Thank you for standing by, and welcome to the GlobalFoundries Business Webinar Series, Enabling AI in the Physical World. At this time, all participants are in listen-only mode. After the speaker's presentation, there will be a question-and-answer session. To ask a question during the session, you'll need to press star one one on your telephone. If you'd like to remove yourself from the queue, simply press star one one again. As a reminder, today's program is being recorded, and now I'd like to introduce your host for today's program, Eric Chow, Investor Relations. Please go ahead, sir.
Thank you, Operator. Good morning, everyone, and welcome to GlobalFoundries' Inaugural Business and Strategy Webinar. Today, we will be providing a market, technical, and strategic update on the opportunities for GF across the rapidly evolving physical AI space. We are excited to showcase how GlobalFoundries' MIPS and ultra-low-power solutions are driving the evolution of smart, autonomous, and connected devices across physical AI applications. On the call with me today are Mike Hogan, Chief Business Officer; Sameer Wasson, CEO of MIPS, a GlobalFoundries company; and Ed Kaste, Senior Vice President of our ultra-low-power product line. Today's slide presentation, along with the recording of the call, will be made available on our Investor Relations webpage. Certain statements on today's call may be deemed forward-looking statements. Such statements can be identified by terms such as believe, expect, intend, anticipate, and may, or by the use of the future tense.
You should not place undue reliance on forward-looking statements. Actual results may differ materially from these forward-looking statements, and we do not undertake any obligation to update any forward-looking statements we make today. For more information about factors that may cause actual results to differ materially from forward-looking statements, please refer to risks and uncertainties described in our SEC filings. At the conclusion of our prepared presentation, we will open the call for questions with Mike, Sameer, and Ed. We request that you please focus the scope of your questions to the topics discussed in today's webinar. I'll now turn the call over to Mike.
Thank you, Eric. Let me start by setting the foundation for physical AI: what it is, how it is transforming the world, and what it means for GF and the broader semiconductor industry. When we look across the history of significant market inflections in our industry, we see that each investment cycle follows a familiar pattern. Infrastructure enables new applications, and those new applications transform our world. Each tectonic shift is defined by the exponential growth in connected devices and the complexity of workloads they support. In addition, each tectonic shift has seen an acceleration in the time it takes to reach widespread adoption. It took about three decades for the dot-com build-out to blossom into the connected world we have today. It took around two decades for the expansion of wireless infrastructure to lead to billions of smart devices today.
We believe the current AI data center investment cycle is merely a prelude to the massive physical AI opportunity around the corner. Where the past was about connecting millions of users and devices, the future is about enabling tens of billions of devices to sense, think, act, and communicate in real time. Physical AI is not just an incremental step, it's a transformative one that will drive new applications and fundamentally reshape how technology interacts with the physical world. Here is how we see this transformation playing out. AI's impact is already moving beyond the digital realm and into the physical world. We see this today across industrial intelligence, autonomous systems, robotics, drones, and smart devices. The journey from basic autonomy to multipurpose intelligence is underway, and the future will have brand-new form factors and intelligent real-world applications that users can only dream of today.
So how do we get there? Physical AI systems must aggregate data from diverse sensors, process information at different fidelities, and make decisions that are executed through motors and complex real-time control planes. Smarter and ever more bespoke microcontrollers operating in complex, multi-threaded real-time environments are key. They must enable efficient control loops and algorithms leveraging the power of an open versus proprietary compute environment. Communications between an array of sensor domains in the outside world elevate platforms from smart to truly intelligent autonomy, all at the lowest possible power in an untethered and connected world. Where all of these performance requirements for enabling the future of physical AI are at the core of GF's diverse technology portfolio today. What does this mean for the technology industry? We expect physical AI to drive a much broader set of workloads than ever before.
Devices must sense, think, act, and communicate in dynamic, unpredictable environments. This means gathering data from the environment, processing it to make decisions, executing commands through actuators, and sharing data internally and externally. The complexity and diversity of these workloads demand new approaches to semiconductor design and system integration. Optimization for most of these foundational elements of physical AI will not depend on the most advanced lithography. To enable this transition, a broad set of semiconductors is required. Customer needs are evolving rapidly, and semiconductor requirements are becoming more complex, with foundry at the core of this transition. Let's dive into each one of these key areas. First, sense. AI is driving the need for sensing capabilities across virtually every application. This means a surge in semiconductor demand for sensors: more types, more volume, more precision, creating substantial opportunities for precision analog and highly integrated solutions. Second, think.
To support localized AI processing, devices require quick, efficient decision-making in ultra-low-power optimized architectures. This transformation is spreading intelligence across devices, moving from centralized processing to a massively distributed intelligence model. Next, act. Executing decisions in real time requires speed, precision, and reliability. This means dense input-output, IP algorithms, and architectural innovation. The rapid expansion of automation is fueling substantial growth in integrated control solutions, from motor controllers to actuators, and finally, communicate. Connectivity remains a critical enabler of physical AI, and semiconductors are central to achieving secure, multi-standard communication. The proliferation of connected devices is driving substantial demand for low-power RF and connectivity solutions, ensuring billions of links operate securely and efficiently. To sum it all up, physical AI impacts the semiconductor market by driving increased demand for analog precision, multimodal integration, ultra-low power, optimized compute, low latency, secure connectivity, and advanced sensing circuitry.
Integrated memory, configurable workloads, and architectural innovation are essential. The market and our customers are asking for more content and complexity, more memory and optimized processing, and more advanced packaging and integration. Ultimately, with the breadth and depth of our product portfolio and the strength of our differentiated technology, we believe GlobalFoundries is uniquely well-positioned to meet these critical AI needs. In the next 15 minutes, my colleagues Sameer Wasson and Ed Kaste will explain just how well-positioned GF is to enable this significant shift. With that, over to you, Ed.
Thank you, Mike. Now let's talk about how GlobalFoundries is an essential enabler of the physical AI transition. GlobalFoundries' portfolio is strongly matched with the needs of the ongoing AI infrastructure build-out, where we have highly differentiated solutions, including silicon photonics for AI data center connectivity, advanced packaging for 3D heterogeneous integration of different technologies for advanced compute and connectivity, and BCD and Power GaN for data center power delivery. With data transmission and power being two of the most important factors impacting the scaling of AI infrastructure, these product lines are already seeing strong design line momentum. Our other product lines, including RF and feature-rich CMOS, are also particularly well-suited for connecting and enabling applications from cloud to edge. We're incredibly excited the GF's portfolio fit is even stronger for the era of physical AI.
The requirements for semiconductors that are purpose-built to sense, think, act, and communicate, outlined earlier by Mike, are directly in the crosshairs of GF's expertise and where the entirety of our investments are focused. This is especially true in our ultra-low-power CMOS and MIPS product lines, which we will be highlighting today. Our FDX platform is ideally suited for applications that are optimized for power, and particularly leakage power, above all other things. GF's differentiated FinFET platform provides an increased level of performance while still aggressively managing power. Both platforms are fully optimized for integrated solutions requiring high-performance RF connectivity, optimized memories, dedicated interface devices, and more. These platforms are already the solution of choice for home and industrial IoT applications, and you'll find them inside many of the leading industrial automation systems, home security cameras, smart speakers, and connected devices.
Now, in the pursuit of continued enhancement of our technology platforms, we are further layering on the full suite of MIPS RISC-V processor IP, subsystems, and software to target the physical AI opportunity. I'll pass it to Sameer, who will dig deeper into these capabilities, as well as where we see a ton of synergy.
Thanks, Ed. MIPS is powering the next wave of physical AI by delivering the real-time, event-driven computing engines that intelligent systems need. Our processors combine a 40-year legacy in risk innovation with pioneering multi-threading capabilities, making them uniquely suited for compute that must respond instantly to the physical world. This isn't just theory. MIPS cores have been battle-tested in the field. MIPS has had a heritage of proven success, powering the Sony PlayStation 2 and Nintendo 64 game consoles. MIPS today is a fully modern platform, driving the brains inside the Mobileye's EyeQ6 driver assistant chips. In fact, MIPS CPU IP has shipped in hundreds of millions of ADAS SoCs for cars worldwide. Even today, some of the leading cloud hyperscalers use our designs in their infrastructure. This track record means MIPS comes to GF with proven credibility and scale. Why do industry leaders choose MIPS?
Because our technology delivers deterministic real-time performance, the ability to sense an event and act on it immediately, which is exactly what physical AI applications demand. Whether it's an autonomous vehicle making a split-second decision or an industrial robot reacting on a factory floor, MIPS processors provide the fast and reliable intelligence at the edge to make it happen. We achieve this by marrying our patented multi-threaded microarchitectures with an open RISC-V instruction set. That gives customers high performance without the proprietary lock-in and a rich ecosystem of tools. In other words, we've taken a software-first core design approach. We engage with customers early, often virtually, to ensure MIPS IP is a perfect fit for their needs. By the time the chip is fabricated, our customer's software is already running smoothly on our cores, a modern engagement model that speeds up time to market and builds deeper partnerships.
MIPS brings tremendous strategic value to GF. We contribute a proven world-class processor IP portfolio that enables real-time AI in the physical world, backed by decades of innovation and broad adoption. Together with GF's geographic reach and manufacturing excellence, MIPS is positioning GF to lead in high-growth areas like autonomous driving, smart devices, and industrial automation. We're providing the critical computing engines that will drive these physical AI workloads, and that's why MIPS is such a powerful and timely addition to GF's customer offering. Let's go deeper into how we conceive platforms for physical AI, because this is where the real differentiation begins. At MIPS, we don't believe one-size-fits-all architecture. Physical AI is fundamentally different from cloud AI. It's not about centralizing compute in a data center. It's about distributing intelligence across the machine.
Whether it's a robot navigating a warehouse, a drone flying autonomously, or a vehicle making real-time decisions on the road, each function, sensing, thinking, acting, communicating, requires its own specialized compute part. This is why we build platforms from the grounds up. Our processor IP is workload-driven, modular, and optimized for real-time responsiveness. We start with software, because that's where intelligence lives. Our Atlas Explorer platform lets customers tune multimodal AI models, co-design hardware and software, and validate performance before a single chip is taped out. That's the real game changer. Now, when you combine this with GF's ultra-low-power process technologies, you unlock something powerful. GF's FDX and FinFET node give us the lowest power and highest integration in the industry. Their embedded memory, RF integration, and advanced packaging allow us to build dense, efficient SoCs that meet the latency, power, and cost constraints for edge devices.
Think about a robot operating on battery in a dynamic environment. It needs to sense with precision, interpret multiple sensor modalities, and act with accuracy and communicate instantly. That's not just AI. It's physical AI, and it demands distributed intelligence, low-power AI cores, high-quality mixed-signal analog, and real-time communication backlinks. These are exactly the areas where MIPS and GF excel. Together, we're not just building chips. We're building platforms for our customers that are purpose-built for the physical world, and that's how we're enabling the next wave of intelligent machines.
Thanks, Sameer. To bring these concepts home, let's start with a real application where physical AI is already having an impact: the software-defined vehicle, which is increasingly managing its own real-time interactions with the physical world. As a consequence of this trend, we're seeing the semiconductor footprint in automobiles surge from a supporting role to the central pillar of innovation. Modern vehicles now integrate hundreds of chips, powering everything from ADAS to infotainment and battery management. As autonomy, connectivity, and electrification accelerate, we see a corresponding increase in the semiconductor content per vehicle, which, in the vast majority of cases, do not require EUV-class semiconductors. In the last five years alone, the average semiconductor content per vehicle has risen from $700-$1,000 and is expected to continue growing to approximately $1,400 through the end of the decade.
High-performance radars, for example, are at the heart of ADAS systems as precision sensors are taking the place of a driver's vision, and data is analyzed by AI. These are incredibly demanding systems requiring a sensing range of 400 meters, angular resolution below 1/10 of a degree, and operating frequencies up to 120 gigahertz. They have to perform this functionality in under three watts of power to maximize energy available to move the vehicle and avoid overheating. The GlobalFoundries FDX platform is already the industry-leading solution for this application, thanks to superior RF performance, leakage management, and highly reliable operation up to 150 degrees Celsius. Now we have the additional benefit of MIPS IP on this platform, combining deterministic real-time processing with industry-leading efficiency and safety compliance. The multi-threaded architecture enables parallel execution of vision, sensor fusion, and decision-making workloads without sacrificing latency, critical for ADAS responsiveness.
This is a real winning combination. While the transformation of things that we know and love is accelerating, these trends will also give rise to new applications that previously would not have been possible without them. The software-defined vehicle paradigm is rapidly extending into robotics, with humanoids as a very exciting example. Just as software-defined vehicles decouple hardware from software to enable continuous feature upgrades, humanoid robots require an even higher degree of flexibility to support evolving AI models, sensor fusion algorithms, and autonomy stacks. The same factors driving the semiconductor content growth in vehicles are therefore magnified in the humanoid use case. Multimodal sensors emulate human senses, enabling rich environmental perception. A distributed intelligence architecture acts as the nervous system, processing data for ultra-low latency and real-time decision-making, and communicating status throughout the body.
Actuators serve as the muscles, delivering precise fluid motion for real-time interaction with the physical world. All of this must be done continuously and seamlessly while always conserving energy for the next task. This all means that the semiconductor content in a high-end industrial humanoid can reach three to four times that of a modern automobile, creating a massive opportunity for GF to deliver energy-efficient, safety-certified platforms that can adapt over time. Our ultra-low-power platforms are perfectly tailored for running these distributed intelligence systems, delivering the multimodal sensing and edge compute capability required at the right power profile. We continue to drive disruptive memory solutions, such as MRAM and RRAM, with the smallest footprints and fastest access times in the market. These technologies allow our customers to build differentiated products without developing memory IP from scratch, helping them future-proof their designs as traditional memory scaling hits physical and economic limits.
We'll continue to innovate as leaders in RF and high-speed connectivity to increase speed and bandwidth of interconnects within and outside of the application. MIPS brings an even stronger solution into this use case, with standard-setting control loops for real-time motor control to enable the 40-plus degrees of freedom in these robots. This application will bring to bear the very best of GF's capabilities.
At the end of the day, what's validating our strategy and giving us confidence that we are on the right path is the feedback we have received from our customers in bringing GF and MIPS together. We are experiencing a marked increase in customer engagements, and the desire to accelerate partnerships is evident. The opportunity with physical AI is becoming evident for the industry, and the only way to accelerate towards tapping into that opportunity is by forming stronger alliances and relationships across the value chain. With GF and MIPS coming together, we are now able to cater to a broader set of new and existing customers who use both traditional and more custom silicon models.
In addition, with our product offering scaling from process, IP, silicon, and software, we are quickly becoming the design enablement partners for hardware and software engineers at our customers as they develop solutions for the physical AI space.
Okay. You've heard from Ed and Sameer about the strength of GF's technology and capabilities, as well as specific examples of where we are winning in physical AI. Let me now talk about the significant market opportunity this presents and how GF is building upon our strong customer and design win momentum. GlobalFoundries is now at an exciting inflection point where preparation meets opportunity. Over the last decade, we 've worked to expand our product portfolio, increase the differentiation of our technology, and reach efficient scale in our fabs across the U.S., Germany, and Singapore. That preparation has positioned us at the forefront of the next transformational shift, the proliferation of AI into the real world. We currently estimate our physical AI SAM to be $18 billion by 2030, a lower-bound forecast with significantly higher upside potential as new form factors and next-gen applications ramp.
Shown through these four broad market lenses: transportation, industrial, consumer, and medical, these examples of end applications are just the tip of the iceberg. Beyond these customer opportunities, with MIPS and GF's ultra-low-power platform, we have an even stronger business model than ever before. GF continues to evolve with our customers' requirements and is now a diversified, holistic technology solutions provider with growing opportunities, including custom silicon, IP licensing, royalties, and software. Combine the strength of our technology and the unique benefits of our geographic footprint, and there is every reason to expect GF to capture significantly more than our historical market share in this new paradigm. Given the significant value we provide our customers as we ramp our high-margin revenue mix, we fully expect our enhanced business model to drive greater growth and profitability accretive to our long-term objectives.
We've shown you a glimpse of the future. Let me show you how we get there. As we partner closely with customers to understand their needs, we see strong design win momentum across all of our end markets. In the first three quarters of 2025, compared to just two years ago, we've nearly tripled the number of design wins with customers. Of these wins in 2025 year to date, approximately 95% were secured on a sole-sourced basis for GlobalFoundries, a strong testament to our differentiated technology. In the last couple of quarters alone, we announced customer partnerships and wins across smart glasses, satellite communications, autonomous vehicle smart sensors, AI-enabled smart medical, and more. These design wins are what we have accomplished so far, but we are even more excited about the traction we have with leading industry players moving forward.
GF is engaged with a wide breadth of the key players driving innovation and growth in our industry. The level of engagement we are seeing with the key industry leaders is unprecedented in our history. The combination of GF and MIPS allows us to have more meaningful conversations with customers. Through years of partnership and collaboration, we are currently engaged with many of the top industrial IDMs, leading AI fabless companies, automotive OEMs, and U.S. hyperscalers, just to name a few. Many of these relationships are just starting to bear fruit. All of this is laying the foundation for accelerated growth and increased market share in the coming years. Today, I can confidently say that GF has never been in a better position to capture the opportunities ahead of us. We are in the very early innings of a generational transformation, and we are excited to craft this future.
With that, I'd like to hand it over to Eric.
Thanks, Mike. Before we open the call to questions, a friendly reminder to please keep the scope of your questions to the topics from today's presentation. We will be happy to address near-term questions about the business after we report Q4 earnings in February. Operator?
Certainly. And as a reminder, ladies and gentlemen, if you do have a question at this time, please press star one one on your telephone. Our first question comes from the line of CJ Muse from Cantor Fitzgerald. Your question, please.
Yeah, good morning. Thank you for taking the question. I guess first question, as you think about physical AI and your positioning and the design win momentum that you highlighted, how should we think about the lifetime of revenue from these design wins? And then as you think kind of top down, since kind of going public, your revenue CAGR has been about 7%. And we'd be curious how you think about this incremental demand would drive incremental uplift versus that historical 7%.
Yeah, CJ, Mike Hogan here. Appreciate the question. I guess the best way to think about it is break it up into three time domains: the near term, the medium term, and the long term. In the near term, we talked about our design win momentum in 2025, and design win momentum in 2025 starts to show revenue impact in 2027 timeframe, basically. But really strong momentum, 400 design wins in the first three quarters of the year. It's 95% sole source. It bodes very well for our near-term growth prospects. Medium term, as we laid out, we think physical AI is sort of a force multiplier for GF because the types of technology that physical AI requires plays more to our strengths.
We see that as improving gross margin because we have a higher value set of solutions to bring to the market with the addition of custom and MIPS IP. Longer term, we think it really is a tens of billions of devices with very high semiconductor content, and again, very much in our wheelhouse from a technology perspective. We talked about an $18 billion SAM in 2030 in the presentation, and we think that's conservative and a bit of a mix of existing markets that we serve and existing devices becoming more autonomous. Then the unknown, the kinds of devices we probably can't even imagine coming into play. The way we think about it is rough and top , 10% has been our market share in the SAM that we've served.
But when a market opportunity is uniquely well-positioned for our technology, we've seen that 30% or higher. Think of things like automotive radar. So we feel good we've been investing in the right technologies, and this physical AI world, this world of sense, think, act, communicate, is actually moving towards our strengths. And the technologies required to take advantage of that don't really favor the single-digit nanometer nodes that are probably more prevalent in what people are talking about in the data center today. So I think if you add all that together, we should experience much bigger market share growth. We have a better product portfolio to bring to the market with the addition of MIPS and the ability to serve some of the custom silicon needs of our customers. So net-net, much more positive prospects. Ed, do you want to throw anything on there?
Yeah. I mean, I think another way to think about it is to draw some parallels to what we've seen in the automotive market. I think that's a good analogy for what we see happening and how we see it playing out for physical AI, so as the automobile, for example, moved from largely mechanical systems to semiconductors to support the needs for autonomy, connectivity, electrification, the market, the automotive market moved right into GlobalFoundries' strength, and so as a result of that, we've organically outgrown both our peers and the broader automotive unit growth for the last several years. As you'll know, our automotive revenue grew from under $100 million to now approaching $1.5 billion just in the last five years, which demonstrates Mike's point of how market transitions can accelerate our growth when they're very well aligned with our portfolio.
But ultimately, we expect the physical AI to be even larger for GF than what we outlined there for the automotive market because the impact of physical AI will actually impact multiple markets. So that's why we do see this as a force multiplier and a real catalyst for GlobalFoundries' growth.
Yeah, so CJ, to put a bow on it, we expect physical AI will get us at or above a 10% CAGR rate once we start to see the ramp of these solutions coming to market.
Very helpful. Thanks, Mike. I guess maybe a follow-up question for Sameer. Curious, now that part of GF would love to hear how customer engagement is going, how you and MIPS specifically are helping physical AI wins, and how should we be thinking about perhaps the evolution of GF's customer base now that you are on board? Thanks so much.
Yeah, absolutely. Thank you, CJ. Yeah. So I mean, honestly, it's a pretty exciting time right now with physical AI and what's happening in the industry. I think it's a good time for MIPS to partner up and become part of GF as we go forward. Honestly, the feedback from customers has been pretty encouraging, right? I mean, we are not only engaging with the traditional customers, which GF had. MIPS brings in a new set of customers, mainly on the OEM and tier one side as well, which is starting to expand GF's footprint and just dialogue. The dialogue is richer. We are starting to service the customer at multiple vectors. Obviously, very strong on the foundry side, but now starting to add IP, starting to add some custom silicon discussions, and eventually some software and tools as well.
So just the entire service which we are able to provide to the customer is richer, broader, and healthier. And if you think about it, right, as we go execute to our plans, 2026, I expect us to add about $60 million-$100 million of revenue over whatever GF foundry business does. And then obviously, as physical AI starts ramping, that will significantly go accelerate in the outer years. So no, overall, pretty exciting times. Integration is coming along well, and we are looking forward to continue kind of accelerating the path we are on.
Thanks so much.
Thank you. And our next question comes from the line of Mark Lipacis from Evercore ISI. Your question, please.
Hi. Thanks for the presentation and for taking the question. The question I have is, you talk about an $18 billion SAM, and then you showed some examples of what's going on in the car and then a humanoid robot. And I'm wondering, are there other concrete examples that make it obvious to investors what's going on with physical AI? And I'm wondering about the hyperscalers. I think you had mentioned some engagements with hyperscalers, and I think when people think about AI, they think about hyperscalers. And I'm wondering if you could talk about what you're doing in that vertical market. And then I had a follow-up. Thanks.
Yeah. Thanks, Mark. Mike here. Let me kind of frame it this way. I think there's four things that maybe people don't fully appreciate about where GF is headed. And it sort of speaks to some of the questions that you have, and we can talk a little bit more about the hyperscaler question. But we're evolving from just a pure foundry to something that's much bigger than that foundry plus, if you want to think of it that way. Much more holistic technology supplier providing not only foundry solutions, but IP and custom silicon. So I think for us, evolving from pure foundry to something much bigger than that is a strength. And because the physical AI opportunity plays more to our strength, the single-digit nanometer aspects don't work against us in that thought process.
For MIPS, I think there's also a bit of a misunderstanding that MIPS isn't just an IP supplier. They're actually bringing more than an IP and design services. They're really more of a platform player, which is why folks like the hyperscalers are much more interested to have an engagement and a discussion with us about how to solve their overall system solution. So the direction of travel is really from proprietary architectures to open, which favors what we have with MIPS, and from off-the-shelf solutions to things that are more customized and bespoke. So again, this is what hyperscalers are trying to solve: very unique problems, very unique architectures. This is why they need to engage with somebody who brings more than just one aspect to the table.
And as Sameer alluded to and mentioned, we've seen a massive upleveling in the customer dialogue, and that's across all different archetypes of customers. So you asked about hyperscalers, but IDMs, fabless, hyperscalers, tier ones, all are having a very different level of discussion with us. So combined with the rapid innovation that we see going on in physical AI, I think we're uniquely suited to serve those markets. But Ed, do you want to throw anything else on there?
Yeah. I mean, Mike, you've described well the more complete solution that GlobalFoundries can now bring to the market, and of course, when it comes to executing that and delivering, it's across our geographically diverse manufacturing footprint, which is also a strategic advantage, especially as markets value secure supply chains and local-for-local sourcing trends accelerate, so this is another example, like we talked about earlier, where the market is moving towards GlobalFoundries' natural strengths. In other words, this is the way that we're built, and we don't have to do anything new to address this need that is getting stronger from our customers, which also sets us apart from our competition, so ultimately, we're providing more and more value to our customers, whether it's through a unique geographic footprint or through our increasingly differentiated technology portfolio.
Maybe Sameer, you want to. Yeah.
Sorry, go ahead.
I was going to ask Sameer if he wanted to get more into the hyperscaler question that you had.
Yeah, absolutely. So I mean, just to add to what Mike and Ed said, GF is building more businesses beyond the foundry business. Very solid foundry business. We're starting to build an IP business with MIPS coming in, starting to build a custom silicon business, and we're starting to build a software business. This now allows us to engage with hyperscalers on multiple vectors. We have strong engagement on the GaN and power side, which is traditional foundry, but we are starting to engage with hyperscalers on the IP licensing side, on some custom silicon opportunities, and also on the software side. In fact, we are already getting deployed as an IP provider in one of the major hyperscalers as we speak, and we expect that to ramp to significant volume in the 2027 timeframe, which will drive a good amount of royalty fall-through for MIPS and GF, so.
That's very helpful. And then organizationally, how do you prosecute these opportunities? Is this a physical AI business unit, or is this a functional capability that layers into the different vertical markets? And on that same kind of topic, would you expect to break out and report physical AI revenues on an ongoing basis? Thank you.
Yeah. I think for now, you should expect we continue to report the same exact end market segments. And we think each of those end market segments will, at a different rate and pace, reflect the realities of physical AI becoming more and more important. So there's no real change to our structure for our reporting in terms of how we attack those markets. We just believe fundamentally the technologies that we invest in RF and power and now in open instruction set architectures like MIPS RISC-V just position those products better to service all of those markets. So physical AI, I don't think will ever be sort of just a standalone segment, but you're going to see aspects of it. You already see it in automotive. You'll see it in IoT. You'll see it in aerospace and defense and other end markets that we already serve.
So I don't think there's going to be really much change there.
Gotcha. Very helpful. Thank you.
Thank you. And our final question for today comes from the line of Tristan Gerra from Baird. Your question, please.
Hi. Good morning. Silicon Photonics is something that you've talked about on your last earnings call, obviously central to AI along with Advanced Packaging being a critical element of the AI ecosystem. We've seen recently the demand for co-package optics drastically increasing for 2027, 2028, and companies in the CPO supply chain planning on raising capacity. So it's very interesting to me to see packaging being central to your capabilities. Could you elaborate on what differentiates you the most in terms of your Advanced Packaging? How much of the 40% CAGR you've mentioned on your last earnings call through 2030 from Silicon Photonics you expect will come from Advanced Packaging, and the synergies, obviously, between that and your differentiated wafer business that basically creates additional revenue opportunity. Thank you.
Yeah. Hi, Tristan. Thanks for the question. This is Ed Kaste, and I'll take it. So just to speak a little bit more broadly, GlobalFoundries has a full suite of advanced packaging capabilities in-house, from silicon interposer to through silicon vias, wafer bonding, and now die wafer bonding, as well as execution of traditional packaging through our OSAT partners. What's driving that roadmap is a lot of what you described there in the data center. Silicon photonics is a great example where a complex optical assembly is required to take to market our silicon photonics wafer process technology. Also, in the data center, driving the roadmap for things like die wafer bonding and power delivery applications. And while the data center market, in many cases, is driving the roadmap, we see these capabilities being leveraged across all markets. I'll give you a couple of additional examples.
For image sensing, you already have pixel wafers bonded to processor die and a roadmap to keep scaling that. Same in the display market, especially with the emergence of micro displays being used in AR/VR headsets, for example, and memory stacking on logic die, and many more applications like that, and then what's even more exciting for GlobalFoundries is when advanced packaging allows GlobalFoundries to integrate multiple differentiated solutions. I'll give an example of a very high-performance silicon germanium die for RF sitting with ultra-low-power CMOS SoC, which serves to just amplify the differentiation that those components would individually supply when you provide them in an advanced package solution, so across the board, we're very excited about this area.
GlobalFoundries is well down the road on advanced packaging, and we see this as another important driver of our business, not just for physical AI, today's topic, but really for all markets.
Great. That's very useful. Then just a very quick follow-up, and that may be a little bit less relevant to this topic, but also maybe related. 6G is still a couple of years away in terms of initial technology ramp, but a lot of this is going to be AI-centric as well. Do you view this as an opportunity, or is this just too early to talk about it?
Yeah, Tristan, it's Mike. Yeah, we're keeping a close eye on 6G. Right now, we think that that next upgrade in the cellular infrastructure is probably a little too far out to start to talk about how it may or may not impact physical AI. But I think it's a great example of another market that plays to GF's strengths. We see that the bands that are being considered for 6G start to fall in a very nice area for our RF product line. So things like GaN on silicon for RF are definitely going to have a major play in how 6G evolves in the infrastructure side and probably even in the handset side. So I don't know that I connected dots all the way over to physical AI.
The other thing I think that is kind of consistent with the build-out of infrastructure that will enable physical AI is not only 6G, but sort of a low Earth orbit satellite infrastructure that's going up there, and again, the kind of end market that really benefits and values the strength we have in our product line across things like silicon germanium, where we're seeing a lot of uptake as those solutions get deployed.
Very useful. Thank you very much.
Thank you. This does conclude the question and answer session of today's program. I'd like to hand the program back to Eric Chow for any further remarks.
Thanks, Jonathan. So we hope this session has been helpful. We're envisioning this as just the start of a series of deep dives so you can better understand GF's business and opportunities moving forward. Thank you very much for joining.
Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.