Archives For computer chip

The Federal Trade Commission (FTC) on Dec. 2 filed an administrative complaint to block the vertical merger between Nvidia Corp., a graphics chip supplier, and Arm Ltd., a computing-processor designer. The press release accompanying the complaint stresses the allegation that “the combined firm would have the means and incentive to stifle innovative next-generation technologies, including those used to run datacenters and driver-assistance systems in cars.” According to the FTC:

Because Arm’s technology is a critical input that enables competition between Nvidia and its competitors in several markets, the complaint alleges that the proposed merger would give Nvidia the ability and incentive to use its control of this technology to undermine its competitors, reducing competition and ultimately resulting in reduced product quality, reduced innovation, higher prices, and less choice, harming the millions of Americans who benefit from Arm-based products[.]

Assuming the merger proposal is not dropped (it also faces tough sledding in the European Union and the United Kingdom), findings of fact developed at the FTC administrative trial scheduled to begin next August will shed light on the robustness of the complaint’s allegations. Without waiting that long, however, and without commenting on the FTC’s theory of competitive harm, it is useful to take stock of the substantial efficiencies that may be associated with the merger, which can be gleaned from the public record. (The following discussion draws primarily on four sources, see here, here, here, and here.)

The Proposed Merger and Its Efficiencies

Arm has been a key player in the development of next generation processors for the better of the last 30 years. Arm-based processors can be found in most mobile devices, from mobile phones and tablets to some computers. Their ubiquity stems from their power, efficiency, high speed, and low cost. Part of this low cost comes from Arm’s licensing scheme, whereby Arm itself does not produce or deliver any semiconductors. Rather, it licenses their intellectual property to other businesses, allowing those businesses great freedom of implementation in return for zero manufacturing risk for Arm. This means that neither consumers nor businesses can buy an Arm processor to put into their computers, and there is no such thing as an Arm-branded processor. Companies use Arm’s technology to develop, refine, and manufacture their own processors.

Nvidia, also a long-time player in the microprocessor space, takes a decidedly different approach to the semiconductor market, manufacturing and selling its devices both to end users and business alike. Nvidia graphics cards (GPUs) are integrated into various computing machines, from consumer laptops to data-center servers, and all carry Nvidia branding. This approach places significantly greater manufacturing risk on Nvidia but allows for significantly greater control over the integration and operation of their products. Since Nvidia undertakes development of optimization and compatibility in-house, it can ensure that its GPU technology works similarly across devices, a step that Arm does not take.

Additionally, there are two ways in which outside companies may interact with Arm’s IP. The first involves buying the right to produce a stock processor and modifying it to suit the business’ needs. This is the less expensive option and allows businesses to undertake the bare minimum of research and development to make their product work. Arm supplies them with the specifications to manufacture the processor, but the optimization and compatibility testing is the responsibility of the end business.

The second avenue is by purchasing what is known as an “architectural license,” giving the business rights to the underlying processor technology and coding language, but not a processor design. In those cases, the end business designs a processor from scratch, optimizing and integrating as it goes along to make sure the processor is a perfect fit for its device. While this integration generally leads to better results for the consumer, this method requires significantly higher research and development costs, leading to higher prices for the device. 

The second avenue enables businesses to significantly advance the capabilities of their processors beyond what is achievable though an Arm-specific design. Since Arm generally focuses on CPU technology, the integration of the additional components to make the computer work—like the motherboard, hard drives, and GPU—are left to the business. In many cases, these components are pieced together from various sources and may be poorly integrated, leading to lower-powered machines with inferior battery life.

However, businesses like Apple and Samsung have taken advantage of architectural licenses to use Arm processor technology and fully integrate all necessary components to work together seamlessly. This can improve battery life, speed, and efficiency in ways that off-the-shelf components are not capable of achieving. This fully integrated system, called a system on chip (SoC) design, advances computing far beyond Arm’s current offerings and presents a significant competitive threat to the processor market.

Given these circumstances, vertical integration of Arm with Nvidia may present both significant efficiencies and new competition in processor markets. Nvidia, with its expertise in manufacturing and designing integrated systems, may benefit from bringing Arm’s processor design in-house. It could save on licensing costs and use the extra capital to bring fully integrated off-the-shelf SoC designs to the mass market. This could reduce the cost of SoC implementation for computer manufacturers, reduce the time spent designing new computers, and bring the price of computers and mobile devices down for consumers.

Additionally, integration with Nvidia would allow Arm to keep pace with the wave of innovation from Apple and Samsung, among others. Those companies are making significant strides in the mobile-computing market, designing smaller, faster, and more energy-efficient processors that can be put into just about any form factor. Arm is significantly behind the curve when looking toward the next generation of processing technology. Integrating with Nvidia may be what the company needs to become competitive in the years to come.

One argument against allowing the merger to be completed is that Arm is a critical trading partner with nearly every processor manufacturer in the market, including Nvidia. Up to this point, Arm has not been owned by a single manufacturer and has not had an incentive to prioritize working with one manufacturer over another. Should the merger go through, Nvidia would own Arm, including the IP used by other companies, leading to concern from the FTC and other international regulators that Nvidia will be able to foreclose rivals from critical IP.

There is a strong counterargument, however, that Nvidia would be going against its own interest if it seeks to foreclose the market. Arm-based processors have become a dominant processor technology in recent years, integrated into 90% of the mobile-device market and nearly 34% of the overall computing market. This guaranteed revenue stream is a gold mine for the company, amounting to nearly $2 billion annually.

Closing the door to this revenue stream by revoking access to Arm’s IP would surely come back to bite the newly merged company. Foreclosing IP would have the effect of raising prices and reducing the quantity of processors in the market, but also would likely force the market to shift away from Arm-based processers over time. Arm already has been forced to reduce the cost to license its technology in recent years in order to stave off competition from open-source chip designs that are available without a license. Doing anything that impacted the overall computing market would harm consumers, businesses, and the newly merged company alike. Denying IP to the broader market would likely not pass an internal cost-benefit analysis for the merged entity.

Conclusion

We do not express an opinion on the ultimate antitrust merits of the Arm-Nvidia vertical merger. We note, however, that vertical mergers are typically procompetitive. Furthermore, information in the public record about the proposed consolidation strongly suggests that it could generate substantial efficiencies that would enhance competition in markets for next-generation computers and mobile devices, in turn benefiting consumers. FTC theories of merger-related anticompetitive foreclosure (which at first blush appear somewhat counterintuitive) need to be scrutinized carefully in light of specific facts, and should be assessed with a jaundiced eye in light of the powerful efficiency arguments in favor of the Arm-Nvidia merger.