Finance Invest is continuing right along, and I am so excited to bring in my next guest. I've followed his entire career at Qualcomm. That's Qualcomm CEO Cristiano Amon. Good to see you, Cristiano. Excited to bring you on here for Invest and talk about — wow, you just launched a big deal with Starlink. My first question to you out of the jump: I mean, could you eventually be supplying a business like Starlink that is going head-to-head with an AT&T, Verizon, and T-Mobile?
Yeah, of course we can. Well, first of all, very happy to be here. Good talking to you, Brian, and thank you for following Qualcomm all of those years. I've been trying to create a new company and we're kind of on our way there. But look, let me answer your question on the SpaceX announcement. I think first of all, yes, recently the company — we've been really focused on doing a lot of automotive and robotics and compute, and we went to the data center. But I will say in all humility, we know how to do cellular communication technology, and actually we started Qualcomm in the very beginning of CDMA days. We also provided silicon for base stations for the RAN business, the radio access network. We have been talking for a while now since the days when conversation started on open RAN, and then most recently on the transition of the industry to next generation, that networks are going to evolve into different kind of architectures with more AI-native, you know, architectures for the infrastructure. Also build the infrastructure more virtualized and more like the way the internet is. We have been building silicon for this and we're very excited. I think what's happening is SpaceX is building a nationwide terrestrial network. It's a 5G nationwide terrestrial network in the United States, one of the largest — among the largest deployments of infrastructure in the United States. We provide the silicon that goes into the unit. Those are not just small cells, including cell sites with 64 transmit, 64 receive, massive MIMO — we provide the silicon for that. We provide the accelerator cards for the digital part of the RAN, and we have layer one software, and SpaceX is the one actually designing the base station. It's a great partnership, and I think it's going to continue to drive the innovation cycle in the cellular industry.
Cristiano, when does revenue start accruing for this business? And then in the same breath, I mean, it sounds like from your vantage point, Elon is serious about being a player in telecom.
Well, I think what you saw announced is, if by all means, if you're going to build a nationwide terrestrial network, that's a very serious investment. And I think SpaceX — if you look at Starlink and what they have done in communication, that's a very, very important telecom business. It's kind of a natural evolution of that.
Cristiano, it's Deirdre, both in San Francisco. It's great to see you. Brian, I hope you don't mind if I just jump in with a follow on that. You know, Elon Musk has these huge ambitions. You're helping execute on them with Starlink, but at the same time, Cristiano, you partner with some of the major mobile carriers that could be threatened by this move. So how have those conversations gone? Have they asked you about it? Is there any worry that this could cannibalize any other parts of your business?
So let's go break that down. It's an interesting question to get. I'm actually — just look at the carrier market. The carrier market is a very competitive market. If you look at today in the United States, you have three carriers that already compete with each other, and we provide chips that go into devices. Now you see in the United States, I think our Snapdragon 8 in Samsung and, you know, Galaxy go to all those carriers. This is a new business of providing silicon to the infrastructure. It is not exclusive. There are different infrastructure vendors out there. I think in addition to SpaceX, we had already built silicon for infrastructure in Japan for NEC, which is kind of building a network for entity Docomo. We have Viettel in Vietnam. You have a lot of carriers. I actually look at this as a little different. I look at this as opportunities for more innovation for all the carriers in the world.
Cristiano, I love that you're talking about infrastructure, and I think it sets it up perfectly for my next question. I mean, when will investors start valuing Qualcomm as a key player in the AI infrastructure build-out?
Look, we're just starting. It's interesting that, you know, if you look at what happened in the market today and all of this concern about data center — the good thing about Qualcomm, there is no AI correction in Qualcomm from a data center perspective because we're just starting. We're absolutely just starting. We recognized that, you know, it was the next opportunity for Qualcomm as we're diversifying the company. We're building a comprehensive road map. We're going to take the long view on this. We want to be a player for the very long term. We're building a road map across four areas: custom ASIC, CPUs, accelerators with differentiated memory architecture, and connectivity, including the transition to connectivity in optics and photonics. We're doing all of that. We have very clear milestones: $5 billion of data center revenues next fiscal year, $15 billion in fiscal 29.
All day today — it was not a great day to report earnings yesterday — but all day today, I've been answering questions from investors and analysts about any risk to your outline, 5 billion and 15 billion.
I'll tell you right now, the 5 billion — I wish I had more supply. I'll do better than five. It's all supply constraint and not demand constraint. The two hyperscaler agreements that we have right now, the Amazon one was announced, and it showed it's a multi-generation, multi-year agreement that is going to take us to the 5 billion. Most of the customers I have today, including Meta for the CPU, get me more than half the way there on the 15. We have our accelerator coming up. So it's a massive TAM. Demand for compute will continue to increase. We're just starting. So we're very confident in our product technology roadmap and those numbers. I think the real answer to your question is now people understand what Qualcomm is doing. They understand that we created multiple avenues for growth in the company. Now, the numbers have to follow, I think, some of those commitments, and we're just focused on the execution. I kind of like, Brian, that most of the uncertainty on Qualcomm is kind of behind us right now. We renewed all the licensing agreement. Apple's coming off the model. For the next three years, we have a good story. We're just going to be executing on it. And then once people see the milestones, I think they're going to give us value for it.
Cristiano, you referred to the market volatility, which has been extreme over the past few weeks, and it feels like some of the fundamentals have changed. I mean, you refer to demand is still huge, we are still supply constrained, but the political backdrop has certainly changed, and there's a lot of concerns about whether the US is going to be able to build out the infrastructure to serve all of this demand. And right as you are sort of transitioning the company to be focused on data centers, how does the next year look when you're trying to get these data centers built and there's been huge pushback at the political level, both federally and at the local level?
Okay. I am going to give you an answer that is a technology-based answer, not a political-based answer, but I look at this in a very simple way. This technology, it's a very disruptive technology. It's almost like going back into the days of the internet. We can all argue about the speed at which the internet got built out, but I think all the predictions in the early days of the internet of what the internet was going to be, they were correct. It just sometimes takes more time, less time — you know, the speed changes. AI is going to be in computers everywhere. Demand will continue to increase. So there is no change to the demand. The only thing that could change are zip codes — where those data centers are going to be built. But the demand for technology continues. It's a global phenomenon and it's not just unique to the data center. I think everything's going to be running AI, it's true for the devices at the edge. We're just seeing the beginning of the inflection point. So from a technology and demand standpoint, nothing changed.
Cristiano, what has to happen? You mentioned that $15 billion target. That's a big target, a longer-term target. What has to go right for you to hit that number?
Okay. So I mentioned briefly before, I'm going to give a little bit more detail. More than half of the number, existing engagements will take us there. So it's like we're going to have five right now, and those customers that are driving the $5 billion are going to continue to grow. And if you look at the projection just on what's announced, we're kind of more than halfway there on the 15, but we have other things that are coming up. We only talk about custom ASIC and the CPU. We have our accelerator coming up. AI 250 is coming up with our HPC memory. Engagement is very good. We have a second engagement on the CPU, and we're building the connectivity in a roadmap. So like I kind of repeat what I said at the beginning: I wish I had multiple gigawatts of contracts for the market to be concerned about. I wish that was the case, but we're just starting. It's a massive TAM. We have been very conservative on our roadmap, and we feel very confident on the 15 just on the current products we're building and in the customer engagement that we've seen so far.
Cristiano, I'd love to ask you about chip design and custom ASICs and the use of AI. OpenAI says it used AI to design Jalapeno. You had the Anthropic CEO talking about using Claude to design chips as well. So if AI is making it faster and cheaper to design silicon, is that a headwind or a tailwind for you guys? Is it both?
Oh, I'll tell you right now. You know, and I'm actually very happy that you asked me this question. It's really a tailwind. You know, in the beginning of this conversation we talked about how Qualcomm is doing so many different things in parallel, and we have done that, I will say, in a very efficient way. We have been some of the early adopters of AI into our workflows and continue to be the case. We continue to see improvements there. The real IP is: do you know what you wanted to build? Do you have the right workflows? Do you have the experience to do it? And then you use AI to allow you to do much more and accelerate the process, accelerate the optimization. So we have seen significant gains and we'll continue to see that. And if anything, it has been an important tool to help Qualcomm accelerate our strategy to diversify and continue to do different things like what we just announced with SpaceX that people were not expecting.
