Why scrapping Nvidia+Arm deal is ultimately bad for the industry | ZDNet

The largest proposed semiconductor acquisition in IT history – Nvidia merging with Arm – was called off today due to significant regulatory challenges, with antitrust issues being the main hurdle. The $40 billion deal was initially announced in September 2020, and there has been wide speculation that this would eventually be the outcome based on several factors that I believed were either not true or overblown. 

Before I get into that, it’s important to understand why this deal was so important. Nvidia’s core product is the graphics processing unit, or GPU, which was initially used to improve graphics capabilities on computers for uses such as gaming. It just so happens that the architecture of a GPU makes it ideal for other tasks that require accelerated computing, such as real-time graphics rendering, virtual reality, and artificial intelligence. 

A second key point is that GPUs alone won’t run a system, so central processing units (CPUs) are still needed. The CPU boots the system, runs much of the operating system, and performs other tasks. Historically, few systems outside of gaming and supercomputers required a GPU and CPU, but that’s no longer the case. The rise of AI-enabled systems has created massive adoption of GPUs, and they can now be found in self-driving cars, video endpoints, digital whiteboards, and factory equipment. 

Nvidia takes a systems approach 

The third important factor to consider is that putting CPUs and GPUs together and optimizing their performance is not trivial. This is where the combination of Nvidia and Arm could have paid big dividends. Unlike most silicon manufacturers, Nvidia extends its value beyond the chip and into systems. It’s done an outstanding job of putting together all the components required to power a specific system. For example, its DRIVE family is a set of engineered systems specifically for self-driving cars and includes Nvidia’s GPU along with CPUs, software, operating system and SDKs. The car manufacturer can drop this into a vehicle and build AI features instead of having to do system design. 

CLARA in health care is another example.  The combination of CPU and GPU would have enabled Nvidia to better optimize its engineered systems; that’s good for everyone and would accelerate the pace of innovation. 

CPUs and GPUs are different markets 

One of the arguments against bringing Nvidia and Arm together is that it would give Nvidia an even more dominant position in the silicon industry. This is where markets are being mixed. While it’s true that Nvidia is the dominant shareholder in GPUs (with more than 80 percent of the market), Arm is a small player compared to Intel and AMD in CPUs with market share of under 10%. 

Intel has bumbled its way around GPUs for more than a decade but had trouble getting out of its own way. Nvidia’s acquisition would have made Arm stronger, but Intel would still be the dominant chip vendor. The silicon industry is comprised of multiple submarkets, and treating it as one shows a gross lack of understanding of the difference in these components. 

Another concern regarding the Nvidia takeover is that many tech companies have turned to Arm to power their own devices, with Apple being the latest. Arm has enabled Apple to build thinner, lighter, more powerful devices. In his post on CNBC, writer Sam Shead wrote: “If Nvidia stopped other companies using Arm’s chip designs in their semiconductors, then analysts said the implications could have been huge.” While the statement is true, there’s no evidence Nvidia would have done this. 

Open ecosystems are core to Nvidia 

Nvidia’s value is built on its ecosystem, and the company has always maintained the highest levels of openness to maximize the number of vendors that could use its technologies. Some critics point to the Broadcom–Cumulus partnership falling apart when Nvidia acquired Cumulus, but that was initiated by Broadcom. In fact, it’s my understanding that Nvidia was fully onboard with Broadcom–Cumulus, but Broadcom looked at the Mellanox acquisition of Nvidia as a threat. In this case, Broadcom put joint customers in a tough spot and forced them to switch silicon providers. Nvidia has always believed in choice and that ultimately the best technology wins. 

With Arm, Nvidia was emphatic that it would continue to invest in Arm and ensure that other firms could continue to benefit from the company’s chip designs, but it looks like that fell on deaf ears. The deal appears dead, at least for now. Arm and Nvidia are both important companies to the future of computing and AI, and the combination would have enabled many industries to accelerate innovation and that would have benefited everyone.