In a joint regulatory filing, Arm and Nvidia stress the consequences for Arm's growth if the proposed merger is blocked.
Arm faces significant hurdles to growth as a standalone company if Nvidia’s proposed acquisition falls through, according to documents released as part of the U.K. regulator’s ongoing merger inquiry.
The 29-page document details the joint Arm-Nvidia response to the U.K. government’s decision last November to refer the deal for further investigation to the U.K.’s Competition & Markets Authority (CMA). The response emphasizes that without investment from Nvidia, Arm would be seriously disadvantaged in its bid to grow in data center markets and compete against Intel Corp. and x86 incumbents. The filing also explains why an Arm stock offering is a non-starter while noting that Arm faces stiff competition from emerging RISC-V competitors.
With SoftBank’s investment nearing an end, Arm finds itself at a crossroads, making its acquisition by Nvidia a “unique, once-in-a-generation opportunity to expand and enhance Arm’s ecosystem, benefitting the U.K. and all Arm licensees,” the filing asserts.
In dismissing the IPO option, Arm said a stock offering would suffocate its ability to invest, expand and innovate, noting that capital markets demand a focus on short-term revenue growth and profitability.
“SoftBank considered and rejected an IPO in 2019 and again in early 2020 because the markets would not give SoftBank the necessary return on its investment,” the filing states. “While Arm’s licensees such as Apple, Qualcomm and Amazon have enjoyed skyrocketing revenue growth and profits, as well as soaring market valuations, Arm has lately endured comparably flat revenues, rising costs and lower profits that would likely present challenges for a 30-year-old public company.
“The capital markets would expect Arm to make significant strategic changes, including cutting costs to maximize Arm’s value.”
Moreover, it continues, Arm’s original market and largest revenue source, mobile, is saturated.
“Data center and PC, two markets that SoftBank targeted with its investments in Arm, are far more difficult to crack. Unlike Arm, the x86 incumbents in data center and PC (Intel and AMD) benefit from an established ecosystem of developers, software, systems and peripherals. They are also vertically integrated, enjoying profits generated from multiple levels of the technology stack, allowing them to make massive R&D investments.”
Hence, “any competitor following an IP-only licensing model, like Arm, is at a major ecosystem and economic disadvantage. In addition, Arm does not have the systems building expertise, the software engineering scale, or the R&D resources of x86 vendors like Intel and AMD. Even under the most optimistic projections, standalone Arm could not generate the revenue necessary to invest and compete toe-to-toe with the entrenched x86 incumbents.
“To date, Arm has only managed to achieve limited inroads in data center, mainly licensing to Amazon, which makes custom chips for its own use, and startup Ampere Computing, the only entity that offers merchant Arm central processing units (CPUs) for datacenter.”
The situation is similar in the PC market, the joint filing notes. Arm’s designs lack significant market penetration, and two PC customers have stated they will not use Arm’s cores in next-generation SoCs, instead relying on internal designs.
“Arm has great engineering talent in the areas where it focuses,” the filing continued. “But as a standalone IP licensing business, and without access to further capital, Arm has inherent scale, scope and economic limitations that would impact Arm’s future as a standalone licensing firm.
“As a publicly traded company, Arm would likely not have the financial resources to invest sufficiently in early-stage revenue businesses. Nvidia is particularly concerned that these pressures would drive Arm to de-prioritize data center and PC and to instead focus on its core mobile and growing IoT businesses.”
“The result would be a concentrated CPU market largely controlled by Intel/AMD (x86), with the remainder controlled by powerful and far more profitable Arm architectural licensees….” [The identity of those licensees was redacted from the filing].
The filing also argues that Nvidia would provide not only investment but also data center expertise. “Arm’s limitations as a standalone, IP licensing-only entity, have become apparent over many years. Arm licensees, including Broadcom, Qualcomm and more recently Marvell, have tried and failed to penetrate data center and PC. As mere licensees, they were neither able to direct Arm to make the necessary investment in data center and PC CPU [markets, nor able to infuse Arm with the ecosystem-building expertise it needs.”
“The merged entity would have every incentive, and the ability, to dramatically increase investment in Arm R&D across the board, rather than facing the difficult choices of where to de-invest and face further customer and competitive pressures.”
Nvidia would also provide expertise in SoC design, software, accelerators and system designs, thereby attracting software, hardware and system developers to the Arm data center ecosystem. “Nvidia would port its platform solutions to Arm, and help developers optimize code for Arm-based accelerated systems,” the filing added.
Arm and Nvidia’s businesses “are highly complementary and relate to different levels of the semiconductor value chain. Arm primarily licenses CPU IP for low-power mobile devices, a sector in which Nvidia is not active,” the companies argued.
Arm said it has lost ground to Intel and RISC-V as the regulatory review drags on. “Intel is already competing head-to-head with Arm to supply CPU IP for important customers, targeting cloud service providers.” It has also partnered with Qualcomm on foundry services, thereby mounting a “strong challenge to Arm’s efforts in the data center and PC markets. Customization was Arm’s greatest selling point in the data center, as Arm does not have the mature ecosystem or massive R&D resources that Intel can bring to bear. Now, Arm’s handful of data center customers can create custom x86 CPUs that will benefit from the massive x86 code base.”
The filing also asserts the RISC-V community is exploiting the regulatory delays. “The past year saw a flurry of activity in RISC-V, a threat to Arm in automotive, IoT and SmartNICs. In June 2021, for example, SiFive announced its ‘P550’ high-performance CPU IP based on RISC-V, which compares favorably to Arm’s contemporary CPU IP block (Cortex-A75), in a smaller package. In December 2021, SiFive announced that its next-generation microarchitecture (available in 2022), the ‘P650,’ which targets ‘high-end servers and other applications requiring large arrays of multiple processor cores.’”
Other RISC-V announcements have followed as western regulators scrutinize the Nvidia-Arm deal.
The filing also disputes U.K. regulator’s assumption that licensees such as Apple and Qualcomm demonstrate Arm’s ongoing success. To the contrary, it notes, “Architectural licensees do not use Arm’s CPU designs. Arm architectural licensees create their own proprietary CPU designs using their own engineering teams in the United States.
“As a result, these architectural licensees compete with Arm’s own engineers. Their success will concentrate the market, because they do not license their designs to anyone else. Apple’s industry leading M1 processor, for example, was designed entirely by Apple, not Arm. It is captive and available only to Apple.”
“As a result, the architectural licensees pose a threat to Arm’s implementation IP business,” the filing asserts.
Stop romanticizing Arm’s past
After arguing that the “the merged entity would have no ability to foreclose competition,” the filing closes with this: “Deal opponents romanticize Arm’s past and either ignore or disparage Arm’s most powerful competition. But if Arm had market power, it would have sizable revenue growth and would be enormously profitable.”
Rejecting the merger “would not promote competition,” the closing argument continued. “Rather, it would prevent Arm from bringing competition into areas that have been long dominated by x86. The alternative outcome urged by deal opponents would result in a standalone, profit-maximizing business without any guarantees about licensing policy or investments. It would likely result in less investment in the UK, less resources for Arm, less innovation and less competition worldwide.”
The full Arm-Nvidia regulatory filing is available here.
This article was originally published on EE Times.
Nitin Dahad is a correspondent for EE Times, EE Times Europe and also Editor-in-Chief of embedded.com. With 35 years in the electronics industry, he’s had many different roles: from engineer to journalist, and from entrepreneur to startup mentor and government advisor. He was part of the startup team that launched 32-bit microprocessor company ARC International in the US in the late 1990s and took it public, and co-founder of The Chilli, which influenced much of the tech startup scene in the early 2000s. He’s also worked with many of the big names—including National Semiconductor, GEC Plessey Semiconductors, Dialog Semiconductor and Marconi Instruments.