to try to lower costs, especially when it comes to inference, which, unlike large language model (LLM) training, is an ongoing expense.
Broadcom played a pivotal role in Alphabet's development of its highly touted tensor processing units (TPUs), which have helped improve AI workload performance and reduce costs. This success has led to other companies turning to Broadcom to help them develop their own custom chips, or XPUs as Broadcom likes to call them.
Broadcom has said that its first three XPU customers -- believed to be Alphabet, Meta Platforms, and TikTok owner ByteDance -- represent between a $60 billion to $90 billion serviceable market opportunity in fiscal 2027 (ending October 2027). However, the company surprised investors last quarter when it said a fourth customer was far enough along in the process to place a $10 billion order for the back half of fiscal 2026. This customer is widely believed to be OpenAI.
This is why Broadcom is also set to be a big winner from Oracle's data center buildout. OpenAI is one of Oracle's big new customers, and in addition to using Nvidia GPUs, will undoubtedly also be using its own custom AI chips developed by Broadcom. Broadcom is on track to generate only around $13 billion in AI chip revenue this fiscal year ending in October (estimated by taking 65% of its $20 billion AI revenue forecast), so the Oracle news is just a huge opportunity for the company moving forward.
