I talked today with NVIDIA CEO Jensen Huang and Arm CEO Simon Segars on the largest and what I consider the most important semiconductor acquisition- ever. While Huang and Segars are quite different personality-wise, their points of view on our video call about the deal were consistent, well considered and communicated. I am thoroughly impressed at what the two companies could create over the next decade and, unlike other industry analysts, have thought that way since July when rumors started swirling. I’d like to map out the news and my opinions on the news.
The following are the deal highlights per the press release:
- Price: $40 billion, NVIDIA stock and cash
- Accretion: immediately accretive to NVIDIA non-GAAP gross margin and EPS
- Cambridge investments: Create “world-class” AI research and education center for healthcare, life sciences, robotics, and self-driving cars. Also, build an Arm/NVIDIA-based AI supercomputer for research
- Softbank ownership: Will keep 10% stake in new entity
The following are the operating highlights:
- Arm operating structure: Arm will operate as an NVIDIA division
- Arm locality: Arm will continue to be headquartered in Cambridge
- IP locality: Will keep registration in the UK
- Licensing model: continue to operate its open-licensing model while maintaining its global customer neutrality
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.
NVIDIA CEO Jensen Huang is excited about the future vision of the combined companies. He told me, “We’re about to enter a phase where we’re going to create an internet that is thousands of times bigger than the internet that we enjoy today. A lot of people don’t realize this. And so, so we would like to create a computing company for this age of AI.”
Software is a key part of the combined company’s vision. Arm CEO Simon Segars framed it well when he told me, “We’re moving into a world where software doesn’t just run in one place. Your application today might run in the cloud, it might run on your phone, and there might be some embedded application running on a device, but I think increasingly and with the rollout of 5g and with some of the technologies that Jensen was just talking about this kind of application will become spread across all of those places. Delivering that and managing that there’s a huge task to do. And it all requires a computing architecture that can scale from the tiniest sets all the way up to the biggest supercomputer, and we can address that.”
The new NVIDIA-Arm combination now plays in nearly every market segment in the datacenter, edge of datacenter, personal computers, smartphones and the IoT. My imagination runs wild with the possibilities of:
- More “big-core” Arm datacenter general-purpose processors
- Bigger CPU, GPU, NPU, and networking datacenter combinations
- CPU-GPU-NPU combination with shared memory systems for the HPC market
- NVIDIA-based smartphone and tablet GPU/NPU IP
- Arm-based SoC for Windows notebooks
- Arm-based big core CPU for highest-performance Windows desktops
Huang told me that first thing that the combined company will do is to, “bring NVIDIA technology through Arm’s vast network.” So I’d expect NVIDIA GPU and NPU IP to become available quickly to smartphone, tablet, TV and automobile SoC providers as quickly as possible.
When asked what would change, I liked Huang’s response as it is the first place I would go. He said, “What will change is the rate of our roadmap. We know for sure that data centers and clouds are clamoring for the Arm microprocessor, the Arm CPU. Energy efficiency directly translates to computing capacity, computing throughput, and the cost of provisioning service.” There’s that big core CPU I talked about. Arm has done an exceptional job with its stepped-up datacenter investments and I view AWS’s Graviton2 as the highlight so far in general-purpose computing. While Huang made it very clear its GPUs will continue to support x86 and POWER platforms, I can imagine more NVIDIA datacenter GPUs connecting to Arm-based CPUs in the future. It’s a mammoth opportunity.
Arm IP customers jumping ship?
As for the notion of Arm IP customers jumping ship to something different? I believe Huang is a pragmatic person, is making a monumental $40B investment, and isn’t interested in messing with what made Arm great. Sure, the combined company will test the waters on a few things, but only if its Arm licensees don’t “go big” after a market segment.
I think many will misjudge how Huang will operate Arm. I believe he respects the Arm business model and doesn’t want to mess with it.
Passing regulatory scrutiny
As for the notion of this deal not passing regulatory scrutiny? First off, Arm licenses IP and NVIDIA makes chips, two very different business models which, when combined, don’t appear to create a monopoly. NVIDIA GPUs have 0% smartphone, 0% TV share, and a very small GPU IoT share. Arm has 0% datacenter GPU, 0% PC GPU market share, and small Chromebook GPU market share. Having watched Huang get quick approval for Mellanox, even in China, gives me confidence the company can hammer this one through.
I fully expect NVIDIA’s competitors to object hard to regional regulators which could result in some agreements and special requirements, but I think this deal goes forward.
I believe the NVIDIA-Arm deal is an incredibly exciting one that has the combined entity operating in every conceivable market and compute unit. If you believe there will be trillions of IoT end points and AI is important to shepherd the power on the device, the edge, or the datacenter, then this deal should excite you, too.