Analysis: Embraced By IT World, Nvidia Is Now World’s Most Valuable Company
Nvidia couldn’t have reached the apex of the corporate world without the support of the cloud service providers, server vendors, data center infrastructure software providers and independent software vendors who have increasingly embraced its ‘full-stack computing’ platform.
The Santa Clara, Calif.-based company hit a $3.34 trillion market cap on Tuesday, allowing it to overtake Microsoft, as its stock price rose more than 3.5 percent. The only other company close to Nvidia in market cap is Apple, which is now valued at $3.28 trillion.
[Related: Analysis: As Nvidia Takes AI Victory Lap, AMD Doubles The Trouble For Intel]
Nvidia couldn’t have reached the apex of the corporate world without the support of the cloud service providers, server vendors, data center infrastructure software providers and independent software vendors who have increasingly embraced the company’s growing portfolio of GPUs, CPUs, networking gear, server platforms and software.
While Nvidia has long stood out due to continuous advances in its GPUs—which have been key to providing the accelerated performance needed to make generative AI advancements possible—a great deal of its recent success has come from its focus on evolving from a chip designer to what CEO Jensen Huang calls a “full-stack” computing company.
Starting with the foundation of its CUDA programming model that has allowed developers to embrace GPUs for a growing number of demanding workloads, Nvidia has spent the past several years building out this stack, either by acquiring companies or developing internal capabilities, so that it could tightly integrate and tune each component, with the goal of optimizing performance and efficiency while keeping development barriers low.
These components range from GPUs and CPUs to data processing units (DPUs) and high-speed interconnects, and the company is packaging them into its own server platform designs, such as the HGX, for cloud service providers and OEMs to adopt for their own products.
This has increasingly put Nvidia behind the steering wheel when it comes to the way servers and data centers are designed.
Take, for instance, Amazon Web Services’ Project Ceiba, which the company said in March will become “one of the world’s fastest AI supercomputers.” This supercomputer will be made up of Nvidia’s GB200 NVL72 systems, which are multi-node, liquid-cooled, rack-scale servers that each contain 36 of Nvidia’s upcoming GB200 Grace Blackwell as well as its BlueField-3 DPUs.
The GB200 NVL72 systems will also become the basis for instances powered by Nvidia’s upcoming Blackwell GPU architecture from the likes of Microsoft Azure, Google Cloud and Oracle Cloud Infrastructures. And these cloud service providers plan to offer Nvidia’s DGX Cloud service, which includes the company’s Nvidia AI Enterprise commercial software platform as well as enterprise support, on top to aid with AI development.
Nvidia’s influence is also strong with server OEMs, and that was most recently demonstrated by Hewlett Packard Enterprise, which on Tuesday launched a comprehensive generative AI solutions portfolio co-developed and co-branded by the AI chip giant.
Called “Nvidia AI Computing By HPE,” the portfolio’s name further underscores how central Nvidia has become to modern data center infrastructure offerings, covering products not just for compute but also storage and networking. It even comes with a joint enablement and training program with combined market development funds for channel partners.
It’s this kind of influence that allowed sales of Nvidia’s data center GPUs and platforms to skyrocket last year, which led to the company surpassing semiconductor stalwart Intel in annual revenue. The company’s latest earnings results show that it’s not losing momentum yet.
But can Nvidia keep this up for the foreseeable future? There are several companies hoping to chip away at Nvidia’s influence, and they include AMD and Intel, who both believe the market is looking for more open alternatives for powering AI and other demanding workloads. Even AWS, Microsoft Azure and Google Cloud are building competitive data center accelerator chips.
Even if Nvidia loses some influence in the long run, it will have succeeded in remaking the IT infrastructure world in its image and forcing others to figure out how they can do things better.