Arm Holdings plc (NASDAQ:ARM) Q4 2025 Earnings Conference Call May 7, 2025 5:00 PM ET
Company Participants
Ian Thornton – VP, IR
Rene Haas – CEO
Jason Child – CFO
Conference Call Participants
Mark Lipacis – Evercore
Andrew Gardiner – Citi
Ross Seymore – Deutsche Bank
Joe Quatrochi – Wells Fargo
Srini Pajjuri – Raymond James
Sebastien Naji – William Blair
Timm Schulze-Melander – Redburn Atlantic
Vivek Arya – Bank of America
Vijay Rakesh – Mizuho
Lee Simpson – Morgan Stanley
Operator
Good day, and thank you for standing by. Welcome to the Arm Fourth Quarter and Fiscal Year ended 2025 Webcast and Conference Call. At this time, all participants are in a listen-only mode. After the speakers’ presentation, there’ll be a question-and-answer session. [Operator Instructions] Please be advised that today’s conference is being recorded. I would now like to hand the conference over to your first speaker today, Ian Thornton, Vice President, Investor Relations. Please go ahead, sir.
Ian Thornton
Thank you very much indeed, operator. Thank you, and welcome, everybody. I’ll be standing in for Jeff Kvaal today. Welcome to our earnings conference call for the fourth quarter of fiscal year ended March 31, 2025. On the call today are Rene Haas, Arm’s Chief Executive Officer and Jason Child, Arm’s Chief Financial Officer. During the call, Arm will discuss forecasts, targets and other forward-looking information regarding the company and its financial results.
While these statements represent our best current judgment about future results and performance, our actual results are subject to many risks and uncertainties that could cause actual results to differ materially. In addition to any risks that we highlight during this call, important risk factors that may affect our future results and performance are described in our registration statements on Form 20-F filed with the SEC. Arm assumes no obligation to update any forward-looking statements.
We will refer to non-GAAP financial measures during the discussion. Reconciliations of certain of these non-GAAP financial measures to their most directly comparable GAAP financial measures, as well as a discussion of certain projected non-GAAP financial measures that we are not able to reconcile without unreasonable efforts and supplementary financial information can be found on our shareholder letter. The shareholder letter and other earnings related materials are available on our website at investors.arm.com.
And with that, I’ll turn the call over to Rene.
Rene Haas
Thank you, Ian, and welcome, everyone. Q4 marked a record-breaking close to a strong year for Arm, driven by strong demand for power-efficient AI compute from cloud to edge. We crossed a major milestone in Q4 revenue exceeding $1 billion for the first time ever in our history. For the full year, revenue topped $4 billion and royalty revenue surpassed $2 billion, also a first.
We delivered record royalty of $607 million this quarter, reflecting the growing value of every chip shipped with Arm inside. And licensing revenue hit an all-time high of $634 million, driven by new deals, including a multi-year AI partnership with the Malaysian government. Our royalty growth is broad-based, comes from all major markets, data center, automotive, smartphones and IoT, showing the strength of our diversification strategy. Arm is now increasingly the first choice for AI cloud deployments. We expect up to 50% of new server chips at hyperscalers to be Arm-based this year.
Arm NVIDIA’s Grace Blackwell, Armv9 is now in full production. Google’s Axion Armv9 now deployed in 10 regions used by 40 of their top 100 customers, offering up to 65% better price performance than current-generation x86. And Microsoft Cobalt 100 supports major workloads for Databricks, Siemens, Snowflake and internal services like Teams and Copilot. And over 50% of new AWS CPU capacity in the past two years is powered by Arm-based Graviton.
We’re also seeing strong momentum in custom silicon with companies turning to Arm for CPU, GPU and NPU solutions. This is driving both license and royalty growth. NVIDIA’s AI desktop, DGX Spark powered by the Grace Blackwell superchip with Armv9 CPUs is gaining traction, reinforcing strong demand for Arm based AI infrastructure. Royalty revenue growth was driven by broader adoption of Armv9 CPUs and compute subsystems in smartphones. Our smartphone royalties jumped 30% year-on-year, far outpacing the modest 2% growth in shipments, proof of our rising value per device. We launched the first Armv9 edge AI platform, combining Cortex-A320 and Ethos-U85 NPU to run billion parameter models adopted by leaders like Infineon, NXP, Renesas, Qualcomm and STMicroelectronics.
GM and NVIDIA announced a collaboration on Arm-based DRIVE AGX platforms for next-generation vehicles. Our compute subsystems are now shipping volume, boosting both mobile and cloud royalty revenue. We also signed our first automotive CSS license with a global EV leader, enabling custom silicon for next-gen vehicles. Our common CPU architecture from car to cloud enables OEMs to deliver cloud features to vehicles.
On the software front, we now support over 22 million developers, the largest such community in the world. Kleidi AI, our core AI software layer, has now surpassed eight billion cumulative installs across Arm-based devices. The AI revolution is accelerating and Arm is at its heart.
And with that, I’ll turn it over to Jason.
Jason Child
Thank you, Rene. We closed a very strong fiscal year with another record quarter. Total revenue was $1.24 billion, which was in the upper end of our guided range. Royalty revenue grew 18% year-on-year to a record $607 million and was above our expectations. This upside was driven primarily by several flagship smartphone launches across multiple vendors based on Armv9 and CSS.
Based on the most recent royalty reports, revenues from smartphones continued around 30% year-on-year, which is well-above analyst estimates of sub-2% industry unit growth. Year-on-year royalty revenue grew strongly across all of our end markets. Licensing revenue increased more than 50% year-on-year to a record $634 million. Much of the licensing strength is tied to demand for Armv9, our most advanced CPU technology or AI more broadly. In addition, we continue to diversify our customer-base with a large multi-year agreement with the Malaysian government to accelerate the development of an Arm-based AI ecosystem in the country.
License revenue varies quarter-to-quarter due to normal fluctuations in timing and size of multiple high-value license agreements and contributions for backlog. As always, we recommend that you look at annualized contract value or ACV to best understand the underlying licensing growth rate.
ACV in Q4 was up 15% year-on-year, which was at the high end of our recent run rate of low-teens and is above our long term expectation of mid to high single-digit growth. Remaining performance obligations or RPO, was down sequentially as Arm recognized revenue associated with licenses signed in prior quarters. As you know, Arm’s revenues today come from technology developed years or even decades ago, and our costs today are investments for future revenue streams.
In the fourth quarter, R&D spending led our non-GAAP operating expenses to $566 million. Operating expenses were slightly lower than expected as the timing of some expenses will fall into Q1 of fiscal ’26. This translated to a record $655 million of non-GAAP operating profit and non-GAAP EPS of $0.55, which was at the high end of our guidance range of $0.48 to $0.56.
Let me spend a moment on tariffs and the macro uncertainty. Based on our current visibility, we expect a limited direct impact on our royalty and licensing revenues. We have less visibility into the indirect impact on end demand. In our royalty business, we estimate that 10% to 20% of our revenues stems from shipments into the US. In licensing, we have found in the past slowdowns such as COVID that impact is minimal as our customers invest through near-term slowdowns given lengthy chip development timelines.
Finally, based on commentary from our partners, we believe the impact of Poland demand on our royalty business has been limited.
Turning now to guidance, our guidance reflects our current view of our end markets and our licensing pipeline. For Q1, we expect revenue between $1.0 billion and $1.1 billion. At the midpoint, this represents revenue growth of 12% year-on-year. We expect to start the year with strong 25% to 30% royalty growth. As a reminder, Q1 ’25 benefited from very strong licensing quarter — a very strong licensing quarter, which presents a difficult comp for Q1 ’26. Even at the low end, Q1 guidance equates to our second highest quarter in revenue ever. As previously mentioned, revenue growth today enables us to increase investment in R&D essential for our long-term success. And we are accelerating investment in our next generation of technologies. We expect our Q1 non-GAAP operating expense to be approximately $625 million. This includes the impact of some Q4 expenses that will now fall into fiscal ’26. We expect non-GAAP EPS to be in the range of $0.30 to $0.38.
Given the uncertainty in the global trade and economic picture, we have lower visibility than is traditional to start the year. As a result, we do not consider it prudent to issue full year guidance. Nevertheless, we reiterate our confidence in healthy growth in the coming year and years to come. Our confidence stems from our visibility into customer design pipelines, contracted royalty rates, rising demand for custom silicon and AI from the edge to the cloud. Given this view, we expect to continue to invest in R&D aggressively to support our customers and partners. This is a moment to press our advantages to ensure AI is everywhere and runs on Arm.
With that, I will turn the call-back to the operator for the Q&A portion of the call.
Question-and-Answer Session
Operator
[Operator Instructions] And your first question comes from the line of Mark Lipacis from Evercore. Please go ahead.
Mark Lipacis
Great. Thanks for taking my question. Jason, just wanted to start with you on the tariff side. I just want to make sure we understand very clearly. So there’s — I’m assuming that there’s no real impact on the cost side and any impact on the revenue side would be indirect. And you talked about licensing issues not being kind of a big deal on your side. It would be just based on the COVID example. But on the royalty side, it would just be — is it an impact that potentially hits you if the cost of the end devices go higher. Is that all we’re talking about here? If you could just share that framework, spell that out, I’d appreciate it. Thank you.
Jason Child
Hey, Mark, thanks for the question. So yeah, you largely got it right. And that is we don’t have direct impact on tariffs since tariffs today are really applied to end products and are not applied to services, which is what we’re providing. That said, we haven’t seen any impact on tariffs — from tariffs in Q4, haven’t seen any impact and don’t expect any impact in Q1. For the remainder of the year, as you said, it’s really going to be on the indirect side. Is there going to be some sort of impact as supply chains may be affected as they react to whatever tariff impacts eventually work out to be. So as a result, we don’t have as much visibility from our partners as we may have had in the past. But I think overall, the key thing to remember is, if you look at royalties right now, we grew 18% year-on-year in Q4. We’re actually taking up our guidance in Q1 to 25% to 30%. So from our perspective, all the business fundamentals still look very strong even with this kind of uncertainty around whatever tariffs will be applied and exactly how they’re going to be executed.
Mark Lipacis
So it’s a negative — just to be clear, it’s a negative demand elasticity if the cost of the end unit goes higher. That would be the indirect impact?
Jason Child
That would be the likely indirect impact on kind of 10% of our royalties that are in the US. So if it’s a 10% or 20% impact on demand, then it could be a couple percent impact on royalties probably at most. But again, until we know exactly how this is going to apply, it’s hard to know exactly.
Mark Lipacis
Got you. Thank you. Very helpful.
Jason Child
Thank you.
Operator
Thank you. Your next question comes from the line of Andrew Gardiner from Citi. Please go ahead.
Andrew Gardiner
Hi, good afternoon. Thank you for taking the question. Just a quick clarification. You guys haven’t quoted the v9 percentage in your release or in your prepared comments. It looks like from the presentation, it did indeed step up. I think it’s been steady for a few quarters at around 25%. It looks like from the slide, it stepped up a bit to 30%. Is that correct? And can you give us a bit of detail as to where you’re seeing it? Is it indeed the smartphones or perhaps data centers contributing to that as well? Just a bit more detail there would be helpful. Thank you.
Jason Child
Yeah. So you’re right. It did step up to north of 30%. So, increase from the 25% it had been in for the previous few quarters. So where the band is coming from, yeah, I would say, first of all, CSS, which is now starting to become a more meaningful part of our royalties. All the CSS that we sell are all in v9. So this last quarter, I think two quarters ago, we saw the very first instances of CSS and now those are becoming more material. Those CSS are, we have one in client and then we have one in infrastructure. We expect royalties from more CSS to come later this year. And so overall, I would expect that that will become a material driver of our growth. And in fact, one of the things we mentioned was we’re not going to give the v9 adoption rate every quarter going forward. And that’s because as you’ve noticed for the past couple of quarters, it’s gotten a bit noisy. The whole reason we wanted to give that metric was to try to make sure that you could have some correlation to what future royalty growth could be. We’re now seeing that, I think, CSS adoption, which is still very, very early days. As CSS adoption and royalty start to flow through, you’ll see growth probably be more tied to that. But overall, the fundamentals look to be very much in the early days on CSS, but have confidence that we’ll continue to see increasing penetration of CSS as a percentage of royalties along with v9 throughout this year.
Andrew Gardiner
Thank you, Jason. And just quickly related to that, I mean, just in terms of the newer deals you’re signing with CSS, the pricing still holding sort of roughly a double in terms of the rate relative to a more standard implementation of v9?
Jason Child
Yeah, that’s correct. And I would kind of add-on to what Rene said last quarter, which is, first generation is at a step change versus v9. Second generation of CSS will be at an increase from the first version of CSS. So there’s really a kind of a step change when you first implement a CSS and then there’s going to be continual increases as you get updated technologies and updated kind of optimized CSS, which are typically going to be rolled-out, depends on the category bit at least annually or semi-annually — or I’m sorry, at least annually at least every couple of years depending on the category.
Andrew Gardiner
Thank you.
Operator
Thank you. And as a reminder, please limit yourselves to one question only and rejoin the queue for any follow-up. We will now go to the next question. And your next question comes from the line of Ross Seymore from Deutsche Bank. Please go ahead.
Ross Seymore
Hi guys, thanks for letting me ask a question. This one is for Rene and a longer-term question. It seems like your strategic progression has been from charging IP per chip and then as you go from v8 to v9, et cetera, like Jason just talked about, the royalty rate goes up. Now you’re moving into the CSS side of things, doing more subsystems. You also in your investor letter talked about moving to signing deals directly with some of your customers. I think in this instance, you talked about an auto EV maker. I guess the question is, is the logical strategic conclusion of the direction you’re heading that you’re going to have those OEMs as your true end-customers, whether they be hyperscalers, auto companies, et cetera? And if that’s true, what does that mean about how your existing customer base, the fabless semiconductor companies would feel about that on the negative side? In the positive side, what would it mean for the TAM increase that you could possibly address as a company?
Rene Haas
Yeah. Thanks for the question. What we are seeing increasingly across the board, whether it’s in the automotive sector, particularly in hyperscalers, but even broadly across other markets, is customization of silicon is a significant way for companies to not only differentiate from a performance standpoint, but unlock some very unique features, whether it’s at a blade or rack or a system-level, for example, in a hyperscaler and/or in an automotive application. Majority of these platforms are moving to Arm, specifically regarding the software base and everything around the software ecosystem. So when you combine the fact that the majority of the software on the platform they’re developing is Arm and the customization benefit that the end OEM is looking for, increasingly, that is driving our relationship with these partners in a much more accelerated fashion. I expect that trend to continue. So I think particularly when you add AI workloads on top of the existing compute workloads that already need to run on these devices, you still need to run an operating system, you still need to run a hypervisor, you still need to run an IVI instrumentation panel. Then when you add-on top of it an AI workload, that increasingly will make the chip not only more complicated to develop, but more time to develop. And as a result, having a direct relationship with us to develop the product with them is the trend. So I expect that to continue. To your question regarding traditional fabless semiconductor companies, I think that market will still exist, but I think you’re seeing a much more of a demand for customization, particularly at the OEMs, largely because as I said, the chips are really complex to build. The software is really intense and most of that software runs on Arm.
Ross Seymore
Thank you.
Operator
Thank you. Your next question comes from the line of Joe Quatrochi from Wells Fargo. Please go ahead. Your line is open.
Joe Quatrochi
Yeah, thanks for taking the question. I guess I just wanted to go back to just trying to understand not giving a fiscal ’26 guide and just trying to understand the context there. It seems like obviously there’s a lot of fluidity in the market in terms of outlooks for where — what could happen to indirect demand implications. But I guess how does that differ to some extent in terms of your visibility into normally giving a fiscal year guide at this time of the year.
Jason Child
I’ll take that. This is Jason. So I would say the difference is typically when we put in a full year guide now, there’s always going to be some questions, and that’s why we’ll have a range. In this case, if you look at all of our partners and customers and then see kind of what they’re predicting and what they’re projecting, almost none of them are providing full year guidance. And so what’s happening then is the amount of signal I’m getting from partners, whether it’s through guidance or from other reporting is just less than I’ve had in the past. And that’s really all on the royalty side. And so overall, I — we do have high confidence in growth. The problem is if I provide a range, I’d have to give you an even wider range than I had last year because not only do I have big deal kind of timing differences that I have to try to account for, which is what our primary range — that’s what our range was primarily adjusting for in the prior years. I now have to have that plus a range on what may happen because of whatever macro impacts may occur due to just all the impacts from the tariff actions. So as a result, we chose to not give a full year guide with a really large range that probably — I don’t know if it would be that helpful. So that was the route we chose to go.
Joe Quatrochi
Okay. Thank you.
Jason Child
Thank you.
Operator
Thank you. Your next question comes from the line of Srini Pajjuri from Raymond James. Please go ahead.
Srini Pajjuri
Thank you. I have a question on the short term and also on the year. In the June guidance, can you speak to the royalty growth by end market, where you’re seeing strength and where you’re seeing any sort of weakness? And then I guess the cloud and networking, that’s the market that’s where I think you seem to be making a lot of progress. I know you said it’s about 10% of your fiscal ’24 sales. Where is it now and where do you see that going by the end of this year? Thank you.
Jason Child
Sure. On the kind of the line-of-business royalty color, I’ll handle that. In terms of the things that have been going strong in the last quarter, we expect to see more of the same this quarter and really for the most part, likely for the year. So, if I look at this last quarter, Rene mentioned, I think in the shareholder letter or we — I think he mentioned in the prepared remarks that we saw smartphones grow about 30%. And that’s certainly much, much faster than kind of what the market is growing. If you look across the entire client business, which also includes PCs and any kind of a device with screens, we’re also seeing strong growth that’s kind of in that zip code of that similar to that smartphone growth in the 30% range. When you look at infrastructure, that has been accelerating as we have seen some of the custom kind of silicon, specifically with the hyperscalers start to get deployed. So that’s been helping drive increasing growth. We’ve seen increasing growth now for the last few quarters and expect that to continue to increase throughout the year, kind of well into the high double digits. The other factor in infrastructure that’s helping us is last year, we had really slow networking business and that was actually down year-on-year. Well, that’s kind of bottomed-out and now we’re starting to see some recovery there. So when you get the combination of that plus the hyperscaler and cloud compute growth as well as things like full deployment of Grace Blackwell, of which all the Grace portion of the Grace Blackwell is on Arm, we have really, really strong demand there and expect that to continue this year. Moving into automotive, we — I think unlike some of the other folks in the automotive space, we’ve seen strong double-digit growth and have for a while. We continue to gain share in IVI and ADAS. Don’t see that changing. And then lastly, in IoT, we certainly have had some, I’d say some slowness last year in IoT and embedded. That is — it looks to be like it’s bottomed out. I don’t know when we’re going to see recovery there. We’re still in solid positive growth territory, but it’s certainly growing, I would say, not as fast as the other three lines of businesses.
Rene Haas
One thing I might like to add to as far as the growth of Arm’s market-share in the data center. As Jason mentioned, now that NVIDIA has transitioned away from the Hopper architecture to the Blackwell architecture, which uses Arm CPU’s Grace, that actually has an accelerant to our overall growth in the data center to general purpose compute. As the data center, AI data centers move to Arm-based silicon for the host node, the leverage from a software standpoint to general purpose compute is quite significant. So we’re already seeing very rapid adoption of Arm in the data center, which we’ve talked about for multiple quarters. But as I mentioned in the opening, the 50% of new server chip designs and hyperscalers be Arm-based is really driven by A, Grace Blackwell acceleration and B, the leverage that it brings us in terms of general-purpose compute. It just makes more sense for a hyperscaler to standardize across Arm. So that’s what’s contributing to the growth there.
Srini Pajjuri
Got it. Thank you.
Operator
Thank you. One moment please for your next question. And your next question comes from the line of Sebastien Naji from William Blair. Please go ahead.
Sebastien Naji
Yeah, thanks for taking the question. It was interesting to see the licensing deal signed with the Malaysian government this quarter. I don’t traditionally think of sovereign government as a target licensing customer for Arm. So could you help us maybe understand the structure of this deal and whether this was more of a one-off or whether we should expect more of these licensing wins at the sovereign level going forward, particularly as things like sovereign AI really start to take-off?
Rene Haas
Yeah. Thank you. The licensing deal we did with the Malaysian government last month that we signed or I think it was March — excuse me, was quite a milestone event. Is I would commend the Malaysian government for being very progressive in their vision. Malaysia for the last 50 years has a long history in semiconductors, primarily around assembly and back-end test. Increasingly, the Malaysian government is looking to diversify their country’s technology footprint and investing in startups, particularly around AI and cloud. So the arrangement between the government and Arm allows start-up companies to get access to Arm technology at the CSS level to rapidly design chips. And a very progressive agreement, as I mentioned, targeted at startups, broadly targeted to accelerate chip development in the sovereign country. While I’m not going to forecast any new deals, I do think there’s going to be a very nice potential knock-on effect of other governments looking at this and seeing potential benefits for them. But we’re super excited and pleased to have done this agreement with Malaysian government. And as I said, they showed great progressive vision and it’s a fantastic outcome for everyone.
Operator
Thank you. Your next question comes from the line of Timm Schulze-Melander from Redburn Atlantic. Please go ahead.
Timm Schulze-Melander
Yeah. Hi there. Thanks for taking my question. Just maybe a bit more color on the royalty progression through calendar ’25, if you could, please. You’ve talked about hyperscalers. Just wanted to ask as we move from Grace to Vera, what would that or what might that change for Arm? And then just as I think about the sequential trends rather than the year-on-year trends, would the June quarter be something where we should expect it to be sort of seasonally the weakest quarter of the year. I know you’re not providing full year guidance, but maybe just sort of some sequential direction color would be really helpful. Thank you.
Jason Child
Yeah, this is Jason. I’ll start on this question. So, first on terms of the shaping of royalty, yeah, we’re again, obviously not ready to give full year guidance. But in terms of how I would expect it to sequentially grow, but pretty strong growth, obviously, as indicated by the year-on-year that we’re forecasting in Q1. But sequentially from Q1 to the remaining quarters, it’s — I expect it to be somewhere in the kind of flattish into Q2. And yeah, I think that is probably seasonally it is one of our kind of seasonally lower quarters. And then I expect to see somewhere in the 10-ish — 10% to 15% kind of sequential growth in each of the last two quarters in the back half of the year, somewhere in that range. Obviously, we’ll provide more specific guidance as we get later into the year.
Operator
Thank you. Your next question comes from the line of Vivek Arya from Bank of America. Please go ahead.
Vivek Arya
Thank you for taking my question. I actually had a few kind of related questions on the licensing side. Last year, it was one of the upside drivers. But when we look at Q4, it was modestly below expectation. I’m curious what caused that? Is there a macro impact on licensing? And you did report a fairly strong mid-teens ACV. I think you mentioned it’s better than the mid-single-digit to 10% you had thought before. So is that kind of ACV growth sort of representative of what you might see in licensing for this year? And if you would allow me, anything related to China licensing activity that could change given all this tariff confusion?
Rene Haas
Yeah. Thanks for the question. I’ll just take part of that as a macro statement and let Jason get into the details. For the year, licensing revenue, as I mentioned, was an all-time high of over $600 million. And that was driven by a number of different factors, but chiefly the fact that, number one, CSS demand continues to be extremely strong. Number two, Arm’s growth across all platforms that drive huge software leverage, but probably also around AI, because one of the one of the things we’ve seen specifically around AI is that the demand for more and more compute has only risen. That’s causing customers to accelerate their product cycles to do more sooner. Probably the best anecdote I can give of that, again, is the kind of work that NVIDIA is doing in the data center with Grace Blackwell seeing a refresh of products into the data center at a rate that typically was multiple years, not an annual basis. And that’s really being driven by the fact that these AI workloads require more and more compute and more and more compute means more and more demand for new products and more and more demand for new products means increased licensing growth, which is why broadly, we had a record licensing year in the last year. On the specific detail, I’ll let Jason go through that.
Jason Child
Yeah, in terms of the Q4 performance, the 53% year-on-year license growth, that was very much kind of where we had expected to come in. So I’m not sure, I thought consensus was actually slightly below that number, but whatever. I think in general, we are very happy with the progress. And I think if you look at full year, license grew 29%. Now, that’s pretty big and certainly well ahead of what we had expected and kind of guided to maybe at the beginning of the year. Now, I would say for this next year, we’ve said in the past that we expect and I think I said in the prepared remarks, we expect ACV to be kind of a better way to forecast what maybe annual growth should look like because it kind of takes out some of the lumpiness. We’ve said in the past that we were typically looking to mid to high single-digit growth as being the right target. I think that’s still the right target. ACV right now is running 15% because we’ve been overachieving very much due in part to AI. We’ll see how this year progresses. We may be able to exceed that and stay in that double-digit range that we’ve been in. But for now, I would still use that kind of mid-to-high single-digit range as the right growth rate.
Vivek Arya
Thank you.
Operator
Thank you. Your next question comes from the line of Vijay Rakesh from Mizuho. Please go ahead.
Vijay Rakesh
Yeah, hi Rene and Jason. Just a quick question on the CSS side. On the 13 customers, what’s the split of data center versus mobile? I saw you had the first auto CSS in there as well. And I have a follow-up.
Jason Child
Yeah, the breakout we said is it’s roughly — this last quarter, we had the first CSS. So we have 13 now. We had 12 as of last quarter. I think we said last quarter that it’s roughly half client, which is mostly mobile. There is some PC in there, but then also then the remainder is all infrastructure. So think of it as six clients, six infrastructure, one auto basically is where we’re at right now.
Vijay Rakesh
Got it. And then just longer term, when you look at the Stargate and Crystal Intelligence actually where you guys work with OpenAI and NVIDIA and others, can you talk to what the opportunity is on the license side or the royalty side? And what — how is that roadmap looking? Are you guys brining chip-on-an-ASIC? Can you kind of color that a little bit? Thanks.
Jason Child
Yeah, the Stargate project, which was announced in January is going incredibly well. I think just given all the advances you’ve seen through OpenAI’s technology with ChatGPT and the reasoning models, the need for more-and-more compute is quite evident. We were announced that day as a technology partner, which was primarily around Grace Blackwell, which will be the initial chips involved. So right now, that’s all we’re saying publicly Vijay, around our technology deployment there. But we’re very, very excited about the opportunity because it’s obvious from just what you see with the benefits of these algorithms, it’s very, very early days and to get more and more benefit, you need more and more compute of which we’re happy to provide. And lastly, I would say that even though Stargate is a lot of power and it’s a lot of compute, energy efficiency really, really matters, which is why Arm is the only CPU provider inside of Stargate.
Vijay Rakesh
Got it. Thanks.
Operator
Thank you. We will now take our final question for today. And your final question comes from the line of Lee Simpson from Morgan Stanley. Please go ahead.
Lee Simpson
Great. Thanks for squeezing me in here. I just want to ask, Rene, maybe a more broad question on the chiplet space. It does seem according to reports that Arm has been associated with making acquisitions in the chiplet space as has some of the key licensees of the Arm architecture. I think when you look at the chiplet system architecture, the AMBA fabric that you have, it does look as though this is an increasingly important space for you. So I’m trying to understand why is that? Is it a waypoint perhaps to the eventual own chip-making Arm or is there something else that we’re missing here as a big value play? Thanks.
Rene Haas
Yeah. So, regarding unannounced products, I’m not going to announce anything here. But what I can say regarding these very, very large SoCs that use multiple chips, aka chiplets, having a common bus for interfaces on and off a die is table stakes. AMBA is the de-facto standard for interfacing in the Arm CPU at the chip level. So it’s only natural that you’re going to see those standards provide themselves as default as you get into the chiplet space. Chiplets are also frankly one of the large-value propositions around doing CSS. If you can imagine some of the chiplets that are being built that use CSS, the vast majority of the silicon area and transistors that are on the chiplet are actually using RMIP, which is why the earlier question regarding custom chips that were being done by OEMs, what’s really driving that, the chiplet approach is a component of it and the CSS is a big component of it. So to your point about having AMBA bus and connectivity, it all — it all hangs together.
Lee Simpson
That’s great. Thank you.
Operator
Thank you. That concludes the Q&A for today. I will hand back to the room for closing remarks.
Rene Haas
Thank you, and thank you for all the questions. They were very, very spot-on. Just in summary, we are so thrilled to have completed our fiscal year with records of $1 billion-plus quarter, $4 billion-plus in revenue and record license and record royalty. I think what you’re seeing is the continued manifestation of the strategy we’ve talked about for the last number of quarters and years, that is Arm is everywhere. Increasingly demand for the Arm architecture is requiring us to deliver more. We’re seeing that with our compute subsystems. And with the advent of AI workloads running in the data center, running on a PC, running on a smartphone, running on the automobile or even running in earbuds, the demand for Arm technology has never been greater. So we are incredibly excited about the future. AI is changing everything and you can’t run AI without Arm. Thank you.
Operator
Thank you. This concludes today’s conference call. Thank you for participating. You may now disconnect.