Joe Quatrochi

Updated the 2026 mix to low to mid-60s and flagged memory rising in the December quarter.

Gave specific lead-time ranges and noted customer conversations have moved to 2H27 deliveries on larger products.

CFO acknowledged first-half supply chain constraints from faster-than-expected ramp slope but said KLA is much better positioned for second-half and 2027, with tremendous focus on ensuring capacity.

CEO confirmed yield-driven demand drives both process control sales and services, with Intel publicly increasing metrology usage and the whole industry squeezing yield from existing capacity.

Bren identified optical components as the primary constraint rather than DRAM, with decisions made last summer affecting first-half shipments. Rick noted supply constraints are industry-wide, not KLA-specific, so no share loss risk.

Bren confirmed March as the likely margin trough, with improvement driven by mix, tariff burden diminishing through process optimization, and memory DRAM cost situation expected to be transitory. Pricing is value-oriented rather than cost-based.

Bren corrected the high-teens assumption, noting KLA's packaging share went from 1% to approaching 6% of the $11B market, with continued growth expected as density and process complexity increase.

Bren cited China merchant mask investment, requalification from high design mix, single-die reticle inspection growth, and broadband plasma Gen 5 print check adoption.

Rick explained the traditional ramp-then-run-off model is broken because multiple designs create compounding ramps, with some customers even back-porting more process control to prior nodes.

Bren attributed margin improvement to favorable product mix from new platforms resetting cost of ownership, plus scale benefits in advanced packaging offsetting prior inflation pressures.

Rick sees no clean room gating factors, with silicon at the leading edge being the governing factor and steady capacity growth expected over the next several years.

Rick confirmed the services impact from China export controls is front-loaded in March quarter and then services grow from that new base.

Rick explained HBM drives process control intensity up 100-150 basis points from the historical 9-10% range due to bigger die, less redundancy, and more complex logic circuitry.

Wallace detailed rising process-control intensity at 2nm -- more inspection points and higher sampling as process windows tighten.

Reaffirmed the flat first half and second half ramp, added the mid teens year over year end market view, and stressed the breadth of the socket base.

Confirmed sampling with some revenue already, and Tony Balow set expectations directly by ruling it out as a 2026 model driver.

Michael described varied content across battery management, GPU power, sensors, actuators. Dollar content varies by application (chips vs modules). Declined to confirm specific content figure.

Tony noted 4Q flat at 55.5% (low end), Q2 incrementally raised on better backlog visibility. H2 headwinds warrant caution. Michael noted module yield improvements are ongoing.

Tony noted PC business expected down from strong 2025 due to selective margin participation. Memory impact too early to assess.

Tony noted both categories trending well but lines are blurry. Michael added they use the same power supply now, making separation difficult.

Bernie confirmed no fundamental change in positioning since Q2 call, suggesting previous enterprise data expectations still hold.

Tony confirmed 2026 revenue ramp but not individually a needle mover. Called it the start of a wave of full solution design wins. Michael noted AI-assisted robotics is the catalyst.

Michael highlighted server refresh cycles driving consistent growth. MPS expects to gain an even higher percentage of the server market next year.

Michael noted growth was evenly spread across DDR5, SSD, HDD, notebook, and desktop segments. Tony indicated no sharp Q2 falloff expected.

Bernie attributed notebook strength to AI-related design wins but acknowledged the Q1 ramp was atypical and lacked definitive explanation.

Tony declined to discuss architectural specifics, citing customer confidentiality. Michael noted customers explicitly don't want lateral vs vertical disclosed.

Jason reiterated high 30% to low 40s percent for the first generation and said 50% is expected over the next couple of years as work is brought in-house.

Rene answered directly, citing improvement across wafers, substrates, test capacity and memory, with demand now north of $2 billion.

Jason acknowledged mobile unit growth turned negative but expects data center demand to more than offset weakness. Automotive continues double-digit growth. Overall royalty outlook positive.

Jason quantified that a 20% unit volume decline would impact smartphone royalties by only 2-4% because constraints hit low-end v8 devices while premium CSS/v9 is protected. Total royalty impact would be 1-2%, more than offset by cloud AI growth.

Rene explained that agentic AI workloads are persistent, always-on, and power-constrained, making them ideal for CPUs. Hyperscalers are responding with higher core counts. Combined with power efficiency requirements, this creates a strong tailwind for Arm.

Rene explained DreamBig's Ethernet controller IP is critical for high-speed data center networking, broadening Arm's offering to end customers in scale-up and scale-out environments.

Rene explained the jump from ~18% to ~50% was driven by Grace Blackwell replacing x86 in AI data centers plus continued general-purpose compute gains from Graviton, Axion, and Cobalt.

Jason explained partners are providing less visibility than normal due to tariff uncertainty, making any full-year range too wide to be helpful.

Jason confirmed strong CSS momentum and first material CSS revenue quarter but deferred FY26 quantification to next quarter's guidance.

Christophe said opportunities are easier to enable in 2028 than near term but declined to give new 2027 or 2028 output numbers.

Roger said price increases are agreed with some 2027 effect and full impact from 2028, citing volume, mix and pricing as tailwinds and ramp costs and upgrades as swing factors.

Roger said 85 tools reflects today's demand-supply balance and ASML would revisit and roll up its sleeves if customers request meaningfully more.

Roger confirmed Q4 gross margin step-down reflects more High NA rev rec, lower upgrade contribution, and the absence of 1H one-off cost benefits.

Christophe pointed to ongoing tariff uncertainty (incl. the H20 flip-flop), the 232 review, and customers waiting as long as possible before committing, leading ASML to become more cautious on 2026.

Roger said the remaining 3600 prebuilds will work down through 2025 and there are no concerns about the salability of the inventory.

Christophe said tariff announcements have not shifted customer business conversations and Roger added that customer clean-room space, not inventory, is the real gating item, so opportunistic pull-ins are not even feasible for most customers.

Roger treated the Q4 High NA cost favorability as a one-off, already embedded in the 51–53% 2025 gross-margin range, and reiterated 2H margin is expected to step down as High NA revenue recognition skews to the back half.

Christophe said ASML is provisioning for upside in its supply plan and will decide whether to actually build the incremental tools as the AI capacity opportunity firms up through the year.

Devgan described the three-layer framework and 4 super agents but gave no TAM number; Wall added guidance assumes no step function.

Devgan said there is real potential for 11% share to increase with agentic AI. Big customer CEOs want to invest in more automation and compute.

Anirudh said every year is a record for hardware, expects trend to continue. Big customers buying almost annually. Current Z3 supports 1T transistor designs. Taking share across hardware, IP, EDA, 3D-IC, and first to market in agentic AI.

John Wall attributed Q3 benefit to small restructure and healthy hardware gross margins. Q4 slightly higher due to new acquisition expenses. Overall year slightly ahead of expectations.

John Wall said $50M revenue raise and $0.12 EPS raise driven by strength across all regions, not just China. Apples-to-apples vs prior guide since restrictions rescinded.

Anirudh prioritized verification (SimAI 5-20x improvement), digital implementation (Cerebrus), package/PCB design (Allegro X AI), and analog migration.

John Wall confirmed no material RPO impact from China restrictions. Acknowledged hardware weakness in China drove 2024 decline. Anirudh noted strong design activity in Chinese EV companies.

Hassane acknowledged prioritizing AI data center pulled from auto and industrial short term, but said there was no real customer impact and supply catches up in H2.

El-Khoury said increased content from more products mapping to rack opportunity. SiC JFET, vGaN, SPS, point-of-load all expanding addressable content. Investment over the last year capitalizing on AI opportunity.

Trent said FCF margin target is 25-30%, delivered 24% in 2025. Expects to expand FCF margin. Plan to return 100% of FCF to shareholders via $6B repurchase authorization.

Trent said lead times extended to ~20 weeks from mid-teens. No significant change to short lead time orders. Order patterns continuing to improve giving confidence in stabilization.

El-Khoury declined to provide SiC guidance but said it's tracking expectations. Gaining share with NIO new brand and broader China EV deployment through a leading Tier 1.

El-Khoury confirmed AI data center will continue growing in 2025 with designed-in platforms ramping. Treo will accelerate portfolio introductions.

El-Khoury said non-core exits are product-driven, not region-driven. No local pricing practices by region.