Compiled by Dolphin Research: $Intel(INTC.US) FY26 Q2 Earnings Call Trans

I. Results Recap

1. Q2 beat across the board, seventh straight beat: revenue of $16.1bn, $1.8bn above the midpoint. AI-driven businesses grew over 70% YoY and contributed ~70% of revenue, with data center hitting a record.Non-GAAP GPM was 41.8%, ~+280bps vs. guide on revenue upside, better yields, and higher ASPs from mix and pricing; Non-GAAP EPS was $0.42 vs. $0.20 guided. Revenue growth is the strongest in ~15 years.

2. Q3 guide: revenue $15.8–16.8bn (midpoint $16.3bn); Non-GAAP GPM 42%, tax rate 11%, EPS $0.38. Full-year Non-GAAP opex ~$16.5bn.NCI is expected at ~$250mn in both Q3 and Q4; NCI of ~$1.1bn/yr in 2027–2028 (GAAP). Mgmt reiterated that keeping GPM above 40% each quarter is the top financial goal this year.

3. Cash flow and balance sheet: Q2 CFO of $7.0bn. Cash and ST investments were ~$30bn at quarter-end, plus a $10bn revolver, for total liquidity of ~$40bn.The company proactively delevered to protect IG ratings; ~$10bn of non-core assets remain monetizable (no disposal plan yet). If upside outperforms, capital markets financing is possible.

4. CapEx raised materially: on strong demand signals, 2026 CapEx is guided to $20bn+, up meaningfully vs. initial plans (+~$3bn this year), with tooling spend +40% vs. 2025. 2027 CapEx will be well above 2026, with the vast majority going to the U.S.Cumulative U.S. fab and equipment investment in 2021–2026 will approach $100bn, higher than any other semi company over the same period.

II. Call Details

2.1 Management Commentary

1. Strategy and supply/demand

a. Demand continues to outstrip supply: the industry faces one of the most acute shortages in history across advanced logic, wafers, memory, and substrates, with limited relief near term. Intel benefits from three strategic assets—x86 CPUs, advanced packaging, and a large foundry network.b. As AI moves from training to inferencing to agentic and multi-agent, CPU density in general-purpose servers keeps rising, and core server CPU growth is at a record pace.c. Deeper collaboration with Google Cloud to accelerate an internal AI-first shift; the operating discipline launched 15 months ago is showing up in higher org efficiency, faster decisions, and tighter customer proximity.

2. Intel Foundry and advanced nodes

a. Q2 foundry revenue was $5.8bn, +6% QoQ; 18A output ran ~25% above plan and grew 50%+ QoQ; external foundry revenue was $293mn. Foundry OP loss was $730.8mn, narrowing QoQ on yield gains, cycle-time improvement, and wafer-cost declines as Intel 4/3 and 18A scale.

b. 18A: ramped across multiple commercial/consumer products with yields ahead of plan. Panther Lake single-SKU cost is down ~50% YTD and should drop another ~20% this year, with further declines in 2027.18AP risk production is underway (≈+5% vs. 18A), IP/design compatible with 18A, targeting external customers.

c. 14A: PDK 0.5 is done; PDK 0.9 expected by Oct (a key milestone). 256 SRAM yield, defect density, and transistor performance are all ahead of 18A at a comparable stage.Internal products plan risk production in H2 2027, and full production is set for 2028. External interest has risen notably on the back of PDK 0.9 and yield progress.

d. Advanced packaging: EMIB-T is seeing strong interest and a growing backlog, with yield and reliability on target, aiming for volume for customers by 2027.

3. Data Center & AI

a. Q2 revenue was $6.3bn, +24% QoQ and +59% YoY, well above expectations on hyperscaler and enterprise demand. OP was $2.5bn with 40% margin, up by about $1.0bn QoQ.Q2 server YoY growth hit a record.

b. Purpose-built silicon revenue rose ~20% QoQ and nearly tripled YoY.

c. Launched the first 18A server part, Xeon 6+ (codename Clearwater Forest). Announced rack-scale and disaggregated inference solutions with SambaNova and Foxconn, and new controllers/adapters supporting 10–200Gb Ethernet.

d. Server roadmap: Clearwater Forest, Diamond Rapids, and Coral Rapids (adds SMT). Granite Rapids (Intel 3) demand is exceptionally strong and supply-constrained.

4. Client Computing & Physical AI (CCPG; former PC)

a. Q2 revenue was $8.9bn, +15% QoQ and above plan; OP was $2.3bn with 26% margin, down by ~$173mn QoQ due to inventory write-downs tied to factory-network optimization.

b. AI PC revenue rose 26% QoQ and now accounts for about two-thirds of client revenue. Edge deployments are ~10% of CCPG.

c. 18A is in full ramp with 400+ Series 3 designs. Arc iGPU has 40+ designs, and the Arc G lineup targets handheld gaming; vPro activations surged 1,500% over the last four quarters; edge AI secured 130 Series 3 designs, including robotics 'brain and control' deployments.

d. The renaming to CCPG reflects edge AI opportunities. Mgmt believes edge and physical AI TAMs can rival client TAM over time.

5. ASIC/Design Services and new opportunities

a. Design services revenue nearly tripled YoY. Mgmt sizes ASIC TAM at $100bn+.With end-to-end design, a broad IP portfolio, and advanced packaging/foundry capabilities, Intel is expanding from networking into compute and ultimately accelerators.

b. Announced a next-gen security processor collaboration with Fortinet (a key ASIC milestone). Intel IPUs are shipping to select hyperscalers.

c. Advancing heterogeneous AI and disaggregated inference via a multi-year partnership with SambaNova.

6. Memory strategy: memory is a major bottleneck, and Intel is working closely with the three leading memory vendors to secure supply (top priority). The former $SK Hynix(SKHY.US) (SK Hynix) CEO recently joined.Intel has deep roots in memory and is exploring compute-memory integration, stacking, and efficiency improvements (internal efforts include Z-Angle Memory and cross-batch memory).

2.2 Q&A

Q: What does the CapEx raise (≈+$3bn this year, materially higher next year) imply for foundry customers? Any firm 14A or 18AP orders? What about packaging?

A: The CapEx increase is broad-based and includes advanced packaging. We are constructive on EMIB-T and will keep investing; front-end fabs still require far more capital than packaging, so spend will skew to front-end, though both matter.For customers, the higher investment signals confidence in demand across businesses, especially where LTAs are in place, giving enough visibility to pre-build capacity. We remain highly disciplined and only invest when returns are compelling; new fabs are initial cash outflows (driving next year's CapEx higher) but yield attractive long-term returns, and with longer node lifecycles, ROIC improves materially—already visible on Intel 10 and Intel 7.

Q: A competitor pegs CPU TAM at $220bn by 2030 (~45% CAGR). Do you agree, and can Intel meet it?

A: We won't quote a specific number, but we agree the market is strong and set to grow significantly. We have repeatedly noted CPU-to-GPU ratios are rising and are near parity today, and shipments could skew even more toward CPUs over time.We are well-positioned to capture meaningful share. While exact sizing is hard, client spend levels, signed LTAs, and our visibility support a very robust growth outlook.

Q: Does owning fabs help regain server share this year? How do you view regaining share vs. AMD and ARM over the next five years?

A: Demand is very strong, and our near-term challenge is ramping supply to meet it. On server/DC, our roadmap is strong—Clearwater Forest, Diamond Rapids, and Coral Rapids with SMT—and we continue to lift single-thread and multi-thread performance, with MT gains realized on Coral Rapids, to compete better.On ARM, we are strong partners; beyond ARM-based CPUs, ARM can be a key customer and collaborator on ASIC/foundry, especially on IP. Overall, our competitiveness and roadmap are solid; we still lag in some areas but are catching up fast and aim to leapfrog on parts of the CPU architecture, supported by significant investments.

Q: Is there still a gross vs. net CapEx view? How do you allocate between internal and foundry, and can you disclose how much serves external customers?

A: There is a gross-to-net difference largely from AMIC (advanced manufacturing investment credit, i.e., ITC), currently in the low-single-digit billions ($1–4bn) and rising over time; it's mainly a timing effect. In the U.S., ~$0.35 is credited per $1 invested, and most CapEx is U.S.-based.But there is a lag: fabs must be built and production-ready before equipment credits are claimed, and then filed with the IRS, so refunds trail spend. Strategically, we plan by target wafer starts per node rather than internal vs. external, and invest front-end accordingly; we place POs based on demand drivers while staying flexible as info updates. In packaging, we already have a sizable backlog and must add capacity, partly in-house and partly via substrate vendors, which require prepayments, so that portion is moving faster.

Q: Is client strength all pricing-driven? With potential softness in H2 end demand, what is your client outlook?

A: Client outperformed mainly on ASP, partly mix and partly like-for-like price increases, passing through our own cost inflation. On a YoY basis the market is down: 2025 was strong on the Windows refresh, creating a high base, and memory cost/availability is also a headwind.We continue to skew mix up, supporting ASPs and results. Next quarter, client revenue should be roughly flat; CCPG overall may grow slightly on solid edge growth, with core client flat.

Under the hood, the market is soft and memory supply/demand dynamics are a drag, so the market should decline this quarter. But customers kept CPU inventories tight, so there may be some CPU restocking in Q3, which we will start to feel in Q4, and that is positive for us, as DC urgently needs CPUs and supply is far short of demand, so we will shift as much capacity as possible to data center CPUs to close the gap.

Q: What is the client inventory write-down, and how big is it? Ex that, does the GPM guide imply a QoQ decline?

A: Some products were not completed as 'match sets'; economically, we prefer to reallocate capacity rather than finish them, creating stranded inventory that we wrote down. The guide is roughly flat QoQ: the Q2 write-down does not repeat and is a tailwind, but Panther Lake and Granite costs, while improving QoQ, are a larger mix and below corporate average margins early in life, which is a headwind.As 18A yields improve, both dynamics should turn into tailwinds, and Panther Lake margins should surpass the corporate average, lifting overall GPM. Our top 2026 goal is to keep GPM above 40% each quarter; we did so in H1 and guided Q3 similarly, and will seek further improvement thereafter.

Q: With substantial capacity coming online by quarter-end, does that imply a big Q4 step-up? If Sept supply shortfalls resemble last quarter (> $1bn), Q4 looks strong—fair?

A: We only guide one quarter out. That said, if inventory/supply improves by late Q3/early Q4, it will help, though we still won't fully meet demand and Q4 remains supply-constrained.Internal wafers are also tight. Our supply is an internal wafers plus advanced packaging equation (substrates, glass, memory), and these procurements are bottlenecks—some among the tightest in the chain; front-end improves more linearly while back-end is 'chunky'. With some blockages easing by late Q3, this quarter trends flattish, and Q4 should have upside.

Q: Drop-through in Mar and Jun looked solid; per guide it dips to just above 50%, still in the 40–60% range. Is the framework intact, and what are the puts/takes into next year?

A: Over time, we still expect drop-through in the 40–60% range. Quarterly dynamics will place us at the low end, high end, or midpoint, but the range remains a good rule of thumb.

Q: When will external foundry confidence translate into customer announcements? How much higher is next year's CapEx, and how is it split between external customers and internal capacity?

A: On confidence: 18AP is in risk production and will be ready by year-end, ~5% faster than 18A; 18A yields/volume are strengthening, and Panther Lake is ramping and scaling. More critically, 14A: PDK 0.5 is done and 0.9 is due in Oct, with 256 SRAM yield/defects/perf ahead of rigorous plans; internal 14A products risk in H2 2027 and volume in 2028.Customer feedback is very positive—once they see PDK 0.9 and yields, discussions shift to which products to run and how much capacity we can provide. Our principle is to invest only after yield proof-points, IP readiness, and strong customer engagement. For structure: by 2026 we already have ample 'space' from prior fab builds and need limited fit-out; thus most spend is on tooling, up ~40% vs. 2025, focused on Intel 3, 18E, and 18AP.We are not giving 2027 numbers yet as we finalize scale and, per practice, disclose early in-year; directionally 2027 will be higher. Funding supports both internal and external businesses; we size wafer capacity holistically by customer demand and build accordingly.

Q: How do you balance customer demand with FCF targets? What is a reasonable CapEx framework—do you only build against signed demand?

A: We are prudent. Even with higher CapEx, base biz CFO is solid when factoring AMIC; however, we likely need back-end investments, especially with third parties, which could pressure cash flow next year and make positive FCF more challenging.Still, these investments carry attractive ROIs; as long as we have confidence in growth, pricing, and cost structures, and know these nodes have long lifecycles with strong ROIC, we will invest. A key change under new leadership is avoiding large bets before firm customer commitments; conversely, our confidence for next year reflects strong customer certainty, or we would not be placing POs now.

Q: ASIC run-rate is ~$1.2bn and growing well. How do you view diversification (e.g., Fortinet), growth prospects, and margin structure at scale?

A: This is a very large opportunity with TAM >$100bn. Our differentiation spans advanced design (CPU, XPU), a deep IP portfolio, and unique advanced packaging for radical integration—many new AI approaches need that packaging plus advanced silicon, creating demand for purpose-built chips across many customers.For example, we announced a next-gen security ASIC with Fortinet, and Intel IPUs to select hyperscalers—both major opportunities. The business is up ~3x YoY; run-rate revenue is near $2.0bn and we expect to reach ~$4.0bn run-rate in the not-too-distant future, and with a $100bn TAM our IP and capabilities should secure a meaningful share.

Q: Given changes in memory hierarchy/architecture (e.g., Z-Angle Memory, cross-batch memory), will Intel play a bigger role in memory? Are there internal R&D and long-term opportunities?

A: First, memory is a major supply constraint, and we are working closely with the three leading memory vendors—this is priority #1 for serving customers. Second, Intel has deep heritage in memory, and we recently hired the former SK Hynix CEO; memory is a bottleneck for AI infra and a pain point for customers.We are researching compute-memory integration, stacking, and improving memory efficiency. Many efforts are underway, and we will update as they progress.

<End of document>

Risk Disclosure & Statement:Dolphin Research Disclaimer & General Disclosure