Intel Corporation (INTC) Earnings Call Transcript & Summary
October 23, 2025
What were the key takeaways from Intel Corporation's October 23, 2025 earnings call?
In Q3 2025, Intel Corporation reported revenue of $13.7 billion and earnings per share (EPS) of $0.23, both exceeding guidance and reflecting a 6% sequential increase. The company highlighted its fourth consecutive quarter of improved execution, driven by strong demand in core markets, particularly in AI and traditional compute. Intel's cash position strengthened significantly, with $30.9 billion in cash and short-term investments, bolstered by strategic partnerships and government funding. Guidance for Q4 projects revenue between $12.8 billion and $13.8 billion, indicating a flat quarter-over-quarter outlook amid ongoing supply constraints.
What topics did Intel Corporation cover?
- Revenue Growth and Strong Demand: Intel's Q3 revenue of $13.7 billion was 'above the high end of our guidance range' and up 6% sequentially, driven by strength in core markets. Management noted that 'customer purchasing behavior and inventory levels are healthy' as AI adoption accelerates demand.
- Improved Cash Position: The company reported a significant cash position of $30.9 billion, up from previous quarters, aided by 'accelerated funding from the United States government' and strategic investments from NVIDIA and SoftBank. This improvement provides Intel with greater operational flexibility.
- AI and Data Center Demand: Management emphasized the role of AI in driving demand, stating that 'AI is clearly accelerating demand for new compute architectures.' The data center segment saw revenue of $4.1 billion, up 5% sequentially, driven by strong enterprise demand for AI servers.
- Supply Constraints: Despite strong demand, Intel faced capacity constraints, particularly on Intel 10 and 7, which limited its ability to meet demand. Management acknowledged that 'we are working closely with customers to maximize our available output.'
- Foundry Business Progress: Intel Foundry reported revenue of $4.2 billion, down 4% sequentially, but management expressed confidence in the long-term foundry TAM, citing 'accelerating deployment and adoption of AI' as a key driver for future growth.
What were Intel Corporation's October 23, 2025 results?
- Revenue: $13.7B (vs $13.0B est, +6% QoQ)
- EPS: $0.23 (vs $0.00 est, beat by $0.23)
- Gross Margin: 40% (vs 36% est, +4 percentage points)
- Operating Cash Flow: $2.5B (null)
- CapEx: $3B (null)
- Cash Position: $30.9B (up from previous quarters)
Intel's strong Q3 performance, driven by AI and traditional compute demand, positions the company favorably despite supply constraints. The improved cash position and strategic partnerships, particularly with NVIDIA, are positive catalysts. However, the guidance for Q4 suggests challenges ahead, particularly in maintaining margins and meeting demand. Investors should monitor the execution of the foundry strategy and the transition to new product offerings as key indicators of future performance.
Earnings Call Speaker Segments
John Pitzer
executiveThanks, Jonathan, and good afternoon to all of you who are with us today. At this point, you ought to have gotten the Q3 earnings release along with the earnings presentation, and you can find both of them on our Investor Relations website, intc.com. If you're joining through the webcast today, the earnings presentation can also be seen in the webcast window. Our CEO, Pat Gelsinger, and our CFO, David Zinsner, are here with me today. Pat will start with remarks about our third quarter results and give an update on how we're doing with our strategic priorities. After that, Dave will go over our overall financial results, which includes fourth quarter guidance, and then we'll move on to taking your questions. Before we get started, I want to remind you that today's discussion includes forward-looking statements based on the environment as we currently see it, and because of that, they carry various risks and uncertainties. It also includes references to non-GAAP financial measures that we think give our investors useful information. More details on specific risk factors that could lead to actual results differing materially from what we expect can be found in our earnings release, our most recent annual report on Form 10-K, and other filings with the SEC; those documents also contain additional information about our non-GAAP financial measures, including reconciliations to our corresponding GAAP financial measures where appropriate. With that said, I'll hand things over to Pat.
Pat Gelsinger
executiveAppreciate that, John — and good afternoon, everyone. We turned in a strong third quarter, beating our guidance on revenue, gross margin, and earnings per share. That marks four consecutive quarters of improved execution, powered by growth in our core markets and the consistent strides we're making to rebuild this company. We still have plenty of work ahead, but the moves we're making are the right ones to deliver lasting shareholder value. Cash and liquidity saw a major lift in Q3 — something I've prioritized since stepping into the CEO role in March. That improvement reflects accelerated funding from the U.S. government, significant investments from Nvidia and SoftBank Group, and the partial monetization of Altera and Mobilize. By bolstering our balance sheet, we gained more room to maneuver operationally and put ourselves in a strong spot to continue carrying out our strategy confidently. On a personal note, I feel privileged that President Trump and Secretary Lafnick have put their faith and trust in me. That backing underscores Intel's strategic importance as the sole American semiconductor firm possessing leading-edge logic R&D along with manufacturing. We remain entirely dedicated to furthering the Trump administration's goal of bringing semiconductor production back, and we are proud to have the U.S. government join us as vital partners in that work. On top of that, during this quarter we achieved concrete gains in how well we execute. We're still on schedule to get the company to the right size before the year is out, and beyond that, to reshape our talent composition, bring back an engineering-first culture, and fine-tune the executive and management layers throughout the business. As our staff come back into the office following an unsatisfactory stretch of remote and hybrid arrangements, we're noticing a major uptick in everyday enthusiasm and teamwork. Now I'd like to dig further into the trends driving our core business. Throughout my professional life, I've been fortunate enough to play a part in several rounds of disruptive innovation. I honestly can't think of another moment when I've felt this enthusiastic about where computing is headed and what lies ahead for us. The AI revolution is only getting started, and I think Intel has a real opportunity to take on a far bigger part as we remake this company. That begins with x86, our core franchise, which remains essential in the AI era. Demand for fresh compute architectures, hardware, models, and algorithms is unmistakably being pushed higher by AI. Meanwhile, AI is also bringing back growth in conventional computing, since the data underneath and the insights that come out of it still depend heavily on the products we already sell, from cloud all the way to edge. Our business is getting a near-term lift from AI, and as we execute, that gives us a solid base for durable growth over the long run. Also, thanks to a level of compatibility, security, and flexibility that no one else can match, and because it has the biggest installed base of general-purpose compute, x86 is in a strong spot to drive the hybrid compute setup that AI workloads require — especially for inference, edge workloads, and agentic systems. That makes it an excellent foundation for rebuilding where we stand in the market, so we can revitalize and rejuvenate the x86 ISA and get it ready for the new computing era with strong products and partnerships. Our work with Nvidia illustrates this well. Together we are building a new category of products and experiences that will span several generations and speed up AI adoption across hyperscale, enterprise, and consumer markets. By linking our architectures via Nvidia NVLink, we pair Intel CPU and x86 leadership with Nvidia's unmatched strengths in AI and accelerated computing, opening up innovative solutions that will produce a better customer experience and give Intel a beachhead on tomorrow's leading AI platform. We must keep building on that momentum and take advantage of where we stand by sharpening how we execute on engineering and design. That means bringing in and promoting top architecture talent, and also rethinking our core roadmap so it offers best-in-class features. To move faster on this, we recently launched the Central Engineering Group, which unites our cross-company engineering teams so we can get more out of foundational IP work, test chip design, EDA tools, and design platforms. With this new setup, we eliminate redundant efforts, speed up decision-making, and achieve better alignment across every part of product development. On top of that — and equally important — this group will spearhead a brand-new ASIC and design service business, delivering custom-built silicon to many external customers. That effort will extend the reach of our core x86 IP while also applying our design expertise to provide solutions ranging from general-purpose to fixed-function computing. On the client side, we're still set to deliver our first Panther Lake SKU before this year ends, and additional SKUs will follow during the first six months of next year. That positions us well in notebooks across consumer and enterprise segments with cost-optimized offerings, covering the full range of our PC lineup — starting with entry-level products, moving through our mainstream Core family, and extending to our top-tier Core Ultra family. In the high-end desktop arena, rivalry remains intense, but we continue to make steady headway. Arrow Lake shipments increased as the year progressed, and Nova Lake, our next-generation product, will bring fresh architectural and software enhancements that strengthen what we offer, particularly within the PC gaming halo segment. Looking at this lineup, we believe our PC offerings will be the best they've been in quite a while. When it comes to traditional servers, AI workloads are driving two things: customers refreshing what they already have installed, and adding more capacity. That's being fueled by rapid growth in tokenization, greater strain on storing and processing data, plus the pressure to deal with power and space limits. We remain the top choice for AI head nodes, and demand for Granite Rapids is solid, with deployments at all the big hyperscalers. We're listening closely to customer requests, and performance per watt plus TCO are the key priorities. Like I mentioned last quarter, the most important thing is improving our multi-threading, closing the gaps that exist today and working to regain share. Finally, on the topic of our approach to AI accelerators, my perspective hasn't changed: there's a real opportunity for us to contribute significantly when it comes to building compute platforms aimed at the inference workloads that agentic AI and physical AI are bringing to the fore. That market is going to dwarf the one for training workloads. Our goal is to establish Intel as the preferred compute platform for AI inference, and we intend to team up with a variety of established players along with newer firms that are shaping this emerging compute paradigm. This effort will take several years, and any partnership we enter will happen only once we're able to offer genuine differentiation and products that lead the market. For the short term, our focus remains on bringing AI features to customers via Xeon AI PCs, Arc GPUs, and the open software stack we offer. Down the road, we intend to roll out back-to-back releases of GPUs tuned for inference, arriving each year with more memory and bandwidth to serve enterprise requirements. Now, regarding Intel Foundry, we keep building momentum. Intel 18A is moving forward at a consistent pace. Panther Lake remains set to reach the market during this year. Yields on Intel 18A are improving at a rate we can predict, and Fab 52, located in Arizona and focused on high-volume manufacturing, has reached full operational status. Beyond that, our efforts on Intel 18A keep moving ahead, and our PDK milestones are still being met.
Pat Gelsinger
executiveAt least the next three generations of client and server products will be built on our Intel 18A family. That includes our collaboration with the U.S. government inside the secure enclave, along with other customers who have made commitments. This node matters greatly—it will push wafer volumes far into the coming decade and deliver a solid return on what we've invested. When it comes to Intel 14A, the team stays focused on technology definition, transistor architecture, process flow, design enablement, and foundation IPs; we continue to be active and engaged with prospective external customers, and the early feedback has been encouraging, helping shape and guide our choices. Finally, things are moving along nicely in advanced packaging, particularly areas such as EMIB and EMIB-T, where we hold genuine differentiation. Just like with our Intel products, my belief in what Intel Foundry can achieve in the market keeps getting stronger. The fast buildout of critical AI infrastructure is creating extraordinary appetite for wafer capacity and advanced packaging services—a sizable opening that calls for several suppliers. As we execute, Intel Foundry sits in a distinctive spot to seize this extraordinary demand. During the prior quarter I noted that our Foundry spending would remain disciplined, with attention on capability and scalability, which gives us room to ramp rapidly, and we'll bring on capacity only once external demand is committed. Creating a top-tier Foundry takes years and rests on trust. In this role, we have to make sure our process is simple for many kinds of customers to adopt, since each one builds its products in its own distinctive manner. We need to master the art of delighting customers, because they depend on us to produce wafers satisfying every requirement around power, performance, yield, cost, and schedule. Only through this can they count on us as a genuine long-term partner ensuring their success. That demands a shift in mindset, one I'm pushing throughout Intel Foundry as we set this business up for lasting success. Looking forward, my attention stays fixed on the enduring opportunity in every market we serve now and those we'll move into later. Our approach has taken shape around our distinctive strengths and value proposition, backed by the accelerating, extraordinary need for compute in an AI-driven economy. Our leadership keeps getting stronger. Our culture grows more accountable, collaborative, and execution-oriented, and each day my confidence in what lies ahead increases. I'm eager to keep you informed as we move along this path. Now I'll hand things over to David Zinsner, who will cover our current business trends and financials.
David Zinsner
executiveThanks, Lip-Bu. During Q3, we posted revenue that beat our guidance for a fourth quarter in a row, powered by ongoing momentum in our main markets. Even though we're keeping a close watch on macro volatility, inventory positions and how customers are buying both look sound, and supply across the industry has gotten notably tighter. On top of that, our confidence keeps building that fast AI uptake is fueling expansion in conventional computing and adding to the momentum running through our businesses. On the client side, it's been five years since the COVID pull forward, and we're now seeing the benefit of a refresh across a bigger installed base. Windows 11 migration among enterprises is still underway, and AI PC uptake keeps rising. Over in data center, the accelerating construction of AI infrastructure bodes well for server CPU demand coming from head nodes, inference, orchestration layers, and storage. We're cautiously optimistic that the CPU TAM will keep expanding in 2026, even though we still have work ahead to strengthen how we compete. Revenue for the third quarter came to $13.7 billion, above the top of our guidance range and 6% higher quarter over quarter. Capacity limits, particularly involving Intel 10 and Intel seven, kept us from completely satisfying Q3 demand for both data center and client products. Non-GAAP gross margin landed at 40%, which is four percentage points above guidance, thanks to higher revenue, a better mix, and reduced inventory reserves, partly offset by greater Lunar Lake volume and the initial Intel 18A ramp; third quarter earnings per share came in at $0.23, compared with guidance of break-even EPS, helped by higher revenue, a stronger gross margin, and ongoing cost discipline. Operating cash flow for Q3 was $2.5 billion, gross CapEx during the quarter was $3 billion, and adjusted free cash flow was positive at $900 million. Strengthening our balance sheet ranked among our top priorities for 2025. Along those lines, we closed deals that brought in roughly $20 billion of cash, among them three key strategic partnerships. We finished Q3 holding $30.9 billion in cash and short-term investments. During Q3, we collected $5.7 billion from the U.S. government, $2 billion from SoftBank Group, $4.3 billion from the Altera closure, and $900 million from the Mobilize stake sale; Nvidia's $5 billion investment should close before Q4 ends, in our expectation. Lastly, we paid down $4.3 billion of debt during the quarter, and we'll keep making deleveraging a priority by settling maturities as they fall due in 2026. Turning to segment results for Q3, Intel products revenue reached $12.7 billion, a 7% sequential increase and ahead of what we expected in both client and server. The team did a solid job enabling upside during the quarter despite today's tight capacity conditions, which we believe will carry into 2026; we're coordinating closely with customers to get the most out of the output we have, which includes tweaking pricing and mix to steer demand toward products where supply exists on our side and demand exists on theirs. CCG brought in $8.5 billion in revenue, which is 8% higher than the previous quarter and beat what we had projected, thanks to a TAM that was seasonally stronger, refresh activity driven by Windows 11, and a better pricing mix as Lunar Lake and Arrow Lake ramped up. During the quarter, CCG took another step forward in its relationship with Microsoft by working together on Windows ML and deeply integrating Intel vPro manageability, with Microsoft Intune making it possible to manage fleets securely through cloud connections for businesses regardless of size. The team also hit every key milestone needed to launch Core Ultra three, which carries the codename Panther Lake; we anticipate the client consumption TAM will reach close to 290 million units in 2025, which would be two consecutive years of growth, recovering from the post-COVID low point in 2023. That marks the quickest TAM expansion since 2021, and we are sensibly getting ready for yet another year of robust demand in 2026 as Core Ultra three makes its way into a healthy PC ecosystem. TCAI revenue came to $4.1 billion, a 5% increase sequentially, surpassing expectations, fueled by a better product mix and greater enterprise demand. Momentum in host CPUs for AI servers and storage compute carried on through the quarter, though supply constraints kept additional upside from materializing. Our newest Xeon six processors, codenamed Granite Rapids, deliver substantial advantages, such as up to 68% in TCO savings and as much as 80% lower power usage when measured against the typical server installed today; it's becoming ever more evident that CPUs are essential now and will remain so within the AI data center as AI usage grows and particularly as inference workloads surpass training. Certain data center customers are starting to inquire about longer-term strategic supply agreements to back their business objectives, given how quickly AI infrastructure is expanding. This situation, together with the insufficient investment in traditional infrastructure over the past couple of years, ought to allow the revenue TAM for server CPUs to grow comfortably moving forward. Operating profit for Intel products reached $3.7 billion, which is 29% of revenue, and rose $972 million quarter-over-quarter on the back of stronger product margin, reduced operating expenses, and a favorable comparison stemming from period costs in Q2. Before I turn to Intel Foundry, I want to recognize the relentless work put in by the Nvidia and Intel teams. There's plenty of work ahead of us, but the collaboration we announced this quarter represented the culmination of nearly a year of diligent effort with a company that refuses to cut corners and puts engineering excellence first above everything else. The x86 architecture has served as the bedrock of the digital revolution that drives the modern world. AI represents the next stage of that revolution, and we're working toward ensuring x86 stays at its core. Partnerships such as this one with Nvidia are essential to that pursuit. Turning to Intel Foundry, Intel Foundry reported revenue of $4.2 billion, a 4% decline sequentially. In Q3, Intel Foundry shipped Intel 10 and seven volumes beyond expectations, achieved key 18A milestones, and delivered hardened 18A PDKs to the ecosystem. On the Foundry side, we also pushed Intel 14A forward and kept building momentum, growing the pipeline for our advanced packaging business. Intel Foundry posted an operating loss of $2.3 billion in Q3, which is $847 million better than the prior quarter, largely because the comparison benefits from the roughly $800 million impairment charge we took in Q2. As Lip-Bu Tan mentioned, our confidence in the long-term Foundry TAM keeps rising, helped by faster deployment and uptake of AI and by growing demand for wafers and advanced packaging services. Forecasts point to more than a 10X jump in gigawatts of AI capacity by 2030, which opens up meaningful opportunities for Intel Foundry with external customers, both for wafers and for our differentiated advanced packaging offerings such as EMIB-T; we're still doing the work to win our customers' trust, and with greater balance sheet flexibility we can respond to demand quickly and responsibly as it arises. Shifting to All Other, revenue was $1 billion, with Altera contributing $386 million, down 6% quarter-over-quarter because of the intra-quarter closure of Altera. All Other in Q3 consisted mainly of three pieces: Mobilize, Altera, and IMS. Together, the category produced $100 million in operating profit; now on to guidance. For Q4, our revenue forecast is a range of $12.8 billion to $13.8 billion. At the midpoint, after adjusting for the Altera de-consolidation, Q4 revenue is about flat versus the prior quarter.
David Zinsner
executiveFor Intel products overall, we're looking at a slight increase from one quarter to the next, though that will still fall short of what customers are asking for, since we're still dealing with constrained supply. Breaking that down, CCG should see a small decline quarter over quarter, while DCAI is set to rise sharply, because we're directing wafer capacity toward server shipments rather than entry-level client components. At Intel Foundry, we anticipate revenue will grow compared with the prior quarter, driven by more Intel 18A revenue, and its external Foundry revenue should also climb thanks to Altera's de-consolidation. Turning to All Other, which no longer includes Altera, we're expecting a revenue decrease in line with Mobilize guidance, though IMS's sequential growth will partly counteract that. Using the midpoint of $13.3 billion, our gross margin projection is about 36.5%, a sequential decline caused by product mix, the effect of the initial Core Ultra three shipments — which carry the higher costs typical when a new product ramp is in its early stages — and Altera's de-consolidation. Our forecasts call for a 12% tax rate and EPS of $0.08, both non-GAAP; on a GAAP basis, non-controlled income should come in around $350million- $400 million for Q4, and we're projecting an average fully diluted share count of approximately 5 billion shares in Q4. On CapEx, our expectation for 2025 gross capital investment remains about $18 billion, and we plan to spend over $27 billion of CapEx during 2025, compared with $17 billion in 2024. To close, as we leave Q3 we have a much stronger balance sheet, near-term demand that's solid, and rising confidence in our core x86 franchise, plus the longer-term possibilities in Foundry, ASICs, and accelerators. We're also clear-eyed about the effort required to reach our full potential; we keep bringing in outside talent and empowering our workforce so we can execute better in product and process development as well as manufacturing. We'll keep tight control over what we can manage, respond swiftly as conditions change, and stay focused on creating long-term shareholder value. With that, I'll hand things back to John so we can begin the Q&A.
John Pitzer
executiveAppreciate that, Dave; at this point we're moving into the question-and-answer segment of the call. Just a quick reminder that we'd like everyone to limit themselves to one question, plus a short follow-up if it's relevant; so with that, Jonathan, would you please bring us the first question?
Ross Seymore
analystHello everyone, I appreciate you giving me the chance to pose a question, and well done on those impressive results. Pat, my first one for you concerns the Foundry business. During the quarter, you folks revealed a huge number of partnerships. Your balance sheet got considerably stronger, and the way you sounded in your opening remarks comes across as far more assured regarding the headway being made in Foundry. Do any of those partnership announcements or equity investments factor into that greater confidence, or is it certain technical strengths you're observing that are lifting your optimism for that portion of your business?
Pat Gelsinger
executiveYes, Ross, I really appreciate the questions. I believe a pair of the announcements we issued lean, I'd say, distinctly toward the product area. The first is SoftBank, since they are constructing all that infrastructure — AI infrastructure — which certainly would require additional capacity on the Foundry side. I think that would be the response. At the same time, I'd say that, you know, it's evident from what I get regarding 18A and 14A that we've achieved enormously positive headway. You know, with the consistent advancement on 18A, and Panther Lake would hinge on it. It's clear we're seeing yield in a manner that's more forecastable. I toured Fab 52, which, you know, is completely running for 18A. Regarding 14A, it's clear we're working with several customers on a milestone basis; right now we're also genuinely pushing forward on some of the yield and performance reliability where I'm witnessing gains; what's more thrilling is advanced packaging, where we're likewise seeing significant requirements from a number of the major customers on the Foundry side, from the cloud-capable segment, and on the enterprise side too. On the whole, I'd say we're feeling pretty thrilled about establishing this enduring trust with a number of the customers and expanding it. We're also concentrating on bringing in some of the finest talent, advancing some of the process technology enhancements.
John Pitzer
executiveRoss, do you have a follow-up question?
Ross Seymore
analystYes, I do. This one's for Dave regarding gross margin. You went over the upside for Q3 and the decline quarter over quarter in Q4. Could you take us through a few of the positives and negatives as we consider 2026, just in a directional sense? What I'm getting at, I suppose, is that it appears the largest gain needs to occur in the Foundry gross margin area. Is that the primary factor? What's behind it? When those gross margins rise, does that affect Intel products gross margin at all?
David Zinsner
executiveYes, certainly, Ross. Clearly we won't provide guidance for 2026, though I believe I can offer a bit of color. To begin with, just keep in mind that Altera won't be in the figures for 2026. They were included in the figures for a big portion of 2025. That's likely a point of margin headwind for us, since they added to our gross margins. That will be somewhat of a challenge to get past. I'd say we continue to have faith in that 40%-60% fall through for margins. Naturally, that's a range wide enough to drive a truck through, frankly. Much of that stems from mix. We'll have, clearly, Lunar Lake as a major element at least during the first half of the year, and that product dilutes us. Panther Lake, though clearly it will be a strong cost structure for us eventually, since the wafers are fabricated internally, at first, when you have a new product on a new process, those products are quite expensive initially. It dilutes things early in the year, then improves as the year goes on. I do concur that we ought to see gross margins get better on the Foundry side, no question. In part that's the scale dynamic we'll gain from. Also, as we shift toward a more leading edge mix, 18A certainly, but even Intel four or 3 as well, those products carry better pricing and a better cost structure. Those margins ought to be accretive. How much it improves will mostly depend on how the mix unfolds across the year.
John Pitzer
executiveJonathan, can we have the next question, please?
Joe Moore
analystThank you, that's great. A lot of the prepared remarks caught my attention, specifically regarding the ways your Foundry strategy differs. During the last quarter and again this time, you mentioned that you want customer commitments in place before spending the money. Could you share how those discussions are progressing? From a customer's perspective, I can certainly see the trade-off; they are committing to you. Are they expecting that capacity to be constructed in advance? Do these investments have a bit of a chicken and egg quality, and how are you handling those talks?
Pat Gelsinger
executiveYes, Joseph, I really appreciate that question. On the Foundry front, it's clear we are working with several customers. To build customer trust, you truly need to demonstrate yield improvement and reliability, and you also must have every specific IP they need; this is a service business. You must possess all the correct IP. That's the reason I set up the central engineering group, to secure all the proper IP and align it with what customers require. I believe the optimal path is, you know, to demonstrate performance and yield, and then we can attach a chip so they can genuinely work with it. After that, they can begin deploying their most critical revenue wafer to rely on us, allowing us to drive that success on their behalf. I think those things matter greatly; when it comes to possible investment and collaboration, different customers have different needs, and we are working alongside them. What matters more is securing their commitment to Foundry and their support. I think that is how trust gets built, and that is more important.
David Zinsner
executiveSure, I might just jump in here and add something. I believe our customers get it — moving from putting money to work to actually having product come out the other end doesn't happen overnight. What we're aiming for is getting those commitments locked down with enough lead time to put the capital to work and still satisfy the demand. On top of that, I'd point out that the CapEx we've already spent leaves us in a fairly solid spot. A large share of the assets sit on our books, plus what we classify as assets under construction. We've put substantial money into the shell space as well. With our current footprint — and frankly, by drawing on what's under construction and putting existing equipment back to use — we can see a clear path toward generating a decent volume of supply for our external foundry customers. My sense is that we've got room to maneuver. Of course, should things turn out stronger, we might decide to put in more money and move faster, though we're fairly comfortable in our ability to respond to whatever comes.
John Pitzer
executiveJoe, do you have a follow-up question?
Joe Moore
analystI do, yeah. On a different note, regarding the limited availability of server CPUs as well as other CPUs, we're seeing that out there in the market. I suppose, you know, yet your sequential growth came in at 5%, with year-over-year growth in the single digits. So I guess, where exactly is that scarcity originating? Is it simply that there's stronger demand out there that you can't satisfy? Or is it tied to certain transitions you folks have undertaken? I definitely observe that tightness in the marketplace. I'm not disputing it, I'm just wondering, you know, where you think that shortage is originating and what the path to resolving it looks like.
David Zinsner
executiveYeah, I'd say the scarcity is fairly broad across our business, honestly; we're certainly constrained on Intel 10 and seven. Clearly, we aren't aiming to add additional capacity there, and as demand grows, we're limited; in certain respects, we're drawing down inventory. We're also attempting to sort of shape demand to steer customers toward other products. There are shortages too, even past our own particular difficulties on the Foundry side. For instance, there are substrate shortages that have been widely reported, I think. Clearly, on the demand side, there was a great deal of caution as we entered the year, I think, pretty much everywhere; it appears things will be stronger this year, and likely that persists well into next year. I think everyone is attempting to navigate through it.
John Pitzer
executiveJonathan, can we have the next question, please?
Chris Muse
analystYes, good afternoon, and thanks for taking my question. I suppose this is a follow-up regarding the present demand-versus-supply picture heading into 2026. I'm wondering whether that remark mostly concerns servers, or if clients are included as well. And based on how you see that, what should our expectations be for Q1 patterns compared with typical seasonality, which as I understand it would usually be a decline in the high single digits to low double digits?
David Zinsner
executiveYes, it applies to both. That said, as we mentioned, we are giving up a portion of the small core market on the client side so we can satisfy customer needs more broadly across client, and more specifically within server. That's roughly how we're attempting to handle it. Looking ahead to Q1, clearly we'll likely provide far more detail on this in January. I'd simply note that shortages may actually peak for us during the first quarter, since we got through Q3 and Q4 with some inventory helping us and by pushing factory output as hard as we possibly could. That kind of advantage probably won't be available to the same degree in Q1. Given how extremely tight things will be in the first quarter, I'm not certain we'll defy the seasonal pattern. Beyond that, I believe we'll begin seeing some gains and can catch up as the remaining part of the year unfolds.
John Pitzer
executiveCJ, do you have a follow-up question?
Chris Muse
analystThank you, John. With money coming in from the U.S. government, Nvidia, SoftBank, and others, I'd like to know how your approach has changed when it comes to CapEx spending or other investments in your product businesses, since your cash position and liquidity are now better.
David Zinsner
executiveAbsolutely, we're in a strong position, no question. Looking at this cash, our number one focus is bringing our debt down. That was a goal we truly pursued — when Pat came on board, Foundry was a real source of frustration for him. We've put in a lot of effort to repair that and improve things for him. During this quarter, we eliminated $4.3 billion of debt, and the maturities coming due next quarter or next year should be taken care of, and we will pay them off. As for CapEx, it gives us room to move, though we plan to remain highly disciplined about it. We will certainly keep watching demand. Pat has made this very clear to us. He wants to look the customer in the eye and believe in that demand. Should that demand be there, naturally, we'll ramp up CapEx as needed. On the investment side, we continue to believe $16 billion in CapEx for next year is the right number. That said, Pat and I are constantly evaluating how we deploy that $16 billion to get the best growth and returns for investors, and we'll make those changes. Past that, we'll see how it plays out. Our intention is to be pretty disciplined with OpEx as a percentage of revenue and to drive leverage; we do see opportunities to invest that can, in my view, generate strong returns for shareholders, and we're not hesitant to do so either. We also have $350 million available to put toward these priorities.
John Pitzer
executiveJonathan, can we have the next question, please?
Blaine Curtis
analystHey everyone, thanks. I had a couple of questions. First, regarding CapEx, I believe you repeated the $18 billion figure, but my guess is you spent less in Q3 than what I had modeled. Is that figure still accurate? I'm just curious, as you begin ramping 18A in Arizona, how should we think about the schedule for adding capacity there?
David Zinsner
executiveRegarding the $18 billion, yes, I believe that remains the figure. Naturally, CapEx can be uneven; it hinges on when everything gets—you know, once all the conditions tied to paying an invoice are fulfilled, that's when payment occurs. We'd anticipate landing somewhere within that range. Of course, you know, there's some margin of error around it. It could come in slightly under or slightly over. On 18A, you know, we still need to ramp it. I wouldn't anticipate meaningful capacity additions in the near future. As we've mentioned, we aren't at peak supply for 18A. In fact, we won't reach that point until the decade's end. We do believe this node will serve us for a fairly long time. We'll keep making investments in 18A as time goes on. There will be CapEx spending next year, but I wouldn't anticipate capacity shifting significantly relative to what we currently expect.
John Pitzer
executiveBlaine, do you have a quick call?
Blaine Curtis
analystYeah, thank you, John. I'd like to dig deeper into the gross margin path once 18A layers come in. I realize comparing it to, say, the two nodes before probably isn't a fair comparison, though it might be against one that succeeded. When you mention that yields sit in a good place and are getting better, can we frame where 18A yields stand relative to a successful product from your past, and then sort of consider how that plays out across the first half?
David Zinsner
executiveYeah, I'd say broadly speaking, honestly, I'm not certain yields at older nodes were ever a major priority for us. On this one, we're charting new territory. As for yields, what I'd characterize is that they're sufficient to meet supply, yet they fall short of the level required to generate the right margins. By the close of next year, we'll likely reach that point. Without question, the following year, I believe they'll land at what you'd call an industry-acceptable standard for yields. Regarding Intel 14A, I'd share that our beginning has been excellent. When you examine Intel 14A's maturity next to Intel 18A at the equivalent stage of maturity, we come out ahead on performance and yield. On Intel 14A, our start is even stronger. We simply have to, you know, keep pushing that momentum forward.
John Pitzer
executiveJonathan, can we have the next question, please?
Stacy Rasgon
analystHey everyone, I appreciate you letting me ask a couple of things. I'd like to revisit the topic of those supply limitations. You spent a lot of time explaining how AI has been fueling significant demand in servers and in PCs, yet it seems like buyers aren't interested in your AI offerings; actually, what they really want is the older generation. I assume you've got ample inventory for Granite, for Meteor, and also for Lunar Lake. What's your plan to move customers away from those legacy products when they've shown zero inclination to move away from them up to now, even with the shortages they've faced? How should we be thinking about how those customers eventually shift? Since it's obvious — you admitted this yourself — that you aren't producing any additional capacity for the older parts. Actually, you pulled some of that capacity, didn't you?
David Zinsner
executiveYes, that's a fair question, Stacy. I'd say, you know, claiming AI hasn't performed well would be inaccurate. It grew by double digits on a sequential basis from one quarter to the next. We mentioned a figure of roughly, you know, around 100 million units shipping by the close of this year for AI PCs, and we're going to be right at the front edge of that range. My sense is things are going quite well. Obviously, though, the older nodes have performed well too, and that piece was likely the more surprising one. I believe we simply need to help ensure the ecosystem generates sufficient applications for AI in the PC segment. We engage with the ISVs on a regular basis to push that forward; they're making progress; it's, you know, similar to any market — it begins fairly immature and gradually expands with time. Within our own organization, we're beginning to discover uses for AI PCs; as a matter of fact, IR is developing one right now that we intend to use. I think it simply takes time. Having said that, what's evidently occurring is that the Windows refresh is taking place more substantially than we had anticipated. That isn't really an AI PC narrative. Raptor Lake is another product that serves that need; we're seeing additional upside in that segment of the market too.
John Pitzer
executiveStacy, is there anything else you'd like to ask?
Stacy Rasgon
analystYes, thanks. Dave, I'd like to dig into a couple of points I believe I caught from you regarding 18A. First, I think I heard you mention that yields wouldn't be in a strong position, at minimum through the close of next year. I also believe I heard you state that we won't be bringing on much additional 18A capacity in the coming year. Did I get those wrong? Specifically the second one — how could that hold true while you're ramping Panther? Or is that perhaps a result?
David Zinsner
executiveClearly we're at the very beginning. What I'm getting at is that, when you look at the CapEx plan, we aren't going to be adding supply incrementally for Intel 18A in the next year. Sure, naturally, volume will be ramping up as the next year progresses. I wouldn't characterize Intel 18A yields as being in a poor spot. They're at the level we intended at this stage. We set a target for year-end, and they will reach it. For Intel 18A to be fully accretive on the cost structure side, yields need to improve. That's true of every process. It's just how it goes. I believe it will take the entirety of next year to truly arrive at a point where that's the situation.
John Pitzer
executiveJonathan, can we have the next question, please?
Joshua Buchalter
analystHi everyone, I appreciate you letting me ask something. I'd like to follow up on a few remarks Lip-Bu Tan gave during the prepared remarks regarding fixed function computing and the possibility of backing additional ASICs. Could you give us a fuller picture of how broad this effort is? Does it target potential foundry clients, or does it involve products? If products are involved, which kinds of applications do you anticipate serving through custom silicon? Thanks.
Pat Gelsinger
executiveThat's a fair question. To start, I touched earlier on central engineering. We're pushing forward on ASIC design, and that effort is going to get stronger. In fact, this gives us a solid chance to strengthen and broaden how far our core x86 IP reaches, while also advancing certain purpose-built silicon for a number of our system and cloud partners and customers. Certainly, when it comes to the Foundry and packaging side, that too will assist us with respect to what they need. Overall, my view is that AI is going to fuel substantial growth, particularly as we double down on Moore's Law. That's going to give us considerable help with our x86 uplift. There's a chance for us to construct that entire ASIC design effort to meet certain customer requirements.
John Pitzer
executiveJosh, do you have a quick follow-up?
Joshua Buchalter
analystYeah, regarding the quarter that just ended — clearly, that disclosure about possibly walking away from 14A drew plenty of eyeballs. I just wanted to check: since your balance sheet looks quite different now compared with 3 months back, has anything shifted on that front? I mean, I assume the Q isn't filed yet, so I haven't had a chance to see whether any wording was revised there. I was wondering if anything had changed since the prior quarter, considering all the movement in your balance sheet. Thank you.
Pat Gelsinger
executiveYeah, compared with the quarter before, I'd say it's clear that our work with the customer around 14A has grown. We're working extremely closely with that customer to pin down the technology, the process, the yield, and the IP needs that will serve them. They plainly recognize the enormous demand, and that Intel needs to be strong on 14A. So we're thrilled and feel greater confidence. At the same time, we're bringing in some of the essential talent for the process technology who can truly propel success. That's the reason it gave me far more confidence to push that forward.
John Pitzer
executiveJonathan, can we have the next question, please?
Ben Reitzes
analystHey, thanks everyone. Lip-Bu, could you give us an update on where things stand with Nvidia and when products might arrive? Have customers said anything about how well you're able to explain how significant that relationship is, and about timing and significance, or any additional detail you'd care to share? Thanks a lot.
Pat Gelsinger
executiveThank you. This collaboration with Nvidia matters a great deal. As all of you know, it's a wonderful company. I've counted myself a friend of Jensen's for over 30 years. We're quite enthusiastic about this undertaking involving Intel CPU 86 leadership alongside their unparalleled AI and accelerated computing, and then linking that up with their NVLink; that genuinely will bring about a brand-new category of product across the multi-generations. This is a very intensive engineering-to-engineering effort; it will be fueling some of the new offerings in custom data center and PC products, and those are truly tuned for the AI era. Taking everything together, I believe this means several years of engagement and then serving a market we're eager about while also shaping some of the needs of the AI infrastructure.
David Zinsner
executivePerhaps I'll add just one more point. What truly sets this apart in our view is that it isn't taking away from the TAM we already have — instead, it gives us a chance to grow that TAM. For us, these are excellent opportunities.
John Pitzer
executiveBen, is there anything else you'd like to ask?
Ben Reitzes
analystYes, thank you, John. Lip-Bu, you said a bit about your approach to AI at this point, right, to go after the inference sector and that you see space for, well, Intel offerings. It sounds like you'll be teaming up quite a bit in that area. Does this approach lean more toward collaboration? Does it lean more toward — is there particular Intel IP, you know, for inference that gets you excited? Or does it lean more toward a Switzerland-style play in which you might work with plenty of the companies already in the market to go after a bigger share of the TAM? Thanks.
Pat Gelsinger
executiveYes, that's a good question. Given that AI is fueling so much expansion, we certainly want to be part of it. I think this is just the start, and for us that's a chance. One thing we're zeroed in on is bringing our x86 back to strength and truly customizing purpose-built CPU, GPU needs for the fresh AI workload; then actually tackling the power-efficient agentic side and handling all the various agents. That's a new kind of compute platform of choice, and it will also extend to the system and software. You're also going to hear, I think, that we'll be working alongside some of the established players as well as the newer firms that are pushing some of these shifts.
John Pitzer
executiveJonathan, can we have the next question, please?
Timothy Arcuri
analystMuch appreciated. Dave, it's rare to come across companies with heavy fixed costs that are capacity-limited and running gross margins below 40%. I fully understand that the bulk of this stems from wafer expenses tied to Intel 10 and 7, plus you're still getting low yields on 18A. I'm curious — and this may be a tough thing to pose or to respond to — but could you perhaps skip ahead a little and tell us, what would gross margin look like if you were no longer on 10 and seven and had shifted to 18A? Is there any way to normalize that for us?
David Zinsner
executiveYeah, I mean, clearly you might be eavesdropping on some of my discussions with the team here, since that's absolutely a point I've been emphasizing. You know, I'd frame it as two dynamics, and one of them is what you're pointing to — namely, the steep expense of older processes compared with the improved cost structure that newer processes offer. That's clearly significant. I mean, for Foundry we're sitting in negative gross margin territory. That translates into a meaningful gain if you simply cross over into positive territory. The second element affecting our gross margins comes down to product quality itself. On client, we're in reasonably decent shape when it comes to product performance and competitiveness, aside from a few exceptions, but on a cost basis we aren't where we need to be; improvements have to happen there. That's laid out on the roadmap. The team sees it, yet reaching that point takes multiple years; on the data center side it's even more pronounced. Not only is our cost structure not right, but our competitiveness also isn't right, which keeps us from earning the margins we should from customers. There's work ahead of us there. That's exactly what Lip-Bu and the team he's brought in are hyper-focused on — delivering excellent products at the proper cost structure to push toward better gross margins. To me, I think that's the linchpin in all of this. On the Foundry side, the gains are simply going to arrive, I believe. Our mix will shift higher and higher toward Intel three, four, and then 18A and finally 14A. The cost structures across all of those are actually fairly comparable. You know, it'll simply come down to the fact that the value delivered by those leading edge nodes will be substantially greater, and that alone will materially push gross margins upward. Another point I'd make is that we're seeing substantial startup costs because we've crammed a whole bunch of new processes in, in a rather rapid fashion. Once we reach 14A, our cadence will be more normalized, and you won't see as much startup costs piled on top of one another, which is weighing on our gross margins; that's billions of dollars. My sense is that once you move past a handful of years, that drops away, and that should be a benefit as well.
John Pitzer
executiveTim, do you have a quick follow-up?
Timothy Arcuri
analystYes, I do. Lip-Bu, during the previous call you didn't provide any news regarding when Diamond Rapids would arrive. I realize the entire roadmap, as you know, is being reviewed, yet you sounded — the company sounded — reasonably hopeful regarding Coral Rapids. Could you simply share some kind of update on the data center roadmap at this point? Thanks.
Pat Gelsinger
executiveThanks. Those are good questions. Obviously, Diamond Rapids is receiving more robust feedback from hyperscalers. We're also concentrating on the new offering, Coral Rapids. SMT, the micro threading, will be part of it, and that's capable of delivering greater performance; right now we're in the definition phase, and we'll figure out the roadmap, and we intend to carry it out. That's what will happen going forward.
John Pitzer
executiveThanks, Tim. Jonathan, we have time for one last question.
Aaron Rakers
analystYeah, I appreciate you squeezing me in. Just a few very fast ones. I suppose if we return to the Nvidia relationship, I can understand that the announcement was genuinely connected to the NVLink fusion strategy and bringing that together with the x86 ecosystem. I believe there have also been some recent reports about perhaps employing Gaudi for certain dedicated inference workloads inside a stack of Nvidia. How do you — you know, is this relationship, you know, a beginning point, and ought we to anticipate seeing additional potential integration past NVLink down the road?
Pat Gelsinger
executiveYeah, I think, allow me to respond to that. I think NVLink is sort of more the hub for linking the x86 and GPU; regarding the AI strategy, we clearly are defining the, you know, Crescent Island that we discussed. We also have a new product line in the lineup; they're addressing the agentic and the physical AI and more on the inference side. I think stay tuned; we will update that.
John Pitzer
executiveAaron, do you have a quick follow-up?
Aaron Rakers
analystI do, and I'll keep this really brief. Could you give us an update on how we should be thinking about the NCI, that non-controlling interest expense, as we move through this year and consider it? I believe in the past you've offered some remarks about how we ought to think about that in the 2026. Thank you.
David Zinsner
executiveYeah, for 2026, we're looking at somewhere in the $1.2 billion to $1.4 billion range is probably a good estimate. Obviously, we're focused on that, and you know, we'll work to minimize that as much as possible.
Pat Gelsinger
executiveWith that said, I appreciate all of you being here today. We're in the process of rebuilding Intel, and there's plenty of work still in front of us, yet during Q3 we're achieving steady progress. I'm eager to meet many of you over the course of the quarter and to give all of you a further update in January.
Operator
operatorThanks so much, everyone, for joining us on today's call; that wraps things up for this session. You're welcome to leave the line now. Have a great day.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Intel Corporation transcript — plus 132,000+ transcripts from 7,000+ companies, speaker segments, AI summaries and full-text search — through the EarningsCalls.dev API.
Get the API View API docs →This call discussed
For developers and AI pipelines
Programmatic access to Intel Corporation earnings transcripts and 132,000+ others is available through the
EarningsCalls.dev REST API. Plans from $39.99/month — full transcripts, speaker segments,
full-text search, and the recently-added /api/v1/transcripts/recent polling endpoint for ETL pipelines.
More tools for stock research
OptionsBell: an unusual options activity scanner that tracks options flow across all US stocks and emails you when large trades appear in Intel Corporation or any ticker on your watchlist.
StockMarketScan: a stock screener with alerts: 24 pre-built screens refreshed every 5 minutes, watchlists and email notifications. Both offer API and MCP access and fair monthly pricing.