NVIDIA Reaches Computing Superpower Status

It’s official: NVIDIA will acquire ARM from Softbank for $40B in a stock and cash deal that is the largest semiconductor deal I’ve heard of in that industry.

This acquisition is a historic deal for the computer industry, and what I believe is another inflection point for NVIDIA, as it becomes one of the very few companies that can build a computer from A to Z, only using its intellectual property (IP).

Previous inflection points, in my opinion, were:  hardware Transform & Lighting, programmable shading, the nForce platform, and Computing/AI on GPUs.

Softbank comes out of this with a ~10% (maximum) stake in NVIDIA and $2B in cash, and in a place where it no longer has to manage ARM actively, but instead sit back and watch the synergy between NVIDIA and ARM create something more valuable than the sum of their parts.

ARM has now access to NVIDIA’s GPU intellectual property. Previously, ARM and NVIDIA were potential competitors, with ARM designing its own graphics processors and selling the designs to various clients in the smartphone business, like Huawei or MediaTek. But in reality, it is about much more than GPUs.

And now, NVIDIA controls the most crucial CPU intellectual property of our times. NVIDIA is now virtually independent of any other 3^rd party when it comes to building computing systems, joining a tiny club of companies (Intel + AMD).

Apple, Qualcomm, Huawei, MediaTek… virtually everyone else uses ARM’s intellectual property. And now, which is now NVIDIA’s. The chances are that NVIDIA will not scare-off current ARM customers with any sudden moves, but alters the balance of power in unimaginable ways.

Through ARM’s vast network of clients, NVIDIA’s graphics technologies could finally permeate in the mobile ecosystem (a longtime objective) that was once the target audience of NVIDIA’s Tegra line of product. A mobile GeForce with RTX ray-tracing is now a real possibility.

Tegra was once a top-performer but was ultimately retired because customers preferred buying Qualcomm’s SoC products like the Snapdragon platform.

At the time (~2016), NVIDIA chose to shelve its smartphone business and focus on things that it could dominate, such as GPUs and datacenter products. The high-margin datacenter market is a prime target for NVIDIA: the company made great strides with A.I and high computing (HPC) products. With ARM technologies, NVIDIA could increase the power-efficiency of its products.

In 2016, I said that there were “better business opportunities for NVIDIA” (than mobile) and products like the NVIDIA DGX-1 or the Tesla P100 had lucrative datacenter + AI clients for the company. Obviously that cash did not go unused. This is now a business case-study of “live to fight another day”.

A synergy between ARM and NVIDIA technologies could yield even better performance/Watt against the X86 equivalent from Intel/AMD.

Already, some people are claiming that this is a new NVIDIA monopoly, but at present, I do not think that’s the case, as long as ARM’s clients keep having access to its licenses. If NVIDIA were to cut-off access for Apple and Qualcomm, then things would be different, but I cannot see anything like that happen. It would be foolish.

This is a fantastic strategic move from NVIDIA, as ARM is currently a one-of-a-kind asset. There are alternatives such as the MIPS open architecture, but there is no evidence that this initiative is a threat to ARM in the foreseeable future.

Perhaps this acquisition will motivate ARM competitors, but beyond its sheer technological prowess, NVIDIA has proven again and again that it can create ecosystems that others cannot. That, in my opinion, is the real force behind this $40B acquisition.

Filed in Breaking >Computers. Read more about ARM, Editorspick, GPU (Graphics Processing Unit), NVIDIA, Processors and Semiconductors.