Conferencia de resultados de Intel del T3 de 2026: crecimiento de DCAI y previsiones del T4

Puntos clave

• Intel registró en el T3 de 2026 unos ingresos de 16.500 millones de dólares, un margen bruto no GAAP del 44,1% y un BPA no GAAP de 0,52 dólares, superando sus propias previsiones en las tres métricas.
• Los ingresos de Centro de Datos e IA, o DCAI, alcanzaron los 7.300 millones de dólares, lo que supone un aumento del 17% intertrimestral y de casi el 80% interanual, impulsados por la demanda de infraestructura de IA y la mejora en el suministro de CPU para servidores.
• Los ingresos del Grupo de Productos de Computación para Clientes, o CCPG, cayeron un 9% intertrimestral hasta los 8.100 millones de dólares. Intel atribuyó esta debilidad al aumento de los precios de la memoria, a la limitación de los presupuestos de TI y a una acelerada corrección del inventario de PC.
• Los ingresos de Intel Foundry aumentaron un 16% intertrimestral hasta los 6.700 millones de dólares, mientras que su pérdida operativa mejoró en 236 millones de dólares hasta los 1.900 millones. Los costes de Intel 18A han disminuido alrededor del 40% desde el T1.
• Para el T4 de 2026, la dirección prevé entre 15.000 millones y 16.000 millones de dólares en ingresos, un margen bruto no GAAP del 38% y un BPA no GAAP de 0,25 dólares en el punto medio.
• La dirección prevé que el suministro de CPU para servidores siga siendo limitado a lo largo de 2027 y hasta 2028. Intel también señaló que es probable que 2027 sea un año de inversiones sin flujo de caja libre positivo, ya que el gasto de capital aumentará de forma pronunciada.

Datos financieros clave

Rendimiento operativo y del negocio

La demanda de centros de datos supera la oferta

DCAI fue el principal motor de crecimiento de Intel. Granite Rapids se convirtió en la mayor familia de productos para servidores de la compañía, mientras que Clearwater Forest, el primer producto de servidor de Intel basado en Intel 18A, comenzó a enviarse generando ingresos. Diamond Rapids se encuentra en fase de muestreo y mantiene su calendario para escalar en Intel 18A-P en 2027.

Los ingresos del silicio a medida aumentaron más de un 50% intertrimestralmente y se duplicaron con creces interanualmente. La dirección afirmó que este negocio se ha triplicado en los últimos tres años y podría volver a triplicarse en los próximos años.

Intel estimó que el mercado total direccionable de CPU para servidores podría expandirse desde los 40.000 millones hasta más de 200.000 millones de dólares en cinco años, impulsado en gran medida por las cargas de trabajo de agentes de IA. Se trata de una estimación de la dirección más que de una previsión oficial de la empresa.

El suministro sigue siendo la principal limitación. Intel citó los tiempos de ciclo de Intel 3, la escasez de sustratos orgánicos avanzados y un paro prolongado en la Fab 28 en Israel. La compañía está trabajando para sumar producción de Intel 3 en Oregón y prevé que el aumento de capacidad de sustratos tenga un mayor impacto en la segunda mitad de 2027.

Se intensifica la corrección de los PC

Los ingresos de CCPG disminuyeron debido a que la escasez de memoria y el alza de precios presionaron la demanda de PC y los presupuestos de los clientes. Los clientes OEM aceleraron las reducciones de inventario a finales del T3.

Aun así, los ingresos por PC con IA de Intel aumentaron un 35% interanual. Las unidades de Wildcat Lake aumentaron cerca de un 70% intertrimestralmente y el producto se agotó. Panther Lake continuó su escalado en Intel 18A, mientras que Nova Lake se mantiene por delante de lo programado. La dirección prevé que Nova Lake cuente con el doble de adjudicaciones de diseño que Panther Lake en su lanzamiento.

Intel prevé que el TAM de unidades de PC de la industria caiga cerca de un 12% en 2026, incluida una caída de más del 20% en la segunda mitad. Para 2027, la dirección proyecta un descenso de unidades de aproximadamente el 10%, con una mayor presión en el primer semestre.

Mejoran los costes y el rendimiento de la fundición

Intel declaró que el rendimiento de fabricación volvió a mejorar en Intel 3 e Intel 18A, mejorando el rendimiento de Intel 18A a un ritmo de doble dígito durante el T3. Los costes de Intel 3 cayeron cerca de un 15% desde el T1, mientras que los costes de Intel 18A cayeron aproximadamente un 40% durante el mismo periodo.

Las CPU de cliente en Intel 18A pasaron de representar menos del 5% de los ingresos por CPU de cliente en el T1 a casi un tercio en el T3. Intel 18A-P se mantiene según lo previsto para lanzarse a finales de año, y la dirección señaló que el compromiso con clientes externos va en aumento.

En cuanto a Intel 14A, la compañía publicó su kit de diseño de procesos (PDK) 0.9 para clientes seleccionados en octubre. La dirección prevé el PDK 1.0 en el T1 de 2027 y mantiene como objetivo la producción en riesgo en 2027 antes de la producción a gran escala en 2028.

Previsiones de la dirección

Las perspectivas para el T4 incluyen un impacto estimado en los ingresos de 300 millones de dólares y un impacto de 200 puntos básicos en el margen bruto debido al incidente de la Fab 28. Casi todo el efecto en los ingresos se relaciona con productos para servidores de 10 nanómetros.

La dirección prevé que los ingresos de DCAI crezcan un porcentaje en la franja media del 10% al 20% intertrimestralmente en el T4, con el silicio a medida creciendo más rápido que el segmento en general. Se prevé que los ingresos de CCPG disminuyan más de un 20% intertrimestralmente a medida que los clientes reduzcan inventario.

Intel mantuvo su previsión de gasto bruto de capital para 2026 de 20.000 millones de dólares y prevé que el gasto en 2027 aumente de forma significativa. Los gastos operativos no GAAP para todo el año 2026 se proyectan en aproximadamente 16.300 millones de dólares.

Riesgos y aspectos a vigilar

• Restricciones en el suministro de servidores: La demanda sigue superando a la oferta disponible, y se prevé que las limitaciones en las obleas de Intel 3 y los sustratos de empaquetado avanzado persistan hasta el primer semestre de 2027.
• Interrupción de la Fab 28: La dirección prevé un impacto en los ingresos de aproximadamente 300 millones de dólares tanto en el T4 de 2026 como en el T1 de 2027, según sus evaluaciones preliminares.
• Presión en el mercado de PC y memorias: Intel no prevé que las condiciones de las memorias mejoren durante 2027. La corrección de inventario de PC podría prolongarse del T4 al T1 y potencialmente a parte del T2.
• Presión sobre el margen bruto: Menores ingresos en el segmento de clientes, costes fijos de fabricación, subutilización en Intel 18A y el perfil de menor margen del silicio a medida podrían pesar sobre los márgenes.
• Intensidad de capital: La dirección señaló que es probable que 2027 no genere un flujo de caja libre positivo debido a que Intel acelerará simultáneamente Intel 18A, invertirá en Intel 18A-P y preparará Intel 14A para su producción.
• Rentabilidad de la fundición: Aunque los costes y los rendimientos están mejorando, Intel reconoció que queda un trabajo sustancial por hacer para establecer un negocio de fundición sosteniblemente rentable.

Puntos destacados de la sesión de preguntas y respuestas con analistas

Crecimiento en 2027: El director financiero David Zinsner describió 2027 como un año de crecimiento, liderado por los centros de datos y el silicio a medida. Sin embargo, se prevé que las limitaciones de suministro sean más severas en el primer semestre y que haya mayor capacidad disponible más adelante en el año.

Flujo de caja libre y gasto de capital: La dirección aún no ha finalizado el presupuesto de gasto de capital para 2027, aunque las órdenes de compra de equipos aumentaron casi un 40% intertrimestralmente. Zinsner afirmó que es poco probable que haya un flujo de caja libre positivo en 2027 debido a las elevadas inversiones en múltiples nodos de proceso.

Cuota de mercado de servidores: La dirección indicó que la cuota de mercado de servidores de Intel en 2027 dependerá en gran medida de su capacidad para ampliar la oferta. La empresa está reasignando parte de la capacidad de producción de Panther Lake a obleas para servidores Intel 3 en Oregón, aunque la disponibilidad de sustratos sigue siendo otra limitación.

Trayectoria del margen bruto: Granite Rapids comenzó a incrementar el margen bruto en el T3. Los márgenes de Panther Lake mejoraron sustancialmente, pero se mantuvieron por debajo del promedio corporativo. La dirección prevé una expansión del margen bruto anual si Intel logra hacer crecer los ingresos, señalando al mismo tiempo que el rendimiento trimestral puede variar.

Clientes externos de fundición: Intel prevé posibles anuncios de clientes de empaquetado avanzado en los próximos dos trimestres, sujetos a la aprobación de los mismos. La dirección también informó de interés externo en Intel 18A-P y de una colaboración activa con clientes seleccionados en Intel 14A.

Transcripción completa de la llamada de resultados

Transcripción completa de la conferencia de resultados

Comentarios de la dirección

Operator

Thank you for standing by, and welcome to the Intel Corporation's Third Quarter 2026 Earnings Conference Call. [Operator Instructions] As a reminder, today's program is being recorded.

And now I'd like to introduce your host for today's program, Mr. John Pitzer, Corporate Vice President of Investor Relations. Please go ahead, sir.

John Pitzer

Thank you, Jonathan, and good afternoon to everyone joining us today. By now, you should have received a copy of the Q3 earnings release and presentation. Both are available on our Investor Relations website, intc.com. For those joining us online today, the presentation is also available in our webcast window.

I'm joined today by our CEO, Lip-Bu Tan; and our CFO, Dave Zinsner. Lip-Bu will open on comments on third quarter results and progress on our strategic priorities. Dave will then discuss overall financial results, including fourth quarter guidance, before we transition to answer your questions.

Before we begin, please note that today's presentation does contain forward-looking statements based on the environment as we currently see it. As such, they are subject to various risks and uncertainties. It also contains references to non-GAAP financial measures that we believe provide useful information to our investors. Our earnings release, most recent annual report on Form 10-K and other filings with the SEC provide more information on specific risk factors that could cause the actual results to differ materially from our expectations. They also provide additional information on our non-GAAP financial measures, including reconciliation where appropriate to our corresponding GAAP financial measures.

With that, let me turn it over to Lip-Bu.

Lip-Bu Tan

Good afternoon, and thank you for joining us. The demand environment in Q3 was more balanced than previous quarters. Strong momentum in DCAI drove upside and enabled the strongest revenue growth for the segment in 15 years. This was partially offset by weakness in CCPG driven by the industry-wide impacts from higher memory prices and limited IT budgets.

The industry continues to navigate one of the most significant supply shortages in history, driven by the unprecedented demand for super intelligence. Against this backdrop, I am pleased by the progress we make growing our server supply in Q3. I'm mindful about the ongoing challenges we face in the future quarters to keep pace with the server demand.

The successful completion of our equity offering in August provided us with the necessary financial resources to capture the growth ahead. I'm very pleased to add new sovereign funds and long-term growth investor, in addition to our existing shareholders. We are deploying this capital quickly and prudently. We expect CapEx and purchase commitments for key components to increase significantly over the next several quarters. I want to thank our substrate, memory, foundry, equipment suppliers for their strong partnership and collaborations as we work together to expand supply and support growing demand.

Turning to business units. First, DCAI within our server business, we are making steady progress on our road map. Granite Rapids on Intel 3 has become one of the fastest ramping server products. Clearwater Forest, our first Intel 18A server product, is shipping for revenue, and demand for 2027 is significant, especially in CPU-dense agentic rack-scale infrastructure.

Diamond Rapids, our first Intel 18A-P server product is sampling and on track to ramp in 2027, with strong interest for inference and agentic workloads. We were pleased to recently announce an agreement with Oracle Cloud Infrastructure to deploy our latest generation CPUs to support their AI buildout.

I also want to acknowledge headwinds limiting our progress to meet this extremely high AI-driven demand. First, Intel 3 yield and cycle time continue to improve, but I am demanding significant execution from the team. Second, even as our Intel 3 wafer output ramps quickly and industry-wide shortage of advanced organic substrates for our EMIB packaging is also impacting our ability to scale output.

Finally, early this month, an unfortunate and tragic accident led to an extended shutdown in our Israel factory. While the teams have worked quickly to address the situation, it will impact our Q4 10-nanometer server shipments.

Purpose-built silicon remains our fastest-growing business segment, tripling over the last 3 years, and posed to triple again over the next few years. We see further growth opportunity with both existing and new potential customers, including AI labs, and expanding our IP portfolio to meet the needs of a wider array of customers.

Lastly, we recently unveiled our Crescent Island inference optimized GPU. Early feedback is encouraging. Performance is tracking well. This is a new market for us, and it will take time to scale but customer engagements are progressing towards firm commitments.

Turning to our client business. In recent weeks, our customers began signaling a PC slowdown based on the memory prices and availability. Year-to-date system endpoint prices have increased significantly, placing additional pressure on already limited IT budgets. The pace of this PC correction has accelerated over the last 2 months, as our OEM partners are aggressively managing inventory to match the softer sell-through. We expect the current environment to persist until memory availability and prices begin to normalize.

While the next few quarters will be challenging for the PC market, our road map continues to strengthen and our long-term outlook remains positive. We have successfully ramped Panther Lake on Intel 18A, and demand for the Wildcat Lake is outstripping supply. Nova Lake, our first Intel 18A-P client product, is tracking ahead of schedule, with design wins expecting to be twice those of Panther Lake at launch. We look forward to sharing more at CES in January. As the market normalizes, we expect this portfolio drives meaningful share gain.

Intel Foundry execution continues to improve consistently. During the quarter, Intel 3 and Intel 18A yields both increased again, with Intel 18A improving at a double-digit rate. Wafer output on both nodes exceeded expectation. Intel 18A-P remains on track for launch by year-end across our internal product portfolio, and external customer engagement continues to build. We are engaged with a couple of highly strategic customers, and I'm increasingly confident that our Intel 18A-P will be an attractive node for external foundry customers.

Intel 14A also remains on track. In October, as committed, we released our industry standard 0.9 PDK to select customers. It is our highest-quality 0.9 PDK, meeting industry standard yield and performance criteria at release. As is standard practice in the foundry business, I have consistently maintained that we will not disclose customer details. That being said, I remain highly encouraged by the progress we are making with multiple potential customers on 14A.

On Advanced Packaging, we are pleased by the progress we are making with EMIB and EMIB-T for our internal products. Both continue to see significant external customer traction as well. We continue to improve on execution, working with our supply chain to secure the capacity to meet the needs of our Advanced Packaging customers. We are also seeing robust progress in the readiness of HBI, which is an enabler of 3D chip stacking for our upcoming server products.

Across our manufacturing network, additional capacity continues to come online. Intel 3 tool installation is progressing well in Ireland, as is in the ramp of the 18A in both Oregon and Arizona, including Fab 62, our second high-volume 18A fab. In New Mexico and Malaysia, we are adding capacity to support strong customer ramps for advanced packaging next year.

In New Albany, Ohio, the team recently achieved important construction milestone with installation of the first major structural truss, and we continue to evaluate unique and highly differentiated opportunity to accelerate our Ohio buildout. And I hope to share more on this in the near future.

Looking to the future, I have always emphasized that rebuilding Intel is going to be a multiyear journey, and I remain firmly committed to making it happen. We are only 1 year into this journey. We have made enormous strides in these 18 months, including dramatic resizing of our workforce, eliminating middle management and bureaucracy, streamlining our executive ranks and establishing an almost entirely new leadership team. We have installed once again an engineering-centric culture, streamlining our product portfolio and put in place a more competitive road map for Intel products and Intel Foundry. We still have much more to do.

The resurgence in demand for CPU is meaningful tailwind for us along our transformation journey. Increasingly, energy efficiency is becoming the most important metric for the AI infrastructure buildout. While GPUs are great at thinking tasks, deterministic CPU are far more efficient at finding tasks. While GPU reasoning query run as high as multiple watt-hours of energy, a single CPU query could cause only [ joules ] of energy.

Applying even conservative estimates to the total number of digital agents and agent serve per CPU, it is easy to anticipate a massive dollar and unit TAM for CPUs over the coming 5 years. By our estimates, server CPU TAM, largely driven by agentic workloads, could increase as much as 5x from where it is today, from $40 billion to more than $200 billion over 5 years. This makes the opportunity for Intel enormous.

We're, of course, mindful of fierce competition underway as nearly every fabless company race to meet this unprecedented demand with their own offering. But Intel is uniquely positioned to benefit from this demand in 3 ways. First, as a custodian of x86, the most widely deployed ISA in the world, we are doubling down on x86 innovation, have an exciting Tick-Tock road map of CPO innovations in the pipeline. In addition, for the first time, we are making fully modular verified instances of our x86 CPUs, optimized for the custom workloads available to our customers.

Second, as 1 of the just 3 advanced node semiconductor foundries in the world and the only one domiciled in the United States, we are well positioned to support not only our own demand, but also the demand of wider fabless industry. We are also open to building any CPU at scale for our customers fully agnostic to underlying ISA.

Third, as modern computing platforms increasingly become complex system in the package, our leadership in advanced packaging offers highly differentiated solution, especially to meet the requirements of the massive AI infrastructure buildout underway. This is why I remain optimistic on the opportunities ahead for Intel even as we have work to do.

Lastly, while we remain focused on executing against our near-term priorities, we cannot lose sight of what's to come next. Intel has a deep portfolio of technology and IP that positions us to drive the next wave of computing. We are continuing to invest in breakthrough technologies, including silicon photonics, neuromorphic computing, advanced materials and quantum computing that we believe can unlock the importance of new opportunities over time.

This week, we announced our intent to acquire Classiq, an Israel-based quantum software provider, moving our quantum efforts from foundational research towards commercialization. Classiq's over-arching hardware-agnostic software platform layer provides a gateway for more consumers to explore quantum computing, and it is a wonderful complement to our already strong internal quantum hardware and software stacks, and enable us to further extend our compute platform offerings, to include QPUs in addition to CPUs, GPUs and ASICs. Post acquisition, Intel will become a full-stack revenue-generating quantum computing company.

In summary, while we did well in Q3, there's clearly more work to do. I'm disappointed that we were not able to meet strong customer demand in this quarter. We face near-term challenges, including manufacturing, supply shortage, the unprecedented memory crunch and the resulting softness in PC demand. But I remain confident in our path forward. We need to address these challenges with urgency, without losing sight of the massive opportunity in front of us.

I want to thank our employees, partners and suppliers for their tireless effort in this quarter. With that, I will turn the call over to Dave for additional details.

David Zinsner

Thank you, Lip-Bu. We delivered a strong quarter, exceeding our guidance for revenue, gross margin and earnings per share. This marks the eighth consecutive quarter of exceeding our financial expectations. Our Q3 results reinforce the essential role our server CPUs play in the build-out of Super Intelligence or AI infrastructure. Results also demonstrate our improving execution as we grow supply to capture the strong demand for AI compute.

Third quarter revenue was $16.5 billion, led by record year-over-year growth in DCAI. Strong execution across our factory network and external supply chain drove better-than-expected server CPU output. This strength was partially offset by weakness in client where industry-wide PC consumption is being impacted by memory shortages and limited IT budgets.

Third quarter non-GAAP gross margin was 44.1%, approximately 200 basis points higher than Q2 in our guidance. The sequential improvement was driven by improved costs on Intel 3 and Intel 18A and a more favorable product mix.

Third quarter non-GAAP EPS was $0.52, compared with our guidance of $0.38. The upside was driven by higher revenue, stronger gross margin and disciplined spending, which together produced better-than-expected operating leverage.

Operating cash flow was $5.1 billion. We ended the quarter with approximately $54 billion in cash and short-term investments, in addition to $15 billion of borrowing capacity under our new extended revolvers, up from $10 billion previously.

In early August, we further strengthened our balance sheet through a $23 billion equity offering. The incremental interest income associated with the offering contributed approximately $0.01 to Q3 EPS that was not contemplated in our original guidance.

Now moving to segment results. DCAI revenue was $7.3 billion, up 17% sequentially and nearly 80% year-over-year. Revenue was meaningfully ahead of our expectations as our factory network and external suppliers supported higher output. The work our teams have done to increase supply is having a tangible impact. Based on our current outlook, DCAI revenue is on track to finish 2026 nearly $10 billion above expectations entering the year.

Q3 growth was broad-based across hyperscale, enterprise and neocloud customers. We also continued to see strong momentum in purpose-built silicon with revenue increasing more than 50% sequentially and more than doubling year-over-year. Despite our progress to increase capacity, demand for DCAI products continues to meaningfully exceed available supply.

DCAI operating profit was $3 billion, representing 42% of revenue and increasing approximately $600 million sequentially. The improvement was driven by higher revenue and lower operating expenses. Granite Rapids is now the largest product family in our server portfolio as we continue to increase output from our Ireland facility.

Earlier this month, at the Open Compute Project, Global Summit, we demonstrated select Clearwater Forest configurations that scale beyond 46,000 cores per rack at approximately 133 kilowatts of total rack power, delivering best-in-class performance per watt. We also introduced the Intel GPU D Series, formerly code-named Crescent Island, which is designed to improve the economics of sustained memory-intensive AI inference. Customer interest in both product families is running ahead of our initial expectations.

CCPG revenue was $8.1 billion, down 9% sequentially and below our expectations. We previously noted that memory shortages could pressure the second half PC TAM. That pressure intensified during Q3, particularly late in the quarter, and resulted in a meaningful adjustment in customer purchasing patterns. Despite the near-term weakness, we saw several positive developments. Our AI PC revenue grew 35% year-over-year, even as total CCPG revenue declined 5%. Our client products on Intel 18A continued to ramp well. Wildcat Lake, our cost-optimized Intel 18A client CPU increased units by approximately 70% sequentially and is sold out. Lastly, our edge business was more resilient, supported by continued momentum in edge and physical AI design wins.

CCPG operating profit was $2.3 billion, representing 28% of revenue and approximately flat sequentially, primarily due to lower inventory charges.

Turning to Intel Foundry. Intel Foundry revenue was $6.7 billion, up 16% sequentially, driven by better-than-expected growth on Intel 3 and Intel 18A. External foundry revenue was $411 million, an increase of $118 million sequentially, primarily due to a nonrecurring customer program.

Intel Foundry reported an operating loss of $1.9 billion, an improvement of $236 million sequentially, even as we continued to invest aggressively to prepare Intel 14A for risk production next year. Higher yields, better cycle times and increased factory utilization continue to reduce wafer costs. Intel 3 costs improved approximately 15% from Q1 levels, while volumes increased in the mid-teens sequentially. Intel 18A costs have declined approximately 40% since Q1, reflecting continued yield and cycle time improvements.

Intel 18A yields continued their upward trajectory to industry standard levels, ahead of our internal projections. Since Q1, Intel 18A client CPUs have grown from less than 5% of revenue to now nearly 1/3 of client CPU revenue. These results represent meaningful progress, although we still have considerable work ahead of us to deliver a sustainably profitable foundry business.

Now turning to outlook. Demand signals from our data center customers continue to strengthen. We expect the industry to remain supply-constrained throughout 2027 and into 2028. As Lip-Bu discussed, the adoption of agentic AI is broadening the role of CPU compute. AI workloads are moving beyond crop-based responses toward multistep reasoning, orchestration and autonomous persistent action. Our CPUs play a central role in enabling these workloads.

As a result, we continue to see strong demand for our DCAI products. We now expect industry server CPU units to grow more than 30% in both 2026 and 2027. Growth would be higher in both years if additional industry supply were available. At the same time, the PC market is weakening after remaining resilient during the first half of 2026. We expect the 2026 PC unit TAM to decline approximately 12%, with second half units down more than 20%. We do not expect a sustained recovery until memory availability improves and pricing begins to normalize. For 2027, we're planning for the PC unit TAM to decline roughly 10%. We expect year-over-year declines to be more pronounced during the first half, with the potential for improvement in the second half against an easier comparison.

Our longer-term view of the PC market remains unchanged. Traditional PCs will play an important role in enabling AI applications and driving strong returns on AI infrastructure spending. In addition, we're just now seeing new agentic systems in the market. We also continue to see attractive longer-term opportunities in edge and physical AI as a genetic applications spread across the economy.

Taking these factors together, we're guiding Q4 revenue to a range of $15 billion to $16 billion. At the midpoint of $15.5 billion, we expect a gross margin of 38%, a tax rate of 11% and EPS of $0.25, all on a non-GAAP basis. Our guidance includes an impact of roughly $300 million to revenue and 200 basis points to gross margin based on our early assessment of the incident at Fab 28.

By segment, we expect DCAI revenue to grow mid-teens percent sequentially, with purpose-built silicon growing faster than DCAI overall. Please note almost all of the $300 million impact from Fab 28 is related to 10-nanometer server products.

We expect CCPG revenue to decline greater than 20% sequentially as customers move quickly to reduce weeks of inventory and align purchases with their expectations for 2027 consumption. The sequential decline in Q4 gross margin primarily reflects lower overall revenue against our fixed cost base as we adjust client wafer starts to current demand conditions. In addition, revenue mix, especially the strong growth expected in purpose-built silicon, is creating difficult quarter-on-quarter comparison. These headwinds will be partially offset by continued yield and cost improvements on Intel 3 and 18A.

We expect non-GAAP operating expenses of approximately $4.4 billion in Q4 and approximately $16.3 billion for the full year. We expect noncontrolling interest, or NCI, to be approximately $250 million in Q4, and we currently expect NCI of approximately $1.1 billion in both 2027 and 2028. Lastly, we expect Q4 fully a diluted share count of 5.44 billion, representing a full quarter of shares issued for our August equity offering.

Moving to CapEx, our 2026 gross capital expenditures remains on track to be $20 billion, and we expect 2027 gross capital spending to be up strongly. We have made good progress securing the equipment required to support our growth plans, with firm purchase orders to equipment suppliers increasing nearly 40% sequentially. We're also accelerating [ clean room ] build-outs, including Fab 62 in Arizona, and exploring opportunities to accelerate in Ohio. Despite this increase in capital spending and working capital, we remain committed to deleveraging in 2027.

We will remain disciplined as we invest, and we will not spend $1 of capital unless we have confidence in the health of our products and process technologies and clear evidence of customer demand. Our confidence in all 3 is improving.

In closing, I want to offer a few words outlining the opportunity we see ahead. Our core server CPU market is growing faster than ever, and edge and physical AI represent meaningful incremental opportunities for client-class CPUs even as we navigate the current market conditions.

Importantly, as we improve our product execution and road map, we have opportunities to regain share and grow even faster than our underlying markets. Beyond our core x86 franchise, we are investing in 3 large opportunities where we have meaningful momentum but relatively modest revenue today: purpose-built silicon, advanced packaging and external wafer foundry. Each represents tens of billions of dollars of new markets where our technology, manufacturing assets and customer relationships provide us a differentiated opportunity to participate.

Taken together, we have line of sight to long-term double-digit revenue growth with meaningful leverage and improving margins. We have significant work ahead of us. However, the combination of stronger markets and the cultural changes set into motion by Lip-Bu to improve execution, transparency and discipline gives us increasing confidence in Intel's long-term earnings power and free cash flow growth, a solid foundation for building sustainable shareholder value.

With that, I'll turn it over to John to begin the Q&A.

John Pitzer

Thank you, Dave. [Operator Instructions] With that, Jonathan, can we take the first question?

Operator

Certainly. And our first question comes from the line of Timothy Arcuri from UBS.

Preguntas y respuestas

Timothy Arcuri

Dave, I'm wondering with all the puts and takes and what's going on in PC. I'm wondering if you can give us maybe some guardrails maybe you could say it next year? I mean it seems like DCAI could grow $10 billion to $15 billion year-over-year, but PC, I'm guessing, is maybe down a touch. So can you just maybe give us guardrails for the year and sort of how to think about the shape of the year?

David Zinsner

Yes, Tim. Yes, I'm not obviously going to guide next year, but maybe I'll give you some qualitative information. I would say at a high level, we see next year as a growth year. We feel like the momentum that we saw in data center this quarter, and taking away the impact from Israel, that momentum would have continued into the fourth quarter, we see that continuing next year.

The only little nuance is the first half of the year, we are still a bit constrained. We've talked about putting 40% more capacity in substrates, but a lot of that doesn't hit until the back half of the year. In addition, we are still constrained on Intel 3, and we're going to work towards moving, as Lip-Bu said, some of our supply into Oregon, but it's going to take a while before that comes out of the fab and gets into products. But we're very optimistic and enthusiastic about the opportunities for us next year in terms of growth on the data center business.

In addition, we talked about within DCAI, we also have purpose-built silicon. That business has been growing at a 3x rate over the last 3 years. We see that growing at a 3x rate in the next 3 years. And so we're really bullish about that throughout 2027.

Obviously, the client is going to be the most difficult to call. As I talked about, we see TAM declining in '26 by about 12% on a unit basis. We think that's going to be about 10% next year on a unit basis. So that will be one dynamic. Then there's going to be this kind of inventory correction. We're in the throes of that right now. We felt a little bit of it in the third quarter; we're going to feel a lot of it in the fourth quarter. We think it extends at least out into the first quarter. After that, I think you'll start to see like the inventory correction part of this start to kind of wane and we'll be back into more of like what's going to go on, on the TAM side, which, as I said, we're thinking about 10% down.

The one added piece of this is we come into next year with Nova Lake as our product offering. And we think this is the best set of products we've had in a very long time. We think it covers the gamut of what's required to participate in client. We're very excited about it. We think next year and in the future years, we're in a kind of a market share gain type of environment for our clients. So obviously, unfortunate that we run into this challenge on the memory side and its impact on client, but we are very enthusiastic about the product side.

And I just -- you didn't ask this question, but I'll just extend it out beyond this year. We do think that the PC is going to go through a renaissance despite this. And we think that as AI workloads start to shift around, there's going to be a lot of workload required in the client space, either as a PC or a some kind of agentic box. And so we do see this market as being pretty interesting. It's got to obviously get through this bit of challenge that we have. But longer term, feel really good about the market.

John Pitzer

Tim, do you have a quick follow-up question?

Timothy Arcuri

I do. Yes. So Dave, you did say that CapEx is going to be up a lot next year. I think a lot of us are thinking kind of like $30 billion. But I guess the real question is, are you committed to generating free cash flow next year? Or do you think because of the CapEx burden, that that will be hard?

David Zinsner

Yes, it's a good question. Yes, we haven't exactly locked into the number. We don't do that until we get to towards the end of this year. But we have kind of a directional understanding of what's likely to happen. And we have been putting POs in place with equipment vendors given the lead time. So we have some sense of what's going to happen.

I think next year is an investment year for us as we look at the year. Probably not a dynamic where we're going to have positive free cash flow next year. But it's because we see this as a tremendous opportunity to invest into the growth. And one thing that we have seen is you make the right investments, you ramp the node in the right way, you see profitability and you see return on invested capital. And we talked about this as we were doing the equity offering at the beginning of August, that we think we should be getting at least 20% ROI on all of these nodes as we invest in them.

But this -- next year is going to be one of those years where there's a couple of things going on at the same time. We're ramping 18A, we're ramping 18A to some extent, and we're making the appropriate investments to position 14A to begin production in '28. That's a lot of investment all kind of hitting us in 1 year. So I don't think that this will be a year in which we see positive cash flow.

And by the way, that is why we raised the money that we did at the beginning of August, was to get ourselves positioned to do this. Longer term though, these businesses generate lots of cash flow. And if you look at Intel 7 today, kind of 10-nanometer 7, that's already a profitable node to us. Intel 3 is going to be profitable next year for us. So we know we can make these businesses profitable. We know we can generate good cash flow from them. We know we can get 20% return. We just got to get through what is basically a [ nut hole ] here in '27 to go realize that.

Operator

Our next question comes from the line of Ben Reitzes from Melius.

Benjamin Reitzes

I wanted to ask about some of the issues that you noted for 4Q and that may linger into 1Q and see if you can kind of size them versus prior expectations and then when they get better. You cited execution, obviously, the Israel situation, which is $300 million. Wondering what that's going to be in the quarters after that and when that is solved. And sorry that happened, by the way. And then the client. And in the context that the guide is about [ 1.8 ] below the Street, maybe you can kind of go through those 3 things, how big they are to the cut of [ 1.8 ] and when they get better, that would be great.

David Zinsner

Yes. What was the -- so Israel, the client, what was the third one?

John Pitzer

Just relative to the guide in the Street.

Benjamin Reitzes

The execution, Dave. There was an extra comment around execution, and Lip-Bu and made some comments. I'm wondering how big that was in terms of the guidance cut.

David Zinsner

Okay. Well, on Israel, we've been shut down for about 2 weeks. We feel like we're going to be starting up production here by the end of this week. While we're doing that, we're trying to assess all the wafers that are in the line and what's usable and what's not usable from those materials. So this is somewhat of an early assessment, both really of 4Q and 1Q. But I think the current thinking is that we'll have about a $300 million impact to the fourth quarter. That's likely to be similar in the first quarter.

We think most of the write-offs of wafers or scrapping of wafers is likely to all kind of happen in the fourth quarter. That's kind of what's driving the 200 basis point. But I wouldn't expect to see additional scrapping going on in the first quarter. But just keep in mind, it's early, we're going through all of that as we speak, and we'll, of course, update you as we go along.

I think on the client side, obviously, it's going to kind of reset to this level, and we're going to have -- I think, usually, you have a very similar inventory correction quarter-to-quarter. So I would expect it to generally be similar other than there will be more -- you will have a seasonality effect in Q1 on client as well. So the likelihood is that whatever the level that we're talking about for the fourth quarter, we'll see somewhat of a seasonal decline next quarter.

Beyond that, I think from an execution perspective, I have to think about some of the negative things, but a lot of the execution has been really positive. I mean Lip-Bu has -- and the team have done a very good job driving Intel 3 yields, driving 18A yields and working with the substrate vendors. Lip-Bu was in Japan talking to many of the substrate vendors, to make sure that we were getting allocated the appropriate supply. So I think we're as positioned -- or as well positioned as anyone out there in terms of delivering supply. It's tight. It's going to be tight this next quarter. It's going to be tied into the first half. But I think we've positioned ourselves well that, beyond that, we should start to see some pretty healthy growth in the data center business.

John Pitzer

Ben, do you have a quick follow-up?

Benjamin Reitzes

Yes. Then just with regard to the server velocity, there's shortages and whatnot. But do you have an idea for when you think you can ramp and grow in line with the market that's going on that $200 billion plus? Is that more second half next year based on your commentary?

Lip-Bu Tan

Yes. Good question. I think, Ben, first of all, I think the server, the biggest revenue for us is the Granite Rapids, that's on the Intel 3. Clearly, we are continuing to drive efficiency in terms of yield. And then secondly, we also try to expand the capacity. And I think that Dave mentioned about Oregon, we are planning to have that Intel 3 in the second half of next year to be in the production so that we can supply some of the demand from the customer.

And so I think overall, I think we'll be good to address the Intel 3. And then clearly, we have the Clearwater Forest coming up, and we are in the shipping [indiscernible] on 18A, we should have capacity for that. And then Diamond Rapids, that will be 2027, we have 18A-P, we'll be ready for that. And of course, the Coral Rapids coming up in 2028, that will be on the 14A. So those, I think, we'll be in pretty good shape.

Operator

Our next question comes from the line of Joseph Moore from Morgan Stanley.

Joseph Moore

Yes. I mean along the same lines, does the PC weakness give the opportunity to reallocate wafers and sort of address some of this and accelerate that transition?

David Zinsner

The answer, yes. We had -- what we're taking off-line to put Intel 3 into Oregon is 18A wafers, which we're generally going to build Panther Lake products. So we see the opportunity with some of that capacity freed up to go produce more data center wafers.

Just keep in mind that we're not only constrained on the wafer side, we're also constrained on the substrate side. So it's important we get the substrate piece of this ramped up to be able to meet the demand.

John Pitzer

Joe, do you have a follow-up question?

Joseph Moore

Yes, I do. How do you see server market share next year? I think you talked about 30% unit growth. AMD has guided to 70% revenue growth. The ARM camp seems to be growing quite quickly. Just I understand you're constrained, but how do you sort of see market share playing out?

Lip-Bu Tan

Sure. I think that's a good question. So clearly, we have a road map. We like what we have, in terms of Coral Rapids, we will be competitive to AMD. And the next-generation product, the Tick-Tock, we'll be ahead of the AMD on that. So I think stay tuned, we'll continue to drive leadership. CPU is really our franchise. We are determined to drive success there. And then I think that's a good opportunity for us to really drive that.

David Zinsner

I'd say maybe one other addition is I really see market share next year being a function of supply. And so the more we can improve our supply on a relative basis, the better we'll do in terms of market share.

Operator

Our next question comes from the line of Stacy Rasgon from Bernstein Research.

Stacy Rasgon

I wanted to dig into gross margins a little bit. So I get that the 38% guidance, I guess, would have been 40% without the unfortunate situation in Israel. But as I'm looking into next year, how do I think about the overall drivers? I've got a -- clearly, I have a client flush and maybe utilization issue, which is a negative. I've got depreciation probably going up from CapEx, which is a negative.

And then maybe like as 18A and 3 ramp, I guess the yields are getting better, but I guess you can tell us if the gross margins on those parts are still below 10.7%. I wonder if that mix shift is a headwind plus ASICs and packaging and everything.

And maybe on the other side though, maybe mix towards servers as a positive, maybe pricing helps. I'm just trying to get a feeling of the gross margin evolution in this kind of an environment given all of those puts and takes? It just seems like there's a lot of them. I'm hoping you could...

David Zinsner

That's a good question, Stacy. There are definitely a lot of moving pieces. I think from a margin perspective, the Israel piece, it's, probably on a margin basis, is going to be really more of a Q4 event. I don't expect a significant amount in Q1 in terms of scrapping. Usually, if we can identify it, we'll scrap it in the quarter. We have Xeon demand remains really strong. And I think I mentioned Granite Rapids is now margin accretive for us. So growth in Granite Rapids is helpful to gross margins. And we will also continue to be focused on both Intel 3 and 18A.

Their structural cost is what we can focus on more, as well as yields, throughput, all of those. We did very well in Q3, we expect to do better in Q4. And the goal is to continue to drive that through the year.

You're right on 18A, the loading isn't going to be where we want it. We're fixing a little bit of that with putting Intel 3 in Oregon. But nevertheless, there will be a little bit of a under-loading scenario for 18A that's going to be a little bit of a headwind.

The other headwind for us, which is, I think, a positive viewed the right way, is our purpose-built silicon business, and it's kind of in its early stages, its margins are below the corporate average. But we see tremendous growth there. And Lip-Bu and I really decided to focus on driving the gross profit. And so we're just going to grow the gross profit. That will grow the earnings. It may put a little pressure on the operating margin percentage through next year, but I think it will be goodness at the bottom line.

I think when you put all that together, and it's so many puts and takes, it's hard to quantify, I still lean towards the [ 40 to 60 ] fall-through. That's probably what we're shooting for. So if we have a growth here next year, we should see gross margin expansion given that kind of rule of thumb. It may not work out every quarter that way. But I think when you look at it on a year-over-year basis, it's going to be in that kind of range.

John Pitzer

Stacy, do you have a follow-up question?

Stacy Rasgon

I do, yes. I guess I wanted to ask about PC. So I mean, Q3 was probably the worst Q3 I've seen for PCs in at least 20 years, maybe more, sequentially. So it does sound like customers probably built inventory in Q2. I guess, can you confirm that? And then it sounds like that they're flushing out a bunch in Q4. I'm trying to get -- and maybe even in Q1. It sounds like you think things get better in the second half as, I suppose, memory supply. What gives you confidence that memory supply is actually going to improve? If you listen to [ the AMD ] guys, they seem to be suggesting that it won't.

David Zinsner

Yes, let me -- yes, it's a good question. Stacy, let me be clear. We think that when you're dead-on that Q4, Q1 are going to be likely the 2 quarters in which there's most of the inventory correction, and that impacts us more than the TAM declines. So we will see that in the first, call it, Q4, Q1, maybe stretching a little bit into Q2.

After that, it does recover back to where, theoretically, where billings to us are somewhat equivalent to consumption. So we get a little bit of a recovery from that. But make no mistake, we are not optimistic about next year in terms of memory. We think that it is going to impact the TAM. We think TAM will be down 10% on a unit basis next year. And then we'll have to see in '28 whether things improve. But our expectation is that it does not improve in '27.

Operator

Our next question comes from the line of Vivek Arya from Bank of America Securities.

Vivek Arya

For the first one, if I take your, I think, implied DC guidance for, that puts it at about $8.5 billion or so. And if I annualize that, that's like mid-30s billion in terms of a conceptual target for next year. So I'm curious, is that a doable floor? Can you do better? Is there kind of ASP expansion, just given demand is outstripping supply? Or are there other factors we should keep in mind as we think about how large your data center business could be in '27?

David Zinsner

Yes. I think that's a good starting point. I think that's the right way to think about it. What we would characterize is that the first half of 27 is likely to be tighter in terms of wafers and substrates. But we've put a lot of work in place, both in terms of as I talked about increasing our wafer output of Intel 3, improving yields as a free way to get more product, and the investments we've made in the substrates with vendors unlock that capacity in the back half of the year. So we do expect it to be big growth.

I mean keep in mind also that we roll our purpose-built silicon, our ASIC business in under that business unit. So that's part of what's driving a lot of the growth in the fourth quarter. And again, we expect that to continue in '27.

John Pitzer

Vivek, do you have a follow-up?

Vivek Arya

Yes. I was hoping to get an update on engagements with external customers in your foundry both on the packaging side? And then also more importantly, Lip-Bu, on the kind of the wafer processing side. Are you expecting external customers for 18A-P? Or should we assume that all the engagements will be on 14A? And when do you think Intel gets the confidence to move them from sort of the evaluation to the tape-out or tangible revenue column, so we're all kind of set in terms of when we should start modeling external revenue for Intel over the next few years?

Lip-Bu Tan

Good. Vivek, thank you so much for the question. So I think clearly the first question you have is about the packaging. Our EMIB-T making great progress. We have been very focused on capacity in the substrate, making sure that we can secure the capacity we need and also the yield profile for the capacity from substrate material.

Secondly, we focus on the yield, and we are improving substantially. And then thirdly, it's the cycle time. And then so far, under the leadership of Seok-Hee Lee, he used to be the CEO of SK Hynix, expert in the process integration. He's done fabulous job, very dedicated. I think we get a lot of capacity in the next 2 years in terms of the substrate material. And we're also improving the yield as the they go by.

And then overall, I think we're going to have some -- stay tuned, we're going to have a couple of customers already to announce in the next 2 quarters. And then, so clearly, we always want to wait for the customer. If they want to announce, we will respond to that.

And then secondly, on the 18A-P, we're making great progress in volume production from Panther Lake, and then clearly on the Wildcat Lake. And so I think those, I think, internally, we've been using. The yield, we are quite happy with the performance. And right now, we also have inbound interest from the external customer, and we are intensively working with them, and they are very pleased with the yield. And right now, we are focused on the die size, make sure that we can deliver to them and also make sure that IP ready for that. So I think 18A-P, stay tuned. We have some good progress on that.

On the 14A, we made tremendous progress. I think we announced the 0.9 PDK. We already released to our select customers. They already respond very actively working with us. And then the other part is also, I think, in terms of performance, yield, all the metrics and [ defect ] density and the variation, we are making daily improvement on that. So far, we are very pleased. We should be able to do the 1 PDK by Q1 next year, and we can start engaging with customers with products they like to use our fab. And so I think stay tuned again, we will have time to announce some of the customers when they are ready.

Operator

Our next question comes from the line of Harlan Sur from JPMorgan.

Harlan Sur

Commencing second half of this year, there's a number of new GPU and XTU platform launches. Most of these new launches are going to continue to ramp through the first half of next year. Given the strong CPU attached to these platforms and many of them are Intel x86 DPU-based platforms. And then additionally, in addition to the overall sort of supply dynamics, which remain tight, but the team is bringing on more capacity, yields are improving, as you mentioned. Given all of this, should we anticipate continued sequential growth in DCAI in the first quarter?

David Zinsner

Yes. I mean I would avoid guiding too much the quarter. I think your overall premise is right. There's going to be a strong driver of demand for CPUs. We see that continuing all through next year. We're working on the supply side of the equation to make sure that we can meet what customers need. We will likely not fully meet what they require. But I think we're going to be in a more challenging situation in the first half of the year than we are in the second half. So maybe that's one added data point I'd just make sure that you're using when you're modeling. Outside of that, I'd prefer to wait for another quarter and see how things kind of play out before we guide Q1's data center.

Lip-Bu Tan

Harlan, just to add on that. I think clearly you saw our announcement with Oracle Cloud. And we also have multiple customers, they are all interested to sign up the LTA agreement with us. So that make sure that we can secure the CPU rollout for them. So I think those are encouraging signs that CPU is high demand.

John Pitzer

Harlan, do you have a follow-up question?

Harlan Sur

Yes. On the factors that the team has control over and as a reflection of the continued yield, reliability, manufacturability improvements, I think back in Q2, both Panther Lake and Granite Rapids gross margins were below corporate average. But as Intel 3 and 18A saw further yield improvements in the third quarter, you did mention [indiscernible] Granite Rapids on Intel 3 accretive to gross margins in Q3. Did the team also drive core Series 3 Panther Lake gross margins above the low mid-40% level in Q3 as well?

David Zinsner

So we have definitely made meaningful improvement on the gross margins of Panther Lake. It's not yet at or above the corporate average, but we have made huge strides. And this is just a matter of grinding it out on yields and throughput and driving the right scale and so forth. And these will be accretive gross margins over time. It's just not quite there yet.

John Pitzer

Jonathan, we've got time for one last question, please.

Operator

Certainly. Then our final question for today comes from the line of C.J. Muse from Cantor Fitzgerald.

Christopher Muse

I guess first question, going back to CapEx, you talked up significantly. Curious how you're thinking about spending on WFE versus buildings. I think in the past, you've talked about majority of the incremental spend on buildings. I wanted to confirm that that's still kind of the direction that we're going.

David Zinsner

Yes. So we have, I think, a significant ramp in equipment spend this quarter, and it's expected to be even more significant next quarter. So a lot of what we're spending on now is really equipment. We have made the investments. I would call it, putting -- in a lot of cases, putting the finishing touches on some of the space and just most of the spend is going towards equipment, with the only exception we're still in the heavy construction phase of Ohio. But outside of that, it's really about equipment.

John Pitzer

C.J., do you have a quick follow-up?

Christopher Muse

Yes. On the PC side, I know you've talked about units down 10% next year for the industry. Curious if you could comment on how you're thinking about your mix shift and what impact that might have on your pricing?

David Zinsner

Mix shift in terms of like low end versus high end and so forth? Is that what you mean?

Christopher Muse

Yes, exactly.

David Zinsner

Yes. I mean we have -- I would say we have seen we have seen strong demand in Panther Lake on a relative basis. Obviously, the whole market is coming off, but Panther Lake is holding its own. We also put a lot of effort into a product called Wildcat Lake, which is a cost-optimized part. We've seen significant growth in Wildcat Lake. It's coming exactly at the right time because it's cost-optimized and we're seeing a lot of focus in that regard.

As we go into next year, we'll have Nova Lake. And I think the important aspect of Nova Lake is we've had a bit of a hole in our portfolio for high-end desktop, and Nova Lake filled in that gap. So I would expect us to do pretty well on a mix basis because we should see better traction on the high-end desktop as well. So we'll see how it goes. We do feel like we go into the year as a market share gainer with our Nova Lake portfolio. And obviously, we'll kind of monitor the market and see how things kind of play out.

Lip-Bu Tan

With that, I want to thank everyone for joining us today. We made good progress this quarter on our journey to transform Intel, but we still have a lot to work ahead of us. I'm looking forward to seeing many of you throughout the quarter and providing another update in January.

Operator

Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.