Nvidia's $40 billion Arm bet all about data centers, edge computing, AI

The Nvidia-Arm combination is compelling on multiple fronts and could be "a first class data center platform" but now the parties just need regulators to buy into the deal.

Nvidia is acquiring Arm for $40 billion and the master plan revolves around data centers, supercomputing and edge computing and melding server CPUs with GPUs.

While there will be a lot of focus on Nvidia licensing Arm to rival chipmakers, Nvidia CEO Jensen Huang said the acquisition is really about AI workloads and edge computing.

See: Arm processors: Everything you need to know

On a call with journalists after the deal was announced Sunday night, Huang explained:

Nvidia doesn't design CPUs, we have no CPU instruction set, Nvidia doesn't license IP to semiconductor companies, so, and in that way, we're not competitors. We have every intention to add more IP tools and also unlike Arm, Nvidia does not participate in the cell phone market.

Our intention is to combine the engineering and the tech -- the R&D capacity of both companies so that we can accelerate the development of technology for Arm's vast ecosystem, and one of the areas ... that we very interested in, is to accelerate the development of server CPUs.

Here's how Nvidia and Arm see the circle of licensing go together as the Internet of people turns into the Internet of things and data. 

Jefferies analyst Mark Lipacis said in a research note:

In previous computing eras, one ecosystem typically controlled 80% of the profits of the ecosystem – in PCs it was Wintel, in Smartphones it is Apple. Due to the ecosystem it has created in Acceleration (processor, interconnect (MLNX), software stack including CUDA and vertical market software), we've argued that Nvidia was best positioned to capture 80% of the profits of the Parallel Processing Era. However, with ARM, we think Nvidia is poised to also offer general purpose, heterogeneous data center scale solutions and ultimately capture up to 80% of the value of the serial processing part of the data center ecosystem as well. We also think that ARM gives it a larger moat in the data center.

Indeed, the addressable market for Nvidia and Arm is huge.

Here's how Huang said Nvidia will turbo charge Arm's business.

• "Nvidia's R&D platform much larger than Arm's. We have a lot of processor development ourselves. A lot of tools we use can be tapped into."
• "The infrastructure required to turn a CPU into a data center platform. We have the capabilities to bring all of that to Arm. We'll be able to create a choice for the data center world that is world class, complete and optimized."
• "We can now harmonize roadmaps."

Patrick Moorhead, principal of Moor Insights & Strategy, said:

The NVIDIA-Arm deal is not only the largest semiconductor deal by dollar volume at $40B, but I believe the one with the most significant impact. I think the deal fits like a glove in that Arm plays in areas that NVIDIA does not or isn't that successful, while NVIDIA plays in many places Arm doesn't or isn't that successful. NVIDIA brings incredible capitalization to Arm which, as we have seen since its Softbank acquisition, Arm has increased its market presence and competitiveness. Softbank investment has enabled Arm's thrusts in the datacenter, automotive, IoT and NPU markets.  I believe the NVIDIA adder can only make it stronger as long as it sticks with its commitment to let Arm do what they do best, which is creating and licensing IP in a globally-neutral way which it is committing.  

The neutrality question

For Arm to continue to do well, it will have to do business with current and potential Nvidia rivals.

Huang said that Arm will remain neutral and work with everyone in the industry. The Arm brand and model will be maintained. On a conference call with analysts, Huang said Nvidia had the trust of the chip sector, works with everyone in the industry such as Intel and AMD and can keep Arm neutral. Analysts were repeatedly questioning Huang's take. 

• How Nvidia built Selene, the world's seventh-fastest computer, in three weeks
• Nvidia and Google claim bragging rights in MLPerf benchmarks as AI computers get bigger and bigger
• Nvidia Ampere, plus the world's most complex motherboard, will fuel gigantic AI models

What's unclear is whether another chipmaker can own the IP that is licensed to rivals. Arm fared well because it was neutral and not owned by chipmakers. Arm is the most ubiquitous computing platform.

When Softbank bought Arm, the deal was a bit of an industry curve ball, but it worked because the buyer wasn't a chip industry player.

Arm CEO Simon Segars said he will continue to run Arm's open computing model. "There's so much we can do as a combined company," said Segars, who referenced research to create an Arm supercomputer.

"The customers have reached out to customers. We believe customers are going to be heartened by our genuine attention to keeping the Arm platform. We bought Arm because of its open ecosystem," said Huang. "We are going to bring more choice and turn Arm into a full-fledged data center platform."

Will regulators approve?

The Nvidia-Arm deal looks strong on paper, but analysts have noted that it's unlikely to get regulatory approval.

Huang said the deal is pro competition and good for customers.

Wall Street analysts noted that Nvidia and Arm are going to get caught up in US-China relations. US has been blocking China's access to technology since the Qualcomm-NXP deal was shot down.

Wedbush analyst Matt Bryson said:

We believe these benefits (to a Nvidia-Arm deal) are likely moot as we see the probability this deal will be blessed by antitrust authorities, specifically China's Ministry of Commerce (MOFCOM), as de minimus. Given any market concentration/competitive concerns, we don't see why MOFCOM would approve a deal that would give the US control over another important piece of semiconductor IP, particularly in light of the US's recent actions to block Chinese access to US IP and goods.