Intel's Q3 2025: Contradictions in Foundry Progress, AI Strategy, 14A Process, and Customer Engagement

Date of Call: None provided

Financials Results

• Revenue: $13.7 billion, up 6% sequentially, came in above the high end of guidance
• EPS: $0.23 per share (non-GAAP), versus guidance of break-even EPS
• Gross Margin: 40% non-GAAP gross margin, four percentage points better than guidance
• Operating Margin: 29% operating margin for Intel Products (operating profit $3.7B), up $972M sequentially

Guidance:

• Q4 revenue guidance $12.8B to $13.8B (midpoint ~$13.3B; roughly flat QoQ adjusting for Altera)
• Q4 non-GAAP gross margin ~36.5% at midpoint (down sequentially)
• Q4 non-GAAP EPS ~$0.08; tax rate ~12%
• Q4 non-controlled income ~$350–$400M GAAP; average diluted shares ~5B
• 2025 gross CapEx expected ~ $18B (CapEx guidance reiterated; 2026 planning ~$16B per management commentary)

Business Commentary:

• Revenue Growth and Strategic Priorities:
• Intel reported a revenue of $13.7 billion for Q3, exceeding guidance and up 6% sequentially.

• This growth was driven by strong execution in core markets and steady progress in rebuilding the company, marking the fourth consecutive quarter of improved performance.

• Supply Constraints and Demand Dynamics:

• Intel experienced supply constraints, particularly on Intel 10 and Intel 7, limiting their ability to fully meet demand for both data center and client products.

• Given the tight capacity environment, the company is working closely with customers to balance supply and demand by adjusting pricing and product mix.

• AI and Compute Market Opportunities:

• AI is driving near-term upside to Intel's business, with AI PCs expected to reach 100 million units by year-end.

• Intel is enhancing its x86 franchise and working to position it as a compute platform of choice for AI inference workloads, leveraging partnerships with companies like NVIDIA.

• Foundry and Financial Health:

• Intel's Foundry revenue was $4.2 billion, down 4% sequentially, with improved operating loss compared to the previous quarter.
• The company's improved financial health includes receiving $5.7 billion from the U.S. government and securing strategic investment partnerships, enhancing its cash position to $30.9 billion.

Sentiment Analysis:

Overall Tone: Positive

• Management called out a "solid Q3" with revenue, gross margin and EPS above guidance, the fourth consecutive quarter of improved execution; emphasized a strengthened balance sheet with ~$30.9B cash and strategic investments, and expressed growing confidence in AI-driven demand and Foundry momentum.

Q&A:

• Question from Ross Seymore (Deutsche Bank): Do collaborations/equity investments or technical progress drive increased confidence in Foundry? Response: Pat: Confidence stems from both strategic investments (e.g., SoftBank) and tangible technical progress on 18A/14A yields, Fab 52 operations, and strong advanced-packaging demand.

• Question from Ross Seymore (Deutsche Bank): What are the drivers for gross margin forward (2026) and will Foundry be the biggest driver? Response: Dave: Foundry margin improvement will be driven by scale and mix toward leading nodes (18A/Intel 3/4); product ramps (Panther Lake) are initially dilutive but improve over time; expect a ~40–60% fall-through range tied to mix.

• Question from Joseph Moore (Morgan Stanley): How are customer commitment conversations progressing for Foundry (do customers expect capacity built ahead of time)? Response: Pat: Intel is building trust by demonstrating yield, reliability and IP; customers engage via milestone-based work and attach chips to validate designs before committing capacity.

• Question from Joseph Moore (Morgan Stanley): Where are current supply constraints coming from and how will they be resolved? Response: Dave: Constraints are across Intel 10 and 7 capacity (and external substrate shortages); demand shaping and existing assets/under-construction capacity give flexibility to increase supply over time.

• Question from CJ Muse (Cantor Fitzgerald): Is the demand> supply outlook into 2026 focused on server or also clients and how should Q1 seasonality be viewed? Response: Dave: It's both server and client; Intel will prioritize server/wafer capacity; Q1 may see peak tightness and likely will not buck normal seasonal Q1 decline.

• Question from CJ Muse (Cantor Fitzgerald): With improved cash position, how will you approach CapEx and other investments? Response: Dave: Priority is deleveraging; remain disciplined on CapEx, target ~$16B for next year but will flex and reallocate mix if customer commitments justify more investment.

• Question from Blaine Curtis (Jefferies): Is the $18B CapEx still the plan and when will 18A capacity be added? Response: Dave: $18B remains the 2025 plan though lumpy; don't expect significant near-term incremental 18A capacity—18A is a long-lived node with gradual ramp.

• Question from Blaine Curtis (Jefferies): Where are 18A yields today vs. a 'successful' node and how will that layer into margins? Response: Dave: 18A yields are adequate and will hit year-end goals but need further improvement to be fully accretive; meaningful margin benefits expected over next year+ as yields mature.

• Question from Stacy Rasgon (Bernstein Research): Customers still buying older products—how will you transition them to newer nodes given capacity constraints? Response: Dave: AI PC adoption is growing (sequential double-digit), but Windows refresh and legacy demand remain strong; ecosystem and ISV adoption will drive gradual transition over time.

• Question from Stacy Rasgon (Bernstein Research): Did you mean 18A yields won't be fully productive until end of next year and you won't add much 18A capacity next year? Response: Dave: We will ramp 18A volume but not dramatically add new 18A capacity next year; yields will improve through next year and become more margin-accretive thereafter.

• Question from Joshua Buchalter (TD Cowen): Scope of ASIC/fixed-function efforts — for foundry customers or Intel products and what applications? Response: Pat: Central Engineering will expand ASIC/design services to deliver purpose-built silicon for external customers and internal products, leveraging x86 IP and packaging for AI and other workloads.

• Question from Joshua Buchalter (TD Cowen): Any update on 14A decision since balance sheet changed? Response: Pat: Engagement with customers on 14A has increased, confidence improved, and Intel is investing in talent and technology definition to support it.

• Question from Ben Reitz (Melius): Update on Nvidia relationship timing, customer feedback, and materiality? Response: Pat: Multi-year, engineering-heavy collaboration (NVLink) to integrate Intel CPUs with Nvidia accelerated compute to create new product classes; seen as incremental TAM expansion.

• Question from Ben Reitz (Melius): Is the inference strategy mostly partnering or based on specific Intel IP? Response: Pat: It's partnership-focused plus revitalizing x86 for purpose-built CPUs/GPUs for inference and agentic AI; will partner with incumbents and emerging players.

• Question from Timothy Arcuri (UBS): If you moved off Intel 10/7 onto 18A, how would gross margin normalize? Response: Dave: Moving to leading nodes materially improves cost structure and margins over time, but competitiveness, product quality, and multi-year startup costs also drive margin improvements—this is a multi-year transition.

• Question from Timothy Arcuri (UBS): Update on data center roadmap (Diamond/Coral Rapids)? Response: Pat: Diamond Rapids getting stronger hyperscaler feedback; Coral Rapids (with SMT/microthreading) is in definition and will be progressed as roadmap is finalized.

• Question from Aaron Rakers (Wells Fargo): Will NVLink integration expand beyond NVLink (e.g., other accelerators like Gaudi)? Response: Pat: NVLink is the primary connectivity hub for CPU/GPU integration now; Intel is defining additional products (e.g., Crescent Island) for inference/agentic AI and will provide updates over time.

• Question from Aaron Rakers (Wells Fargo): Outlook for non-controlling interest (NCI) expense for 2026? Response: Dave: NCI for 2026 estimated roughly $1.2B to $1.4B.