Skip to main content

Nvidia’s big ambitions could be its Achilles’ heel in the Arm deal

Nvidia’s big ambitions could be its Achilles’ heel in the Arm deal

/

The chips are down for the $40 billion deal

Share this story

The word “arm” is superimposed on a psychedelic background
Illustration by Alex Castro / The Verge

Nvidia has been trying to buy Arm for $40 billion for over a year now — but this week, the acquisition was hit with its biggest roadblock yet. On Monday, the Federal Trade Commission laid out the case to stop the merger from going through, arguing that the deal would “stifle competing next-generation technologies.” 

It’s the most significant attempt to reign in Big Tech yet under Lina Khan’s term as FTC chair, so there’s a lot at stake, both for the FTC and the electronics industry at large. Arm is a hugely important company; the company’s chip designs touch hundreds of billions of devices, including CPUs and ISPs for modern cars, embedded chipsets for wearable and medical devices, smart home gadgets like thermostats and routers, and of course, smartphone and laptop processors. The question of who controls it will have massive implications for all of them — and the FTC’s case now seems like it’ll be ​​the biggest barrier to the acquisition going through. 

Nvidia already competes in at least some of the same fields as Arm licensees. The FTC and other regulators in the EU and UK are concerned that, because of that involvement, Nvidia could influence Arm’s future product development. It’s particularly worried that Nvidia would use its control over Arm to advance its own interests in emerging markets like data centers and autonomous vehicles, instead of working to ensure that all the companies that use Arm designs to compete with Nvidia in those fields can continue to do so on an equal playing field. 

Crucially, Arm doesn’t actually make its own chips: rather, it sells both licenses for companies to design their own chips that use Arm’s architecture (like Apple’s M-series chips for Macs), in addition to selling entire CPU and GPU designs (like the Cortex-X1 CPU and Mali GPUs found in the Google Tensor and Samsung Exynos 2100). But Arm’s architectures — which offer better power efficiency and customization options for device manufacturers — have steadily become a critical part of the computing landscape. It’s Arm’s technology, not the traditional x86 designs used by Intel and AMD, that has fueled the massive growth of smartphones, tablets, and other mobile devices over the past two decades. 

The FTC makes its case by looking at three specific areas where Nvidia already licenses Arm’s technology: Nvidia’s advanced driver assistance chips and two types of data center chips. One of those data center chips is particularly important: the DPU-based (data processing unit) SmartNIC, a core component of data center networking infrastructure, which provides both processing (thanks to Arm cores) and network interfacing for secure cloud infrastructure. Nvidia thinks they’re going to be a big deal, setting up that DPUs are set to be the “third pillar” of computing infrastructure, alongside traditional CPUs and GPUs. And virtually everyone that makes DPU SmartNICs — even Intel, the biggest x86 chip company around — uses Arm technology. 

Even Intel, the biggest x86 chip company around — uses Arm technology

It’s a similar case for automotive vehicle assistance. Arm’s technology is used in virtually all the chips for enabling Level 2 and Level 3 assistive features in cars (which covers basic automated tasks like acceleration or lane changing while still having a human operator on standby). With the exception of Intel-owned Mobileye, every major chip in this field uses Arm — including Nvidia.

There are a couple of things to note here. The FTC is being very strategic in its examples, choosing areas that have a clear impact on the broader tech industry. But these three licensing deals also represent a very small part of Arm’s business. As a SoftBank presentation from 2020 shows, its bread and butter is focused in mobile phones (where it dominates more than 90 percent of the market) and IoT application processors. On the automotive side, Arm claims to have 75 percent market share but doesn’t distinguish how much of that is for infotainment systems versus the kinds of driver assistance applications that the FTC is concerned about. And its presence in data centers is virtually nonexistent: Arm’s technology has just a 5 percent market share there. 

The FTC is making a broad case here — that a combined Nvidia / Arm would be willing to make moves that would hurt its competitors that rely on Arm’s architecture or designs, because the profits it could generate from better succeeding in these markets would outweigh the losses from licensees. But it’s relying on very specific parts of the business, which seems to open the door for some kind of compromise that would spin off the conflicting parts of the business while allowing the bulk of the acquisition to go through.

The FTC has a long road ahead of it

In practice, that kind of compromise has been hard to achieve. According to The Information, Nvidia offered to spin out an independent licensing company for licensing out designs in an attempt to mollify regulators, but the plan failed due to concerns that Nvidia would still control Arm’s development of new products. So while Nvidia could try to divest itself of, say, Arm’s data center technology, thanks to the common architecture shared by Arm’s products, it would still be in charge of what kinds of features Arm would focus on to include in those chip designs. 

The FTC also highlights the fact that Nvidia would gain access to Arm’s customer list and be privy to sensitive information that companies share with Arm. “Nvidia’s ownership of Arm would fundamentally upend Arm’s status as a neutral partner and, at the same time, enable Nvidia to obtain access to its rivals’ competitively sensitive information.” The regulatory group also notes that Arm licensees that compete with Nvidia might be less willing to work with the company to help further improve Arm, in addition to skewing future development of Arm’s products in directions that benefit Nvidia specifically, rather than the broader Arm ecosystem. 

Of all the FTC’s arguments, Nvidia seems most prepared for this one: back when the deal was first announced, Nvidia CEO Jensen Huang told the Financial Times he could “unequivocally state that Nvidia will maintain Arm’s open licensing model. We have no intention to ‘throttle’ or ‘deny’ Arm’s supply to any customer.” The company has since reiterated that the merger would “boost competition, [and] create more opportunities for all Arm licensees and expand the Arm ecosystem.”

This puts the FTC in a delicate position as it heads to court. But the specific examples that regulators are using to prove it are very specific and less integral to Arm’s overall business strategy than, say, concerns about competition for its major CPU and GPU products. After decades of relatively lax oversight, it’s hard to say how any antitrust case will fare in court — so even the slightest weakness in the case could be fatal.

On the other hand, the FTC also doesn’t necessarily have to win to block the deal. It just has to hold out long enough that Nvidia decides that the deal isn’t worth the trouble, especially since Nvidia’s deal with current Arm owner SoftBank only gives the US chip company until the end of 2022 to clear regulatory approval. 

Behind all of it is a very nervous electronics industry. Google, Microsoft, Qualcomm, and Tesla have all objected to the deal, and the concerns that Nvidia ownership could influence Arm’s development and stifle innovation are very real. The only question now is whether the FTC can make the case work in court.