Nvidia is acquiring Arm from Softbank in a $ 40 billion deal (much of it in stock, thanks to Nvidia’s high flight rating). And while this seems like an interesting fusion of complementary technologies, there are plenty of challenges that could make this a very problematic combination.
First, will the acquisition give a break for many of Arm’s licensees, especially Qualcomm, Broadcom and other chip vendors, and maybe even Apple? Suddenly, an open licensing model belongs to a competitor who makes chips. Nvidia claims it will hold the open model in place and arm work separately, but that’s unlikely in the long run. Arm is historically an IP licensing company that does not have the chip manufacturing capabilities to compete with its licensees. This is very similar to an open source model that is so common in software. This gave him tremendous momentum in licensing his intellectual property. Now it will be owned by a company that sells chips. That creates a completely different dynamic.
Ultimately, Nvidia wants to be a major player in the entire chip area and compete with the big boys (e.g. Intel), especially in the data center with its graphics functions and in AI. This is a big part of the growth over the past few years and is important to success. I suspect Nvidia sees Arm as an opportunity to be a full-service provider for data centers. However, a number of companies have tried with little success to promote ARM as a functioning data center processor and especially as a cloud solution.
Qualcomm, which has a lot of experience and credibility, had an ARM architecture-based data center product (Centriq) a few years ago and was unable to try it. Same goes for AMD. There are still some smaller companies out there trying, and Google and AWS have ARM licenses to do their own thing, but it̵
In 2011, a wireless modem supplier (Icera) was even bought to integrate the modem into its chips. That was primarily aimed at the glowing mobile space, but those efforts have failed. Now it is targeting the data center that has the greatest potential for growth. But will it be successful this time? It remains to be seen how well the company can compete against Intel and AMD with an ARM-based, GPU-centric offering for the data center / cloud. It’s harder to be a full-fledged processor provider than just designing GPUs really well.
Nvidia has the best technology in the GPU area and has sold its GPUs very successfully in the PC and now in the data center area. Will it try to get Arm to introduce its own GPU in place of the current Arm family of GPUs called Mali? What does this mean for Arm IP licensees? Qualcomm made its own GPU design called Adreno because the arm designs don’t have enough functionality, as do Apple and others. Mali really isn’t the most powerful GPU, so improvements Nvidia has to offer could help the market out. However, many chip suppliers do not want to be tied to what they consider to be the ultimate competitor.
One could argue that Arm gives Nvidia a strong position to pursue the cellular market. And the acquisition might help in that regard, but most of the big smartphone vendors are using chips from Qualcomm that directly compete with what Nvidia is trying to do in GPU and AI (Qualcomm is doing its own thing here, not even with the basic arm- IP now available) and Mediatek, which licenses the Arm IP but generally competes on mid-to-low end phones (and there are others too, such as Samsung and HiSilicon / Huawei that make their own Arm IP chips). That’s why I don’t see Nvidia as an important chip supplier for smartphones anytime soon. Even if it does choose to offer a full chip, it will likely be on the high end and have to compete directly with Qualcomm, which would be hard to beat given its market position and product strength.
There is also a strong psychological negative impact on the Nvidia and Arm combination for licensees. Will owning Arm Nvidia give Nvidia a competitive edge in the marketplace as it doesn’t have to pay a license and royalty for the IP it would use to make chips? And will others be willing to subsidize Nvidia by paying Arm a license fee? Are licensees concerned that their proprietary use of Arm IP will be compromised or used by Nvidia? I think this is going to be a problem, not in the short term as you can’t just rip and replace technology, but in the longer term, especially for the biggest players who have the bare minimums to walk alone and ultimately move away from the Arm IP when they do perceive it as a risk. This would likely take at least 2-3 years, but it could happen if licensees are concerned. For example, Apple built its own chips for the Mac to get Intel out of its systems, and Intel wasn’t a direct competitive threat.
Will China eventually allow this acquisition? As listed companies, they come under the jurisdiction of the regulatory authorities, at least in the USA, the EU and China. Arm is a UK based company and concessions have been made to expand UK R&D facilities which could calm a backlash from the EU. However, after the acquisition, Arm can be considered a US company. Will arm tech licensing come under the terms of the Trump boycott of China? Does the licensing of intellectual property to China, particularly Huawei, but possibly Mediatek, have to end (Mediatek is Taiwanese but is still strongly represented in China)? That would be a HUGE disturbance. I don’t know the answer to this question, but it could be a big problem and give China a break so the acquisition can go ahead …
So the bottom line is that this acquisition is no big deal for Nvidia, as there are many longer-term challenges in the market.
Jack Gold is the founder and lead analyst at J. Gold Associates, LLC., An information technology analyst firm based in Northborough, MA that covers the many aspects of business and consumer computing and emerging technologies. Follow him on Twitter @jckgld or LinkedIn at https://www.linkedin.com/in/jckgld.