You are watching Bloomberg Tech. It powers the vast majority of smartphones worldwide and a growing share of AI data centers. Arm, the company underpinning the chips used by Apple, Nvidia, and Microsoft. The UK-based firm is making one of the boldest moves in its history. Arm is going from purely designing chips to making them in a bid to command AI like it's dominated smartphones. It is a high-stakes shift, but the company's chief executive says it could drive strong growth for years to come.
We are shifting away from being largely known for smartphones to being around the cloud and AI data centers, where in a few years, that's our largest business by far. In five years, it will be orders of magnitude larger than what we do in smartphones.
Will Arm's big AI gamble pay off? We put that question to its CEO. We will hear from investors and industry experts on the future of this unique company. Welcome to Bloomberg Tech Europe. This is Cambridge, England, home to Arm, one of the world's most influential chip companies, founded in 1990 with a radically different vision from Silicon Valley's chip giants. Instead of making processors, it would design them, focusing on efficiency, not just raw power. That approach meant Arm's devices could be used in any device and anywhere. When smartphones took off, Arm's designs were already built in. Today, nearly every smartphone in the world, from Apple to Xiaomi. Arm went public at the height of the dot-com boom. 18 years later, it was taken private for $32 billion by Japan's SoftBank. A former Nvidia executive took over as CEO in 2022 and returned Arm to the public markets just a year later. This time, Arm listed in New York only, a major blow to London. Now Arm is making one of the biggest strategic bets in the company's history. The company recently unveiling its first in-house processor, the AGI CPU, designed for artificial intelligence data centers, marking a move from licensing into selling silicon itself. From smartphones to the infrastructure powering AI, Arm's stronghold on global computing continues to grip. Arm may be listed in the US, but it is very much a company with global reach. The company sits at the center of Masayoshi Son's ambitions to take on the AI hyperscalers. Its parent SoftBank owns more than 90% even after Arm's IPO. I traveled to San Francisco to speak with the CEO about the company's pivotal shift.
What's happening inside Arm is we are shifting away from being largely known for smartphones to being around the cloud and AI data centers, where in a few years, that's our largest business by far, and in five years, it will be orders of magnitude larger than what we do in smartphones.
For the first time, Arm is designing its own chips, its own CPUs, the whole package. And you're working with Meta. How significant is that for this company as a strategy?
I think it's very significant. When we looked at the opportunity, particularly in the data center with the growth of AI, but more specifically agentic AI, and agentic AI quadruples the demands on CPUs. This is a market that is heavily underserved at the moment. Number two, customers are asking us, 'What, you deliver and build a chip for us?' To your point, yes, we now have our first product. It is Arm's product, the Arm AGI CPU, designed by Arm, built by TSMC, sold to customers like Meta or SAP or Cloudflare, OpenAI, a litany of companies that we've announced, and the demand has been extremely strong.
What is the potential size of this new revenue stream, of this new market for you?
When we look at the royalties in our traditional business model associated with the cloud and AI business, it's about a $3 billion TAM. When we look at the potential of this product line maybe out for five years, it is north of $100 billion. So it is profoundly different in terms of the scale. It puts us in a whole different zip code.
Semiconductors increasingly become a tool of geopolitics as well. How challenging does that make your role as CEO of a global company that has that skill?
Five years ago, it wasn't something we worried about. We may have had people inside Washington or working with people in Number 10 to make them aware of the technology, but it wasn't top of the radar. Visiting Washington or getting to know the folks inside DCMS inside the UK, that was not something that as a CEO you had high on your priority list. That has changed.
Do you think government understands where the real leverage in the chip stack?
I think they are increasingly learning that. I would say 10 years ago, not so much, but they've certainly added lots of capability in terms of people. If you look at the US government in particular and the current administration, they have added David Sacks, the AI Tsar. He's put a tremendous amount of people to go off and learn that. Same thing under Peter Kyle and Liz Kendall in the UK.
You are stewarding this technology that underpins billions of devices. Do you personally feel the weight of that responsibility? Are there moments when you think, wow?
When you're in it on a day-to-day basis, you don't really think about that kind of scale. But sometimes when I step back and when we are talking with teams preparing materials for launch, and we try to come and get the scale of what Arm does, we come up with a number of 350 billion chips. People like to talk about how many chips does a certain person use? 350 billion chips. There is 117 billion humans that ever lived on Earth, ever, and we are three times that number.
That scale is enormous. One of the fascinating parts of the Arm story is the nature of the business. Headquartered in Cambridge, United Kingdom, majority owned by SoftBank, publicly listed in the US. How do you navigate having to meet quarterly earnings with Masayoshi Son's grand sweeping visions about how this kind of tech plays out?
I have a fortunate job in some senses in that SoftBank, Masa, is a 90% owner of Arm. I have investor meetings quite frankly with my chief shareholder. He is very long on Arm, hasn't sold a share since we went public. He is very long on the potential. That helps make my job in some ways easier in the sense that he has a long-range vision, as do I. Technology, we always know, is a long, long game. It is not a real quarterly game. That allows me to have some freedom to think about the long-term without worrying about a wide cadre of investors. Thankfully, the quarters have been very good and have met or raised guidance nearly every quarter since we went public.
What's going on with Stargate? OpenAI, Oracle seemed to be pulling away with some of their commitments. Is that starting to falter?
Not at all. We put a large investment to OpenAI the last round. They are going to be using a lot of that capital for more compute. OpenAI is a magnificent company. I think the growth potential is literally unbounded. Compute is what will limit them, so they are needing more and more compute. There's been a lot written about Stargate, but Stargate is a complex program. I think the noise made around the Stargate thing stumbling has been a bit overblown.
Apple is a big customer, quite clearly, of Arm. Has Apple lost the AI race or are they going to be a fast follower? What's your take?
Too early to tell. We are still in very early days of this AI race.
Where do you stand on the question of AI and jobs? In a company like Arm, are you going to be in 5-10 years' time, are you going to have more employees or fewer?
The way I think about AI eliminating jobs is when you think about, let's use agentic AI as an example. An agent can go off and run a piece of work for you. And one might argue that you, an agent, that's going to eliminate jobs because all that work is now being done by an agent. Being imaginative about how to go, what more can be done? What products can I design? Things can happen faster. Inside Arm, we are pushing it heavily. I am pushing it very heavily not to reduce OpEx. It is to design products faster and design better products quicker and innovate and have more products. Right now, my limitation for developing great products is the amount of engineers that we have. That is my biggest limitation. Right now, AI has not solved that. It is not to replace the engineers because we've got no shortage of ideas. I got tons of ideas that we talk about in our leadership team meetings where we think about, gosh, we'd love to do X, if only. I am much more of an optimist. I think so long as companies can come up with creative ideas on what to do and build, AI is going to suit us quite well.
I spoke to ASML CEO, he was pretty damning about Europe's place in the semiconductor industry. He said Europe has no real capabilities to compete with the US and China. Do you agree? Where can Europe compete when it comes to semiconductors?
I think Europe is falling behind in this space. I think when you think about the amount of innovation that takes place inside either the US or China, Europe has fallen behind. I think some of it isn't actually at the big company level. I think some of it is at the startup company level. In other words, companies that can get early funding, do innovation, attract really smart people to go off and build companies and be entrepreneurial. For whatever the reasons are, there's not enough capital or that capital is not organized well enough or there is something that is making founders move elsewhere. I don't think it's a lack of talent. I think it is some lack of infrastructure to nurture the companies to start in Europe and stay in Europe.
When it comes to the data center story, when do you think you will be able to declare victory? What for you is success? You have again shifted, what for you is success? You have shifted Arm to have a laser-like focus on data centers to build up that market. When can you call that a win, do you think?
On one level, I think it is a win now in that we got real customers, Meta, SAP, Cloudflare, Cisco, as I mentioned. On level zero, it's a win. Going forward, I think there's a lot of opportunity to do much more. The aperture for opportunity for Arm has just gotten much larger. I mentioned north of $100 billion for this AGI CPU product, but when we think about the broader market, it could be north of a trillion dollars that we could address.
Arm CEO Rene Haas speaking to me in San Francisco. Arm's move to sell its, make its own chips has been largely cheered by its investors. One of those investors supportive of the plan is Clare Pleydell-Bouverie, fund manager at Liontrust. Liontrust's top tech investments include Nvidia, Broadcom, and TSMC. She says Arm's switch to produce its own AI CPUs is a significant moment in the company's history.
