Intel's second-quarter financial results showed robust performance in revenue, profit, and gross margin, marking the seventh consecutive quarter the company has exceeded market financial expectations. Management disclosed that the company is accelerating its transition to an "AI-first" strategy to address an unprecedented wave of computing infrastructure buildout.
As noted, in after-hours trading on the 23rd of the month, Intel reported its second-quarter earnings, with revenue nearly 12% higher than analyst expectations. Revenue growth accelerated to 25% year-over-year, marking the strongest quarterly revenue growth in nearly fifteen years.
More critically, Intel's third-quarter revenue guidance maintained double-digit high growth, with the guidance range implying a year-over-year increase of over 15% to nearly 23%, a range that was entirely above analyst expectations.
Behind this earnings beat, the explosion of the AI business was the biggest driver. On the earnings call, Intel's CEO, Lip-Bu Tan, stated directly:
"Our overall AI-driven business grew more than 70% year-over-year, which includes record data center growth and contributed approximately 70% of total revenue."
Tan emphasized that Intel is in a prime position to benefit from the massive wave of AI infrastructure buildout, holding three strategic assets: x86 CPUs, advanced packaging, and wafer foundry services. He stated:
"Today, we saw the strongest revenue growth in 15 years. Our core message is simple: robust demand for our products continues to outpace our growing supply."
Furthermore, management revealed that 18A process node yields and capacity ramps are exceeding expectations, while key metrics for the next-generation 14A process node are also outperforming internal targets. The company announced it is aggressively locking in supplier purchase orders and accelerating cleanroom construction, sending a clear signal of expansion intent to the market.
Demand Far Exceeds Supply, Capital Expenditure Guidance for This Year and Next Significantly Increased
Facing explosive demand for computing power driven by AI, Intel's current capacity has become its biggest bottleneck.
Due to extremely strong customer demand signals, Intel announced it is raising its 2026 capital expenditure outlook to over $20 billion, a significant increase from earlier this year, and has previewed that 2027 capital expenditure will be substantially higher than 2026 levels.
CFO David Zinsner disclosed that between 2021 and 2026, Intel's total capital expenditure on tools and factories in the US will approach $100 billion, exceeding that of any other semiconductor company during the same period.
Tan did not shy away from pointing out the current pain points in the industry's supply chain:
"The entire industry is facing one of the most severe supply constraints in history, encompassing advanced logic chips, silicon wafers, memory, and substrates. For the foreseeable future, these shortages will persist."
Regarding funding sources, the company has approximately $30 billion in cash and short-term investments, a $10 billion revolving credit facility, and roughly $10 billion in non-core assets that can be monetized, providing ample overall liquidity. Zinsner admitted that if expansion goes well, the company does not rule out raising funds through the capital markets.
Addressing the high capital expenditure, Zinsner also provided the rationale behind the investment, emphasizing spending discipline:
"We only commit capital expenditure when we are very confident it will generate good returns. As we transition to a model that extends the lifecycle of process nodes, these returns are quite substantial."
Looking ahead to the third quarter, Intel provided a revenue guidance range of $15.8 billion to $16.8 billion (midpoint $16.3 billion), with non-GAAP gross margin expected to be 42% and earnings per share of $0.38.
Management cautioned that due to the cyclical nature of capacity expansion, the release of supply growth will be more concentrated towards the end of the third quarter and into the fourth quarter.
Core Foundry Node 18A Mass Production Exceeds Expectations, ASIC Targets a $100 Billion Market
In the area of foundry and process roadmaps, which are under intense market scrutiny, Intel expressed strong confidence.
In Q2, Intel's foundry revenue was $580 million, up 6% sequentially. Driven by strong growth at the 18A node, quarterly wafer output exceeded internal targets by approximately 25%, with sequential growth of over 50%. Tan revealed:
"Since joining the company over a year ago, my confidence in our foundry process roadmap has significantly increased. 18A production ramped significantly during the quarter, with yields continuing to track ahead of expectations. We are currently in high-volume production on 18A for several new products, including Panther Lake and Wildcat Lake."
Additionally, the PDK 0.5 for the 14A node is complete, PDK 0.9 is on track for October, and high-volume production is planned for 2028.
Beyond traditional foundry, Intel's custom chip (ASIC) design services are also experiencing explosive growth.
This business's revenue has nearly tripled year-over-year, is approaching a $2 billion annualized revenue run rate, and has a future target of $4 billion. Tan emphasized in response to an analyst's question:
"This is a huge opportunity. I believe the total addressable market is over $100 billion. With our x86 IP, XPU design capabilities, advanced packaging, and leading-edge process technology, we have a unique advantage to provide customers with custom computing chips."
This quarter, Intel announced a partnership with Fortinet to develop next-generation security processors, a multi-year collaboration with SambaNova to advance disaggregated inference performance optimization, and continued delivery of infrastructure processing units (IPUs) to hyperscale cloud customers.
Tan stated the company will continue to expand its product line, extending from networking to computing and eventually into the accelerator space. Furthermore, the company recently hired former SK hynix CEO Shoxi Li to strengthen its strategic layout in memory architecture integration and computational storage convergence.
Data Center (DCAI) Explosion, PC Market Faces Challenges in Second Half
From a business segment perspective, demand was driven by increased recognition of x86 CPUs in AI infrastructure by cloud and enterprise customers.
The Data Center and AI Group (DCAI) achieved Q2 revenue of $6.3 billion, up 24% sequentially and a staggering 59% year-over-year, with operating profit increasing by approximately $1 billion sequentially.
Addressing the market's concern over the CPU vs. GPU share battle in the AI era, CFO Zinsner gave a very optimistic assessment:
"We've talked in the past about the ratio of CPU to GPU increasing; we now believe they have reached near parity at this stage, and over the long term, on a unit basis, it could potentially lean more toward the CPU."
Management predicts strong double-digit growth in industry server CPU shipments this year and next, a trend expected to continue through 2028.
In the Client Computing and Physical AI Group (CCPG, formerly the PC business), Q2 revenue was $8.9 billion, up 15% sequentially. AI-PC revenue grew 26% sequentially, now accounting for two-thirds of total client revenue.
However, management objectively pointed out a concern for the second half: due to memory chip price increases and capacity constraints, the PC consumer market in the second half is expected to show sub-seasonal performance, with a potential low double-digit decline for the full year.
Even so, the strong momentum in edge AI deployment is expected to provide a positive buffer for this segment.
Intel Q2 Earnings Call Transcript (AI-assisted translation):
Operator:
Thank you for standing by. Welcome to Intel Corporation's Second Quarter 2026 Earnings Conference Call. At this time, all participants are in a listen-only mode. A question-and-answer session will follow the prepared remarks. This conference is being recorded.
I would now like to turn the conference over to your host, Mr. John Pitzer, Vice President of Investor Relations. Please go ahead.
John Pitzer (Vice President, Investor Relations):
Thank you, Jonathan. Good afternoon to those joining us today. Our second quarter earnings release and the associated presentation materials are now available on our Investor Relations website at intc.com. For those participating via the webcast, the presentation materials are also available within the webcast player.
Joining me today are our Chief Executive Officer, Lip-Bu Tan, and our Chief Financial Officer, David Zinsner. Lip-Bu will begin with comments on our second quarter results and our progress against our strategic priorities. Dave will then discuss the overall financial results and our third quarter outlook, after which we will open the call for questions.
Please note that today's presentation contains forward-looking statements based on our current expectations and views of the external environment and, as such, are subject to various risks and uncertainties. The presentation also includes non-GAAP financial measures, which we believe provide valuable insight to investors. Our earnings release, our most recent 10-K annual report, and other filings with the SEC provide additional information regarding specific risk factors that could cause actual results to differ materially from our expectations and also provide reconciliations to and a description of our non-GAAP financial measures.
With that, I will turn the call over to Lip-Bu.
Lip-Bu Tan (Chief Executive Officer):
Thank you, John, and good afternoon, everyone. The second quarter was another quarter of solid execution. Revenue, gross margin, and earnings per share all came in above guidance, marking our seventh consecutive quarter of exceeding financial expectations.
Our core message is simple: robust demand for our products continues to outpace our growing supply; we are improving in design, manufacturing, and execution; and the operational discipline we instilled 15 months ago is beginning to yield tangible results. Today, we are witnessing the strongest revenue growth in 15 years.
The cultural transformation continues, and our organization is already operating with higher efficiency—moving faster, making better decisions, and getting closer to our customers. Our recent announcement to deepen our collaboration with Google Cloud will help accelerate this transformation as we embrace an AI-first mindset across our entire business operations. We also continue to strengthen our leadership team with world-class talent.
The robust global demand for and rapid buildout of computing infrastructure are creating significant opportunities for both our products business and our foundry business. Currently, the industry is facing the most severe supply shortages in history across front-end logic, silicon wafers, memory, and substrates, and these shortages will persist for the foreseeable future.
Intel is well-positioned to benefit from this persistent, strong demand thanks to three strategically significant core assets: our x86 CPU product line, advanced packaging technology, and extensive wafer foundry network.
As AI expands from training to inference and increasingly evolves toward autonomous and multi-agent systems, the density requirement for general-purpose server CPUs continues to increase, and our core server CPU business is growing at an unprecedented rate. Demand signals from our customers have further strengthened our confidence. As Dave will detail later, we are significantly increasing investment to support this favorable demand outlook.
Regarding Intel Foundry
Since joining the company over a year ago, my confidence in our foundry process roadmap has significantly increased. I am more convinced than ever of the strategic importance and unique value of Intel Foundry.
In the second quarter, our factories for the Intel 7, Intel 3, and Intel 18A nodes all exceeded internal volume targets, driven by continuous yield improvement, cycle time reduction, and increased wafer starts. 18A capacity grew substantially during the quarter, with yields continuing to track ahead of expectations. We are currently ramping multiple new products into high-volume production on 18A simultaneously, supporting the growing demand for key products like Panther Lake and Wildcat Lake. I continue to raise the bar internally, and the team consistently rises to the challenge. The successful high-volume ramp of 18A on internal products provides important validation for Intel Foundry as it pursues external customers.
At the same time, we have initiated risk production on 18AP, which provides additional performance and power benefits while maintaining compatibility with Intel 18A IP and design, making 18AP a competitive process node for external customers.
Looking beyond 18A, I am very encouraged by the progress on Intel 14A—defect density and transistor performance are tracking ahead of 18A's development cadence at a comparable point. PDK 0.5 is now complete, and PDK 0.9 is on track for delivery in October. We are building and qualifying the IP portfolio for 14A to drive broad adoption of the 14A product family across a wide range of customers. I am pleased to see the growing momentum in customer engagement for Intel 14A, and I am increasingly confident that 14A will be a highly competitive process node across key dimensions like performance, power, density, cost, and schedule.
Based on encouraging external customer progress and growing internal product demand, we remain on track to start risk production of 14A internal products in the second half of 2027 and have made a decision in the second quarter to commit to high-volume production in 2028.
In advanced packaging, customer interest in EMIB-T remains very high. This technology is highly attractive, offering advanced AI chip solutions that current mainstream options cannot match. The backlog for EMIB-T continues to grow, and yields and reliability are on target. Our current focus is to bring this technology to high-volume, high-quality production to support customer product launches in 2027.
Regarding Intel Products
We recently renamed our PC business to the Client Computing and Physical AI Group (CCPG) to reflect the immense growth opportunity in the edge AI market. Our new leadership is driving hard in this direction, and I am very excited about the prospects.
In the core PC client business, Intel 18A is now in mass production across multiple commercial and consumer products, with factory output increasing month-over-month. The successful high-volume production ramp of 18A on internal products also provides important validation for Intel Foundry's external customer engagements. We still have work to do to establish a solid market position in the edge and physical AI ecosystem, but we see it as a significant future growth engine.
The Data Center and AI Group (DCAI) performed well this quarter. Demand from both hyperscale cloud and enterprise markets has clearly accelerated as customers increasingly recognize the critical role CPUs, and specifically x86 CPUs, play in AI infrastructure. Second-quarter server business saw its strongest year-over-year growth ever, with Xeon 6 continuing to be one of the fastest-ramping products in Intel's history, reflecting improved execution and strong customer demand. We further solidified our business outlook in the second quarter by winning new strategic customers and signing long-term agreements.
Our top priority now is to increase capacity and factory output as quickly as possible to meet customer demand while continuously improving our competitive roadmap. We are also deepening our heterogeneous AI strategy through a multi-year collaboration with SambaNova to jointly advance disaggregated inference performance and power optimization.
Additionally, our new design services business continues to make steady progress, with revenue nearly tripling year-over-year. We see a tremendous opportunity to leverage our powerful x86 general-purpose computing product line to create more specialized computing products for the AI era. Our unique advantages in end-to-end design, IP portfolio, combined with leading wafer and packaging capabilities, position us competitively in this rapidly growing field. We are actively expanding our specialized product portfolio, from networking to computing, and gradually extending into the accelerator space.
This quarter's announcement of our collaboration with Fortinet on security processors represents a significant step forward in our ASIC strategy.
Looking ahead, I am energized by the shape of the new Intel. We are operating with greater speed, stronger accountability, and closer proximity to our customers. There is still a lot of work ahead, but our priorities are clear: strengthen product leadership anchored by our x86 computing line, and build Intel Foundry into a world-class wafer and packaging foundry business.
Intel is uniquely positioned to benefit from the overwhelming computing demand driven by the industry's continuous and rapid buildout of computing infrastructure. We are the only company capable of designing and manufacturing complete computing solutions—from general-purpose CPUs and GPUs to specialized ASICs and CPUs optimized for autonomous agent AI. As computing architecture increasingly evolves from single-chip systems to system-in-package solutions, our advanced packaging and wafer foundry capabilities will become even more important strategic assets.
Our strategy is clear, our execution is accelerating, and the opportunity ahead is enormous. The strategy is beginning to show early results, and I am confident that Intel will play a significant role in defining the next era of computing.
I want to thank all of our employees around the world for their focus, discipline, and hard work every day, and I also thank our many customers, partners, and suppliers for their continued trust in Intel.
With that, let me hand it over to Dave to walk through the financials.
David Zinsner (Executive Vice President and Chief Financial Officer):
Thank you, Lip-Bu. We delivered another strong quarter, with robust demand and disciplined execution driving better-than-expected supply performance.
Second-quarter revenue was $16.1 billion, exceeding the midpoint of our guidance by $1.8 billion. Overall AI-driven business grew more than 70% year-over-year, including record data center growth, contributing approximately 70% of total revenue. It's worth noting that while wafer output exceeded expectations this quarter, persistent strong demand still far outpaces our growing supply.
Second-quarter non-GAAP gross margin was 41.8%, approximately 280 basis points above guidance, driven by higher revenue, better yields, and higher average selling prices from a favorable product mix and pricing actions. Non-GAAP earnings per share were $0.42, significantly exceeding guidance of $0.20, driven by higher revenue, stronger gross margin, and solid operating leverage.
Second-quarter operating cash flow was $7 billion, and we ended the quarter with a healthy liquidity position of approximately $30 billion in cash and short-term investments.
Segment Performance
Client Computing and Physical AI Group (CCPG)
CCPG revenue was $8.9 billion, up 15% sequentially, exceeding expectations. The overall client market remained healthy despite widespread component shortages and pricing pressures. AI PC revenue grew 26% sequentially and now represents two-thirds of client revenue. Edge deployment performed solidly and now accounts for approximately 10% of CCPG revenue.
CCPG operating profit was $2.3 billion, representing 26% of revenue, down approximately $173 million sequentially due to an inventory write-down the company took to optimize its factory network to match overall client and server customer demand.
The client group has now scaled 18A to full mass production, with over 400 design wins for Series 3 commercial and consumer products. Against an inflationary backdrop, CCPG's launch of the core Series 3 at A-stepping to provide cost-competitive mainstream computing capabilities was perfectly timed. Integrated Arc graphics solutions continue to gain strong market acceptance, covering over 40 integrated Arc graphics design wins across creator, workstation, commercial, and gaming segments. Following the success of gaming notebooks in the second quarter, CCPG also launched the Intel Arc G-series processors—a new product family designed specifically for the next generation of handheld gaming systems, providing an additional growth vector for the business.
On the commercial side, our market-leading vPro manageability software activations have surged 1500% over the past four quarters, confirming that manageability and enhanced security have become indispensable core requirements in the agentic workplace. We expect enterprise AI adoption to be a long-term tailwind for CCPG, and the AI-driven market opportunity extends far beyond this—the edge and physical AI opportunity is potentially as large as, or even larger than, the client market over the long term. CCPG won 130 Series 3 design wins in edge AI applications, including deployments in applications like robotic brain control.
Data Center and AI Group (DCAI)
DCAI revenue was $6.3 billion, up 24% sequentially and 59% year-over-year, significantly exceeding expectations, driven by strong demand from hyperscale cloud and enterprise markets. The custom chip product line continued its strong momentum, with revenue growing approximately 20% sequentially and nearly tripling year-over-year. DCAI operating profit was $2.5 billion, representing 40% of revenue, an increase of approximately $1 billion sequentially, driven by higher revenue, improved product margins, and lower operating expenses.
This quarter, DCAI launched the Xeon 6+, codenamed Clearwater Forest, Intel's first server-class product built on the 18A process. Additionally, DCAI, along with partners like SambaNova and Foxconn, announced rack-scale and disaggregated inference innovation solutions and introduced new controller and adapter products supporting 10 to 200 Gigabit Ethernet for data center, enterprise, and telecom applications, further enhancing its connectivity portfolio.
Intel Foundry
Foundry revenue was $5.8 billion, up 6% sequentially, driven by strong growth in Intel 18A output—which was approximately 25% above target and grew over 50% sequentially. External foundry revenue was $293 million.
Intel Foundry's second-quarter operating loss improved sequentially by $730.8 million, driven by lower wafer costs from yield improvements, cycle time optimization, and scale-up at the Intel 4, Intel 3, and 18A lines. 18A progress is very good. Intel Foundry has reduced the cost of the Panther Lake primary SKU by approximately 50% year-to-date, with plans for a further ~20% reduction within this year, and expects another significant cost reduction in 2027. This quarter, the foundry business also initiated risk production on 18A-P and achieved the key milestone for Intel 14A PDK 0.9 delivery in October. We increased investment in Intel 14A in the second quarter, preparing for risk production in 2027 and committed high-volume production in 2028.
Third Quarter Guidance
Looking forward, customers continue to signal strong and sustainable spending, fueled by unprecedented demand for AI computing power. Industry-wide supply shortages in wafers, memory, and substrates remain the primary challenge for customers supporting AI infrastructure buildout.
Wafer output across our main process nodes exceeded expectations from 90 days ago, and 18A yield trends are tracking ahead of the targets set in March. Despite strong execution and positive early-quarter trends, supply remains very tight, with the linearity of near-term supply growth weighted more towards the end of the third quarter and the fourth quarter, particularly on the server side.
From an end-market perspective, we expect PC consumption in the second half to be below seasonal patterns, impacted by memory price increases and supply shortages, resulting in a year-over-year decline in the low double digits for full-year 2026, consistent with industry peers and third-party forecasts. Meanwhile, supply improvements, a better product mix, and positive tailwinds from edge deployments provide a degree of positive offset. Since our last earnings call, we have once again raised our expectations for server CPU demand, anticipating strong double-digit industry shipment growth this year and next, with momentum extending into 2028.
Given these factors, we are guiding third-quarter revenue in the range of $15.8 billion to $16.8 billion. At the midpoint of $16.3 billion, we forecast non-GAAP gross margin of 42%, a tax rate of 11%, and earnings per share of $0.38. Full-year non-GAAP operating expenses will remain tightly controlled at approximately $16.5 billion. We expect non-controlling interest (NCI) to be approximately $250 million each in the third and fourth quarters of this year, and on a GAAP basis, approximately $1.1 billion for 2027 and 2028.
Regarding capital expenditure, based on strong customer demand signals, we are raising our 2026 outlook, with capital expenditures expected to exceed $20 billion, a significant increase from earlier this year. We are also actively locking in supplier purchase orders for equipment, accelerating cleanroom construction, and proactively securing substrate and memory supply. We expect capital expenditure in 2027 to be significantly higher than in 2026, with the vast majority invested in our US network. From 2021 to 2026, our total US capital expenditure for equipment and factories will approach $100 billion, substantially exceeding that of any other semiconductor company during the same period. We remain disciplined in tightly aligning spending with customer demand to capitalize on the growth opportunities ahead.
Summary
The second quarter was again strong both financially and operationally. The client market is evolving as expected, and server CPU demand continues to significantly outpace available supply. Emerging markets like physical AI, custom chips, advanced packaging, and external wafer foundry each represent multi-billion dollar annual revenue opportunities in the near future. I am confident in our ability to leverage our broad IP portfolio to solve our customers' most pressing needs and create long-term value for our shareholders.
With that, I'll turn the call back over to John to begin the Q&A session.
John Pitzer (Vice President, Investor Relations):
Thank you, Dave. I would ask that you please limit yourselves to one question and one brief follow-up so we can get to as many people as possible. Jonathan, please take the first question.
Question-and-Answer Session
Operator:
Thank you. Our first question comes from Ben Reitzes of Melius Research. Please go ahead.
Ben Reitzes:
Thank you. Great quarter. I wanted to ask about the capital expenditure of about $3 billion this year and the significant increase next year. Does this mean you have received formal orders for 14A or 18AP? Can you also break down the capital expenditure allocation between advanced packaging and wafer front-end? Thanks.
David Zinsner:
Let me take the second part first. The capital expenditure is spread across various areas, including advanced packaging. As Lip-Bu said, we are very optimistic about the prospects for EMIB-T, so we will continue to invest there. However, front-end fab costs are significantly higher than packaging facilities, so the overall split is still heavily weighted towards front-end, though both are critical to us.
Regarding customers, this increase in investment reflects our confidence across our entire customer base. Specifically in areas where we have signed long-term agreements, we have enough visibility into demand over the next several years to pre-build capacity for various business units. And as I said in my prepared remarks, we maintain strict discipline on capital expenditure. Lip-Bu has also repeatedly emphasized that we only commit capital when we have a high degree of confidence in the returns.
These factories currently under construction are net cash outflows initially, which is why capital expenditure will continue to rise next year. But over the long term, these investments will generate substantial returns—especially in our model of extending process node lifecycles, where returns are particularly significant. The Intel 10 and Intel 7 processes have already proven this point.
John Pitzer:
Ben, do you have a follow-up?
Ben Reitzes:
Yes. Your competitor mentioned today that the total addressable market (TAM) for CPUs could grow to $220 billion by 2030, with a compound annual growth rate of around 45%. How do you view this forecast? Does Intel see the same trend and is it positioned to capture this opportunity?
David Zinsner:
We won't provide specific numbers, but we clearly agree this is a strong market that will grow significantly. Lip-Bu has previously mentioned the increasing ratio of CPU to GPU; we now believe they have reached near parity on a unit basis, and the CPU could potentially exceed the GPU in the future. This is a very high-quality market where we hold a strong position and have the ability to capture a significant share.
As for the exact size of the market, it's difficult for anyone to predict precisely. But based on the customer spending signals we're receiving, the long-term agreements we've signed, and the forward visibility we've gained, the growth is expected to be substantial.
Operator:
Our next question comes from Joe Moore of Morgan Stanley.
Joseph Moore:
Thanks. Regarding server CPU market share, you now own your own fabs, which seems like a significant advantage. Has this helped you gain market share this year? Looking out over the next five years, how do you see the prospects for winning back lost share against competition from both AMD and ARM?
Lip-Bu Tan:
Let me start, and Dave can add. The demand is very strong, and in the context of AI agents and inference, the ratio of CPU to GPU usage is tending towards parity. The bigger challenge right now is expanding supply to meet customer demand.
On the server and data center side, we have a strong product roadmap: Clearwater Forest, Diamond Rapids, and Coral Rapids, which will introduce simultaneous multi-threading (SMT). We continue to improve single-thread and multi-thread performance, with multi-thread improvements coming in Coral Rapids.
Regarding ARM, they are great partners, and I have a good personal relationship with Rene and Masa. We are not only focused on ARM-based CPUs; ARM can also be an important partner and customer for us in ASIC foundry and IP licensing. Overall, we are performing well in the competitive landscape, have a strong product roadmap, and are catching up in certain areas at a very fast pace. We are also investing significant resources in certain CPU architecture directions, aiming to leapfrog, and time will tell.
John Pitzer:
Joe, do you have a follow-up?
Joseph Moore:
Yes. Regarding capital expenditure, is there still a difference between "gross" and "net"? Can you provide more color on the allocation of capital expenditure between internal products and external foundry customers?
David Zinsner:
There is currently a difference between gross and net, mainly driven by the AMIC (Advanced Manufacturing Investment Credit), which is currently in the low single-digit billions and is expected to grow as we progress. Essentially, there is a timing difference—for every dollar we spend in the US, we get $0.35 back in investment tax credits, and the US accounts for the vast majority of our capital expenditure, so the payback is substantial. But there's a lag: tax credits for buildings can be claimed after the factory is built, and for equipment after it's installed and ready for production, after which we file with the IRS. So there is a timing difference between spending and recovery, but the difference between net and gross exists.
Regarding the allocation between internal and external capital expenditure, we prefer to view it from the perspective of total wafer starts, rather than strictly distinguishing between internal and external. We determine the required wafer starts for each node based on demand drivers from various business segments, then place purchase orders with suppliers accordingly, staying flexible as new information comes in. On the packaging side, we have accumulated a large order backlog, so we need to ramp capacity quickly, including expanding internal manufacturing facilities and sourcing substrates from external suppliers. We are accelerating investment in these areas to get ahead.
Operator:
Our next question comes from Stacy Rasgon of Bernstein Research.
Stacy Rasgon:
Thanks. I wanted to ask about the client business. Everyone expected the data center strength, but the outperformance in client was a bit surprising. Was it primarily pricing-driven, or were there other factors? Also, given that the end market appears to be weaker than normal seasonality in the second half, what is your outlook for the client business in the second half?
David Zinsner:
The client business did exceed expectations, but I think the main driver was average selling price (ASP). Part of that was product mix, and part was active pricing actions on comparable products—since our cost base has experienced some inflation, we needed to pass that through to end customers. Compared to last year, the market is indeed declining; 2025 was an exceptionally strong year due to the Windows upgrade cycle, and we are now in a correction phase, further pressured by memory cost increases and supply tightness.
Our shift in product mix towards the higher end contributed significantly to the ASP improvement, helping the results outperform.
Looking to the next quarter, we expect client revenue to be roughly flat, with CCPG potentially seeing slight growth overall, driven by the edge business, while the client side itself is stabilizing. The underlying market is under pressure due to memory dynamics, and we expect that segment to decline this quarter. However, because CPU inventory levels have been relatively low, we could see some CPU inventory replenishment in the third quarter. In the fourth quarter, our own business will begin to feel the impact of softer demand, but the good news is that we need to shift as much capacity as possible towards data center CPUs to close the significant gap between supply and demand.
John Pitzer (to Stacy):
Stacy, do you have a follow-up?
Stacy Rasgon:
Yes. You mentioned the inventory write-down in client. What were these items, and what was the amount? Also, your guidance for next quarter's gross margin is roughly flat. If we exclude the impact of these write-downs, does that mean the gross margin guidance is actually sequentially down?
David Zinsner:
Okay, let me break it down. We had some products for which supporting components were incomplete, and from an economic perspective, it made more sense to convert these products to other lines than to complete the original build. So we took a write-down on these stranded inventories.
You are correct that we guided next quarter's gross margin flat. The absence of the inventory write-down is certainly a positive, but the offsetting factor is this: while Panther Lake and Granite Rapids costs are improving, their mix is increasing. However, since they are still relatively early in their lifecycle, their margins are below the company average. This is a drag on overall gross margin, which offsets the benefit from the lower write-downs, resulting in a flat outcome.
Over the long term, as 18A yields improve further, Panther Lake's margins will improve and exceed the company average, at which point it will become a driver for gross margin expansion. We remain highly focused on continuous gross margin improvement. Frankly, one of my core objectives for the finance team this year has been to firmly establish gross margin in the 40%+ range every quarter. The team has done an excellent job in the first two quarters, and the third quarter guidance points to the same result. From there, we will continue to drive gross margin higher.
Operator:
Our next question comes from Timothy Arcuri of UBS.
Timothy Arcuri:
Thanks. Dave, you mentioned capacity significantly increases towards the end of the quarter, which seems to imply a decent sequential revenue uptick in the fourth quarter. If we assume you are still shipping from backlog, which is over $1 billion, then Q4 should be quite strong. Is that the right way to think about it? Can you discuss the puts and takes?
David Zinsner:
We typically only guide one quarter at a time, as is our practice. But having said that, if we start to see inventory and supply improvements by the end of the third quarter, the fourth quarter could indeed see an uplift. It's worth noting that even with supply improvements, we will not completely close the gap; the shortfall will persist into Q4.
The internal teams have done a remarkable job in ramping capacity, in some ways exceeding Lip-Bu's and my expectations. But our supply depends on a combination of internal wafer output and factors like advanced packaging, substrates, glass cores, and memory. Procuring the back-end supply chain is actually our most challenging area right now. Front-end wafer capacity ramps more linearly, while the release of some back-end capacity is more "lumpy." Therefore, we expect bottlenecks to start easing only towards the end of the third quarter, which explains why revenue is relatively flat this quarter and has upside potential in Q4.
John Pitzer:
Tim, do you have a follow-up?
Timothy Arcuri:
Yes. Dave, the incremental gross margins in March and June were quite good, but based on the guidance, this incremental margin seems to fall back to the low 50% range, which is still within your 40%-60% target range. Is this still the correct reference framework? Also, any thoughts on the outlook for next year?
David Zinsner:
Over the long term, the 40%-60% range is still a reasonable landing zone for incremental margins. Each quarter has its own unique dynamics that affect where we fall within that range, but overall, I think it's a useful reference standard.
Operator:
Our next question comes from Vivek Arya of Bank of America Securities.
Vivek Arya:
Thanks. Lip-Bu, you mentioned increasing confidence in pursuing external foundry customers. When might this confidence translate into tangible customer announcements? Also, how much of the increase in capital expenditure is for external customers versus internal product demand?
Lip-Bu Tan:
Thank you, Vivek. Let me talk about the source of confidence, and then Dave will discuss capital expenditure.
On 18A, 18AP has entered risk production, will be ready by the end of this year, and offers about a 5% performance improvement. 18A yields and volumes are continuously improving, as clearly demonstrated by the ramp of Panther Lake into mass production.
On 14A: PDK 0.5 is complete, and PDK 0.9 for 14A is on track for delivery in October, which is a very important milestone. The yield, defect density, and performance of the 256 SRAM are all tracking ahead of the aggressive plan I set for the team, and they have met all targets. Risk production for 14A is scheduled for the second half of 2027, with mass production committed for 2028.
All these customer engagements and feedback are very positive. The strong internal product demand and the level of engagement from external foundry customers also give me a lot of confidence. When they start to see the 0.9 PDK and the corresponding yield performance, they become very excited and serious about what products to run on 14A and how much capacity they need. This is a very positive signal that customers are genuinely moving forward.
This is also why, as I have said before, only after confirming that yields are on track, IP is ready to serve customers, and customer engagement reaches a certain level, will Dave and I initiate capital expenditure.
David Zinsner:
Let me start with 2026 to break down the capital expenditure. Over the past few years, we have invested heavily in factory space, and we now have very ample capacity, with only a small amount of investment remaining for infrastructure completion. Therefore, the vast majority of current capital expenditure is going to equipment. Compared to 2025, equipment investment in 2026 will increase by 40%, and as you would expect, the focus is on Intel 3, 18A, and 18A-P.
I won't give a specific number for 2027 yet, as details are still being finalized, and we typically provide this at the beginning of the year. However, I have already communicated to investors that we expect this number to be higher than 2026, and I ask for one to two more quarters to finalize the figure.
The direction of capital expenditure will balance both internal and external needs, as we plan overall wafer demand across all business units and build capacity accordingly.
John Pitzer:
Vivek, do you have a follow-up?
Vivek Arya:
Yes. Dave, as you increase investment in the second half and next year, how are you thinking about the balance sheet? Can the success of the CPU business support these investments, or will you need other financing means?
David Zinsner:
That's a good question. Our current balance sheet position is very healthy: approximately $30 billion in cash, plus a $10 billion revolving credit facility, providing about $40 billion in total liquidity. This allowed us to deleverage and maintain our solid investment-grade credit rating.
The continued expansion of revenue, profit, and EBITDA provides substantial cash flow for the business. Additionally, we have approximately $10 billion in non-core assets that we can monetize if necessary, although we are not in a rush to do so, and it serves as a backup option.
We are also seeing customers willing to co-invest with us through prepayments, which helps unlock additional capacity. Of course, if the business expands at a very high speed, we may need to access the capital markets for financing, and we will inform shareholders at that time.
Operator:
Our next question comes from CJ Muse of Cantor Fitzgerald.
CJ Muse:
Good afternoon. Thanks. Dave, I know you won't give guidance beyond one quarter, but can you help us understand the pace of server revenue recovery from the second half of this year into 2027, and how you view the trends in both shipments and average selling prices?
David Zinsner:
Let me provide some context first: server wafers are almost entirely produced internally, with the exception of most ASIC products. We are aggressively increasing wafer starts on our core nodes, especially Intel 3, which is the key node for Granite Rapids. Demand for Granite Rapids is extremely strong; it is the most supply-constrained product in the data center business and is very well received by customers. We are gradually expanding Intel 3 capacity, with a relatively steady ramp planned for the remainder of this year and into next year, but capacity release will have some "lumpy" characteristics.
The challenge is not just on the front end; back-end capacity also needs to be expanded, especially for tight areas like substrates. We made progress in the first half but need to continue efforts.
From an overall market perspective, we primarily discuss growth in terms of shipment volumes, and the server market generally prices on a per-core ASP basis. As servers move towards higher core counts, the ASP per server also increases. This will be an important component of revenue growth and gives us confidence that the compound annual growth rate for this business will be well into double digits for the next several years.
John Pitzer:
CJ, do you have a follow-up?
CJ Muse:
Yes. Coming back to capital expenditure, I know you won't give guidance for next year, but in balancing customer demand with free cash flow targets, is there a framework you can share to help us model? Or is it simply "invest as long as we have customer contracts"?
David Zinsner:
We will be more judicious. Looking at the core business's operating cash flow alone, even with increased capital expenditure, and with the offset from AMIC tax credits, the cash flow profile is actually quite good. However, third-party investments in the back end could be a drag on next year's cash flow, making it challenging to achieve positive free cash flow.
However, all our investments have attractive returns on investment. As long as we have confidence in the growth rates and pricing cost structures, and these nodes typically have long lifecycles that almost always yield a good return on invested capital, we will invest decisively. But we will be very careful not to make large capital bets before customer commitments are secured—this is one of the most important changes Lip-Bu has brought.
You can interpret the current situation from the opposite perspective: based on our confidence for next year and the purchase orders we are placing, this in itself indicates that our confidence in customers is already very high.
Operator:
Our final question comes from Aaron Rakers of Wells Fargo.
Aaron Rakers:
Thanks. First question on the ASIC business. Based on last quarter's disclosure, the annualized revenue is around $1.2 billion with good momentum. Can you talk about the diversification of this business? You announced the partnership with Fortinet. How do you view the growth trajectory and margin profile of this business?
Lip-Bu Tan:
This is a huge opportunity, with a potential total addressable market exceeding $100 billion. We have unique advantages, including advanced CPU-based design capabilities, a strong IP portfolio, and our unique advanced packaging capabilities for high integration and multi-layer interconnect—capabilities that are urgently needed by many new AI technologies. The combination of packaging technology with advanced silicon creates abundant opportunities for custom chip development.
The security ASIC collaboration with Fortinet is a good example, as we advance next-generation high-performance security processors. Additionally, Intel's IPUs are already serving multiple hyperscale customers. The opportunity is enormous. This business has grown approximately 3x year-over-year, and the outlook is very bright.
David Zinsner (adding growth context):
Regarding growth rate, this business currently has an annualized revenue run rate of around $2 billion, and we expect it to reach a $4 billion annualized run rate in the near future. Given the $100 billion total addressable market, combined with our IP strength and overall advantages, we should be able to capture a significant market share. Please stay tuned.
John Pitzer:
Aaron, do you have a follow-up?
Aaron Rakers:
Yes. There are many changes happening in memory, memory hierarchy, and architecture, and there is news about Intel's internal R&D efforts, such as Z-angle memory and cross-batch processing memory. Will Intel play a more important role in these computing scaling architectures? Are there significant opportunities in the memory space in the future?
Lip-Bu Tan:
Good question. First, memory has become a major supply constraint, and we are deepening our collaboration with the three major memory manufacturers to meet customer demand—this is our top priority.
Second, as you know, Intel has a rich history in the memory space. Recently, we brought Shoxi Li onto the team, who was formerly the CEO of SK hynix. Memory has become a bottleneck in AI infrastructure and is a core pain point for our customers. We are also researching how to achieve deeper integration of computing and memory, how to use 3D stacking, and how to improve overall efficiency through better memory utilization.
This is an actively explored area, and we will continue to update you on our progress.
Lip-Bu Tan (Closing Remarks):
Thank you all for joining us today. This quarter, we made good progress on Intel's transformation journey, but there is still a lot of work ahead. I look forward to meeting with you during the quarter and providing a further update in October.
Operator:
Thank you for participating in today's conference call. This concludes the call. You may now disconnect and have a great day.
Comments