Will Arm’s leadership in lower power chips be challenged?
/One of the most important technology companies in the world is often times one of the least discussed and debated. Arm Holdings (or just Arm) is a UK-based company founded in 1990 that has been the silent leader in the expansion of computing outside the world of servers and PCs. If you have a smartphone, a connected appliance, a smart TV, or digital home assistant, you are working with a piece of technology with Arm at its core.
Arm has an interesting business history. It was purchased by the Japanese conglomerate SoftBank Group back in 2016, had an acquisition attempt in 2020 from Nvidia that failed regulatory hurdles, and finally went back to being a publicly traded company on Nasdaq in August of this year. The company just announced its first quarterly earnings report after its IPO, with the market generally pleased with the last 3 months of revenue but wary of the guidance given for the next quarter.
In my view, there are interesting reasons to believe that Arm has great years ahead of it with expansion into new markets and new revenue streams, but also a few key headwinds that put long term growth at risk.
Strengths
Arm is most often associated with the explosion of the smartphone market, and for good reason. With few exceptions, at the heart of every phone today is a chip based on Arm technology, including those from Apple and Samsung. While the handset space has been problematic for companies recently with the softness of sales in China, indications are that the region is picking back up, with a note from Qualcomm in its latest earnings to that point. A reinvigorated smartphone market, with the help of on-device AI adoption, will improve the outlook where the company’s largest revenue stream is tied.
There are many areas of growth for Arm outside of the smartphone market. The PC space is getting interesting as we move into 2024 as Qualcomm announced a new Snapdragon processor last month meant to displace Intel and AMD that use x86-based processors for their laptop platforms. Qualcomm’s success here turns into success for Arm as the new Oryon core powering the chip is based on the Arm architecture. And recent rumors of Nvidia and AMD entering into the Windows-on-Arm chip market gives the Cambridge-based company another avenue to take advantage of this silicon diversification direction from Microsoft.
As the automotive market becomes more about computing and assisted driving, Arm-based products are going to be at the heart of that transition. Qualcomm is one of the leaders in this space as well, mostly utilizing Arm technology. But other Arm partners like NXP, Renesas, and Cadence are involved in the rollout of automotive technologies from ADAS to digital cockpits, and all are integrating Arm processors of some kind. This segment will be growing at nearly a 10% CAGR for the next 5+ years and offers a significant revenue opportunity.
I would be remiss if I didn’t also mention the progress Arm has made in the data center segment with its Neoverse family of IP that is targeted at powering cloud and edge server infrastructure. As more data centers look to find ways to improve or maintain performance while lowering fixed costs of power and space, the advantages that Arm CPUs have in efficiency and scalability shine. And while the data center segment has a been a target for Arm since before the AI revolution really hit us in the forehead, it’s significant partnership with Nvidia, the clear leader in the enterprise AI race, means that Arm-based products like the Nvidia Grace CPU will drive revenue and relevancy.
As part of its recent earnings, Arm said that more than 7.1 billion devices had shipped this past quarter with Arm technology inside them. That’s a stunning number and one that tells you how deeply embedded and how pervasive the Arm architecture is in our connected world. We aren’t just talking about companies like Apple, Qualcomm, and Samsung, but also ones like Toshiba, NXP, and Tata Communications. Even Intel and AMD are utilizing Arm designs for some portion of their product lines.
Another interesting note is that Arm CEO Rene Haas mentioned in an interview with Jim Cramer this week that the strong increase in licensing revenue that Arm saw in this quarter’s earnings report is a “strong indicator for R&D investment.” That is good news for Arm – the more product development that is happening today on its IP means that future products released to market will be based on that R&D work, cycling back to more revenue and market share for Arm.
The Risk
Perhaps the biggest risk to Arm is competition. Not from the world of x86 processors that are simply trying themselves to not be displaced by Arm designs, but another low power architecture. RISC-V (pronounced “risk five”) is a competing instruction set architecture (ISA); that basically means it is based on a different set of computer microinstructions, not compatible with either Arm or x86 designs.
RISC-V is an “open” design, meaning that, at least in theory, it is free to utilize for both academic and commercial use cases. This is obviously a benefit over the Arm architecture where companies must pay a licensing or royalty fee to design their own Arm CPU or to use one of the cores designed by Arm itself. This openness also means that companies and the RISC-V community are encouraged to share best practices for designs, improving performance and time-to-market.
Many tech companies are already using RISC-V for some of their platforms. Qualcomm’s Lu Dai sits on the RISC-V board, as does Nvidia’s Frans Sijstermans. Other representatives from the likes of Google, Huawei, Seagate, and even Intel are there too.
Qualcomm announced a RISC-V based wearable platform, and Google announced OS support for it, just last month. You can find presentations and papers from Nvidia as far back as 2016 referencing work on a RISC-V CPU core, and call outs to “areas of interest” for RISC-V as automotive, silicon emulation, and security. AI-startup Tenstorrent, led by chip guru Jim Keller, is using RISC-V in its design and has signed up Samsung to manufacture it.
So how big is this existential risk to Arm? The truth is that migrating or building a RISC-V core is a very heavy engineering lift, and companies that were building RISC-V cores like SiFive have recently cut 20% of their workforce. And though I don’t have specifics on the pricing model, if a customer is going to license a core based on RISC-V from a company like SiFive, one of the primary advantages over Arm (cost) seems to fall away.
Arm should take this risk to its business model seriously, and it appears to be doing so. There have been shifts in programs offered by the company including a “flexible access” model that offers no-cost access to Arm IP for companies that are in the startup stage, clearly targeting many potential customers of the RISC-V ecosystem.
The other big risk I see for Arm is finding a way to get credit and recognition for its value in the tech space. Companies like Intel, AMD, and Qualcomm themselves struggle to maintain brand recognition and brand value as a component company in the products and services that people use every day. Qualcomm makes the modem in nearly every flagship smartphone today, and Intel powers 75%+ of the laptops sold each year, but very few consumers will recognize that fact or even really care. Arm is even one more step removed for that – Arm’s designs power the Snapdragon processors that enable flagship Samsung handsets. Getting credit for that enabling work, from the investor market as well as the consumer (and to some degree a share of wallet) is a difficult task.
These headwinds aside, the next 5+ years for Arm look strong. It’s leadership in the smartphone market will not be challenged and the areas of growth, from PCs to AI to automotive, offer significant revenue upsides.