CJ Muse

CJ Muse inquired about NVIDIA's future roadmap concerning customized silicon, including CPX and Groq, and the increasing focus on workload or customer-specific solutions, particularly in the context of dielet architecture.

Answer

Jensen Huang, President and CEO of NVIDIA, explained that while dielets are used when necessary, NVIDIA prioritizes monolithic dies to reduce latency and power. He stated that NVIDIA's versatile CUDA architecture and architectural compatibility across GPUs allow for continuous software optimization, benefiting the entire install base.

CJ Muse inquired about NVIDIA's future roadmap concerning customized silicon for large context windows and decode-specific solutions, such as those from Groq, and whether customized silicon or a dielet architecture would be an increasing focus.

Answer

CEO Jensen Huang stated that NVIDIA avoids dielets when possible due to latency and power costs, preferring integrated designs like the two giant reticle-limited dies in Grace Blackwell and Rubin. He highlighted CUDA's architectural effectiveness and compatibility across generations, and mentioned plans to extend NVIDIA's architecture with Groq's innovations, similar to the Mellanox acquisition.

CJ Muse questioned whether NVIDIA sees a realistic path for supply to catch up with the surging demand for AI infrastructure over the next 12-18 months, given the current sold-out status and upcoming product generations.

Answer

Jensen Huang, President and CEO, NVIDIA, explained that NVIDIA's extensive supply chain, including TSMC and memory partners, has been well-planned for significant growth. He highlighted three ongoing transitions—Accelerated Computing, Generative AI replacing Classical Machine Learning, and Agentic AI—all driving exponential demand for NVIDIA GPUs, making it challenging for supply to fully catch up.

CJ Muse questioned the sustainability of AI infrastructure buildouts and asked if NVIDIA anticipates supply catching up with demand over the next 12-18 months, given current sell-out conditions and upcoming product launches.

Answer

Jensen Huang, President and CEO, NVIDIA, explained that NVIDIA's extensive supply chain planning with global partners has been robust. He attributed sustained demand to three simultaneous transitions: accelerated computing, generative AI transforming hyperscaler workloads, and the rise of agentic AI, indicating that demand continues to outpace supply.

CJ Muse of Cantor Fitzgerald inquired about NVIDIA's high-level growth vision into 2026, considering the long lead times for products like Rubin and the multi-year nature of data center projects. He also asked for commentary on the growth dynamics between the networking and data center segments.

Answer

Founder, President, CEO & Director Jensen Huang explained that the primary growth driver is the shift to reasoning and agentic AI, which requires orders of magnitude more computation. He highlighted that the Blackwell NVLink 72 rack-scale system was built for this moment. Huang projected a $3 to $4 trillion AI infrastructure opportunity by the end of the decade, noting that CapEx from just the top four cloud service providers has already doubled to $600 billion in two years, signaling the beginning of a massive build-out.

CJ Muse of Cantor Fitzgerald asked about the scale of unannounced large GPU cluster investments, similar to recent sovereign AI deals, and how these significant orders are affecting lead times and demand visibility for the Blackwell platform.

Answer

President and CEO Jensen Huang confirmed that NVIDIA has more orders now than at GTC and that numerous AI factory projects remain unannounced. He stated that the supply chain is scaling to meet this demand, including new onshore facilities. Huang emphasized that AI is becoming essential infrastructure for every nation, which provides strong, long-term visibility for demand.

CJ Muse asked about Applied Materials' 2026 WFE outlook, comparing it to peers' guidance, and inquired about the drivers of share growth for the company's semiconductor equipment business. He also followed up on gross margin expectations for the current sub-cycle, considering rising volumes and value-based pricing.

Answer

CEO Gary Dickerson stated that Applied Materials expects its semiconductor equipment business to grow more than 20% in calendar 2026, weighted towards the second half, with strong momentum continuing into 2027, driven by AI investments in leading-edge foundry logic, DRAM (including HBM), and advanced packaging, where Applied Materials holds strong leadership positions. He also highlighted the company's inflection-focused innovation strategy and the EPIC platform for value creation and margin growth. CFO Brice Hill added that the Semiconductor Systems business gross margin is approximately 54% and is expected to improve as the portfolio shifts to more valuable products.

CJ Muse asked about Applied Materials' 2026 semiconductor equipment business growth exceeding 20% compared to broader WFE forecasts, inquiring about the drivers of market share growth. He followed up on the company's gross margin trajectory, considering rising volumes, value-based pricing, and improved customer margins.

Answer

CEO Gary Dickerson explained that AI is fueling growth in leading-edge foundry logic, DRAM (including HBM), and advanced packaging, where Applied Materials holds strong leadership positions and expects to gain share. CFO Brice Hill added that the equipment business gross margin is approximately 54% and is expected to improve as the portfolio shifts to more valuable products.

CJ Muse sought more detail on the Q4 weakness, asking why export licenses are now a challenge for China, the specific cause of the foundry visibility issue, and what trends are being seen in HBM.

Answer

CFO Brice Hill clarified that the Q4 China forecast is lower due to customer delivery schedules and digestion, not new license issues. He noted the company is conservatively assuming no revenue from its significant backlog of pending license applications. The leading-edge logic weakness is due to a less linear ramp than modeled, not a specific issue like clean room availability.

C.J. Muse inquired about gross margins, asking for an update on the ~48.2% floor, the impact of tariffs, and the long-term trajectory considering value-based pricing, cost reductions, and manufacturing diversification.

Answer

CFO Brice Hill confirmed that a low-48s percentage (48.2-48.3%) is the correct current operating level for gross margin, with only a modest impact from tariffs due to a flexible global supply chain. CEO Gary Dickerson emphasized that the company is focused on driving sustainable, long-term margin improvements through both cost reductions and better value capture for its innovations.

CJ Muse asked for more details on GlobalFoundries' silicon photonics business, specifically regarding the expected doubling of revenues in 2026, any changes in customer mix, and the overall growth trajectory for the Communications Infrastructure & Data Center (CID) segment. He also inquired about the incremental revenues and implications for gross margins and OpEx from all 2025 acquisitions.

Answer

Tim Breen (CEO, GlobalFoundries) explained that AMF brought new customers, broadening the silicon photonics portfolio, with most revenue from pluggable applications also driving demand for high-performance silicon germanium. He confirmed co-packaged optics (CPO) is on track for a 2027 scale ramp. Sam Franklin (CFO, GlobalFoundries) added that CI&D grew 30% in 2025, driven by silicon photonics, optical networking, and satellite communications, expecting similar growth in 2026. Mr. Franklin projected MIPS to deliver $60M-$100M and AMF at least $75M in 2026, both margin accretive (roughly a point of incremental margin in 2026). Mr. Breen noted MIPS and Synopsys are targeted to become a multi-year, incremental billion-dollar business.

CJ Muse asked for details on the expected doubling of silicon photonics revenue in 2026, including changes in customer mix, and the overall growth trajectory for the Communications Infrastructure & Data Center (CID) business. He then followed up on the incremental revenues, gross margin implications, and OpEx framework for all 2025 acquisitions.

Answer

CEO Tim Breen confirmed new customers from AMF broadened the silicon photonics portfolio, with most revenue from pluggable applications and co-packaged optics on track for a 2027 scale ramp. CFO Sam Franklin noted CI&D grew 30% in 2025, driven by silicon photonics, optical networking, and satellite communications, with similar growth expected in 2026. For acquisitions, Mr. Franklin stated MIPS is expected to deliver $60M-$100M and AMF at least $75M in 2026, both margin accretive by roughly one point. Mr. Breen added that both photonics and custom design/IP (MIPS/Synopsys) are targeted to become multi-year billion-dollar businesses.

CJ Muse inquired about the drivers behind the expected 20% growth in non-wafer revenues for 2025 and whether similar growth is anticipated for calendar 2026. He also asked about the Smart Mobile Devices business outlook for calendar 2026, considering the pricing reset and hopeful unit volume gains, and any potential headwinds.

Answer

CEO Tim Breen explained that non-wafer revenue growth is driven by a higher number of design wins leading to more tape-outs, and the MIPS acquisition contributing IP revenue, expecting this trend to continue. For Smart Mobile Devices, Tim Breen clarified that the pricing reset for dual-source business is complete, with no further step-down expected, and anticipates increased volumes. He highlighted differentiated solutions in RF front-end, CBIC technology, and new form factors like smart glasses as mixed tailwinds, along with the long-term benefit from onshoring strategies of handset players.

CJ Muse from Cantor Fitzgerald asked for more color on the drivers behind the projected 20% growth in non-wafer revenues for 2025 and whether similar growth should be expected into calendar 2026. He also inquired about the smart mobile business's potential to grow in calendar 2026, considering the pricing reset and hopeful gains in unit volumes, or if headwinds persist.

Answer

CEO Tim Breen clarified that non-wafer revenue includes reticles, non-recurring engineering, technology services, and licensing, with MIPS being a new driver. He attributed growth to a higher number of design wins leading to more tape-outs and the MIPS acquisition. Regarding smart mobile, Tim Breen stated that the pricing reset for dual-source business is complete and not expected to recur, with expectations for increased volumes. He highlighted differentiation in RF front-end (CIPIC silicon germanium technology), new form factors like smart glasses, and the long-term tailwind from onshoring customers diversifying supply chains as growth drivers, making the market ripe for growth in differentiated areas.

CJ Muse of Cantor Fitzgerald asked for a quantification of the expected gross margin uplift in Q4, considering strong non-wafer revenue and ASP headwinds. He also inquired about the outlook for 2026 gross margins and the growth trajectory for the smart mobile segment.

Answer

CFO John Hollister anticipated significant Q3-to-Q4 gross margin improvement driven by a stronger product mix, non-wafer revenue, depreciation roll-off, and better utilization. CEO Tim Breen expressed a bullish long-term outlook for the smart mobile market, citing refresh cycles, new form factors like smart glasses, and GFS's opportunity to gain share in areas like haptics, display, and connectivity.

CJ Muse asked about the evolution of customer engagement and contracts within the current tight market environment. He also followed up on the company's capital allocation plans after the SanDisk share monetization.

Answer

Irving Tan, CEO, detailed Western Digital's customer-centric approach, including dedicated teams for hyperscale customers, which has deepened relationships and improved demand visibility, leading to longer-term LTAs. He noted a structural change in pricing reflecting the value delivered and a focus on predictable pricing for sustainable value creation. Kris Sennesael, CFO, confirmed the ongoing focus on share repurchases, having already utilized $1.3 billion of the $2 billion authorization.

CJ Muse questioned how Western Digital's customer engagement and contractual agreements are evolving in the current tight market environment. He also followed up on the company's capital allocation plans after the SanDisk share monetization.

Answer

CEO Irving Tan highlighted a customer-centric approach with dedicated teams for hyperscale customers, leading to deeper relationships, better demand visibility, and longer-term LTAs. He emphasized a focus on fair value exchange, predictable pricing, and sustainable value creation. CFO Kris Sennesael confirmed that Western Digital is actively pursuing its $2 billion share repurchase authorization, having already repurchased $1.3 billion (approximately 13 million shares), and will continue to utilize this program.

CJ Muse sought to quantify the incremental NAND demand driven by AI infrastructure build-out, excluding KV Cache, and asked for an updated demand growth CAGR for 2026-2028. He also questioned SanDisk's capital return strategy, specifically regarding share repurchases, given the significant debt paydown.

Answer

CEO David Goeckeler indicated that the best proxy for AI-driven demand is the accelerating exabyte growth in the data center, with forecasts for 2026 increasing from mid-20s to mid-40s, and now to high 60s, primarily driven by AI. CFO Luis Visoso affirmed that debt reduction is a priority, with $2 billion reduced to $600 million. He stated that the primary focus remains on investing in the business and building prudent cash reserves, with updates on capital return plans to be provided at the appropriate time.

CJ Muse asked for a quantification of incremental NAND demand from AI infrastructure build-out (excluding KV Cache) and the projected demand growth CAGR for 2026-2028. He also questioned SanDisk's capital return plans, particularly share repurchases, given the significant debt paydown.

Answer

David Goeckeler (Chairman and CEO) indicated that the best proxy for AI-driven demand is the data center exabyte growth, which has significantly increased to high 60s for 2026. Luis Visoso (EVP and CFO) stated that debt reduction, investing in the business, and building prudent cash reserves are current priorities, with no immediate changes to the capital allocation strategy or plans for share repurchases.

CJ Muse of Cantor Fitzgerald inquired about the drivers behind the Q1 gross margin guidance, seeking details on underutilization, cost reduction trends, and product mix.

Answer

EVP & CFO Luis Visoso explained that while underutilization charges are decreasing significantly to $10-15 million, the primary headwind is approximately $60 million in fab startup costs for the BIX-eight node. He noted these startup costs represent a temporary 300 basis point impact and are expected to decline significantly in Q2 and Q3, becoming a tailwind to margins.

C.J. Muse asked about Seagate's gross margin trajectory, given strong supply-demand dynamics, and whether average pricing per exabyte could move flat or positive year-over-year, considering the historical philosophy of sharing gains.

Answer

CEO Dave Mosley explained that pricing is dictated by demand, and with current strong demand, flat to slightly up pricing is possible, driven by the strong value proposition of new drives. CFO Gianluca Romano added that Seagate is executing better than the Investor Day model, aiming to optimize product, sales, and profitability, and expects continued progress in the right direction for gross margins.

CJ Muse asked about Seagate's gross margin trajectory, given strong supply-demand dynamics and the company's philosophy of sharing gains while also raising like-for-like pricing. He inquired about a framework for incremental gross margins and whether average pricing per exabyte could move flat or positive year-over-year.

Answer

CEO Dave Mosley stated that pricing would be dictated by strong demand, making flat to slightly up pricing certainly possible, especially as new drives offer strong value propositions. CFO Gianluca Romano added that Seagate is executing better than its Investor Day model for incremental margins above $2.6 billion revenue and aims to continue optimizing profitability over the next two to three years.

C.J. Muse asked if Seagate could pivot supply from consumer/VIA to cloud to address tightness, if SSDs are cannibalizing HDD demand, and what would prompt Seagate to add capacity given the vision for sustainably higher demand.

Answer

Dave Mosley, Chair and CEO, stated that customers are not changing architectures due to current tightness, and there's no evidence of cannibalization by SSDs. He emphasized that customers are seeking predictability and aggressive product transitions. Gianluca Romano, CFO, added that the gap between supply and demand is widening, not being filled by other technologies. Dave Mosley also noted that some Edge IoT supply is being shifted to cloud, improving Edge IoT profitability.

CJ Muse sought clarification on March seasonality, the ability to pivot supply from consumer/VIA to cloud, concerns about SSD cannibalization due to HDD tightness, and the potential for adding capacity to support sustained demand.

Answer

Seagate's Chair and CEO, Dave Mosley, stated that customers are not changing architectures due to current conditions, and there's no cannibalization. He also noted that some supply is being pivoted from Edge IoT to cloud, muting seasonality. CFO Gianluca Romano added that the gap between supply and demand is widening.

CJ Muse of Cantor Fitzgerald sought to understand the drivers of short-term revenue growth, given the September guidance, and asked when the HAMR ramp would enable the company to support its long-term target of low-to-mid-teens growth.

Answer

CEO Dave Mosley explained that near-term supply is dictated by long-term planning and a focus on technology transitions. CFO Gianluca Romano pointed out that the September guidance already reflects ~15% year-over-year growth, in line with their model, and that future growth will come from increasing exabytes via technology rather than unit volume.

CJ Muse from Cantor Fitzgerald questioned Intel's CapEx strategy for 14A, asking why the company isn't more aggressive in placing equipment orders now, given bullish AI-led demand and potential longer lead times if orders are delayed until late 2026. He also asked about the technical feasibility of High-NA for future HBM and its adoption timeline (14A or 10A).

Answer

CFO David Zinsner explained that Intel is aggressively acquiring tools for Intel 7, 3, and 18A, but holding back on 14A capacity until foundry customers are secured, maintaining financial discipline. He emphasized that short-term supply improvements are being driven by better yields and cycle times with existing tools, which don't require CapEx. Zinsner confirmed that High-NA is targeted for the 14A process, with different variants.

CJ Muse from Cantor Fitzgerald expressed concern that waiting until late 2026 to place 14A equipment orders might lead to longer lead times, questioning why Intel isn't more aggressive given competitors' actions. He also asked about the technical feasibility demonstration of high-end A for future HBM, clarifying if it's targeted for 14A or 10A adoption.

Answer

David Zinsner, CFO, clarified that Intel is aggressively acquiring tools for Intel 7 and 18A, increasing wafer starts on those nodes. However, 14A CapEx is being held back until firm foundry customer commitments are secured. He also emphasized that improving yields and cycle times with existing tools is a key, CapEx-free way to increase supply. David Zinsner confirmed that high-end A for HBM will be part of the 14A process.

CJ Muse asked about the expected growth of compute for Broadcom generation to generation in custom silicon and whether the company sees a broadening opportunity for its existing five customers to have multiple XPU offerings, referencing a competitor's XPU offering for massive context windows.

Answer

Hock Tan (President and CEO, Broadcom) confirmed that custom accelerators allow for multiple XPU offerings for each customer, tailored for different aspects of training and inference (e.g., Sparse Core, dense matrix multipliers, memory capacity/bandwidth variations). He emphasized the fascinating and varied nature of creating specialized hardware for different workloads, leading to multiple chips per customer.

CJ Muse sought clarification on the Q4 gross margin guidance, specifically how a 70 basis point sequential drop is achieved given lower software revenue sequentially and higher contributions from lower-margin wireless and XPU products.

Answer

Kirsten Spears, Chief Financial Officer, explained that the Q4 gross margin will be impacted by the higher mix of XPUs and wireless revenue, which generally have lower margins. She also noted that software revenue will be 'coming up just a bit' in Q4.

CJ Muse from Cantor Fitzgerald followed up on the inference opportunity, asking about the optimal workloads for custom silicon and the potential long-term mix between inference and training for the XPU business.

Answer

CEO Hock Tan explained that the value of custom accelerators applies to both training and inference. He stressed that the fundamental advantage of custom silicon is the ability for a hyperscaler to co-optimize their proprietary large language models with the hardware over a multi-year roadmap. This deep integration, he argued, unlocks performance levels unattainable with merchant silicon.

CJ Muse of Cantor Fitzgerald followed up on the inference opportunity, asking about the specific workloads best suited for custom silicon and what the long-term business mix between inference and training might look like for XPUs.

Answer

CEO Hock Tan explained that the fundamental value of custom silicon applies to both training and inference. He emphasized that the key advantage is not just cost, but the ability for a hyperscaler to co-optimize its software algorithms with the silicon over a multi-year roadmap. This tight hardware-software integration, he argued, unlocks performance levels unattainable with merchant silicon.

Asked how Broadcom manages to work with multiple competing hyperscalers on their frontier models, helping them optimize while also maintaining confidentiality and allowing for differentiation.

Answer

Hock Tan explained that Broadcom provides fundamental semiconductor technology and helps partners optimize it based on their specific direction and models. The optimization involves balancing key variables like compute, power, and memory. Broadcom's visibility is limited to what the partner shares, ensuring each partner's proprietary work remains separate.

CJ Muse asked for a framework regarding AI drivers in industrial and data center segments, seeking specific numbers or percentages for faster-growing parts of the business like semi-test analog, optical, power, and energy.

Answer

CEO Vincent Roche highlighted that the data center business grew 50% in FY25 to a $1 billion run rate, and the ATE business grew 40% to an $800 million run rate, both expected to see double-digit growth. He detailed opportunities in electro-optical interfaces (800 gig, 1.6 terabit), sophisticated power management, and vertical power technology. CFO Richard Puccio concurred, noting increasing CapEx forecasts from hyperscalers.

CJ Muse requested an update on how to model the non-Android handset business for calendar 2026.

Answer

CEO Cristiano Amon stated there was no change to what had been previously communicated regarding share within Apple.

CJ Muse asked for an update on how to model Qualcomm's non-Android handset business for calendar 2026.

Answer

CEO Cristiano Amon stated there was no change to what had been previously communicated regarding share within Apple.

C.J. Muse asked about the drivers of the IoT business's strong growth in contrast to weakness from peers, and also inquired about the potential margin impact if a large customer were to transition to internal solutions, as noted in the 10-K.

Answer

CFO Akash Palkhiwala attributed IoT strength to new product launches in XR and PC, and an inflection in the industrial market towards AI-ready solutions. Regarding the large customer risk, he reiterated the existing three-year agreement for 2024-2026 phone launches, with a planning assumption of share ramping down to 20% for the 2026 launch, noting any better outcome would be upside.

CJ Muse asked for a framework regarding gross margins throughout calendar 2026, considering the MI355 to MI400 transition and the move to full rack scale. He also probed AMD's growth expectations through 2026 and beyond, specifically how the OpenAI partnership and other large customers contribute to the projected tens of billions in revenue by 2027, and the breadth of customer penetration.

Answer

Jean Hu, EVP, CFO, and Treasurer of AMD, stated that data center GPU gross margins typically improve with new product ramps, normalizing after a transition. While not guiding 2026, the priority is top-line revenue growth and gross margin dollars, aiming to increase the gross margin percentage. Lisa Su, Chair and CEO of AMD, emphasized the significance of the OpenAI relationship and its multi-gigawatt scale. She noted numerous other deep customer engagements, including OCI and Department of Energy systems, indicating expectations for multiple customers to achieve significant scale with the MI450 generation, supported by a dimensioned supply chain.

CJ Muse asked for an outlook on the Instinct accelerator ramp into 2026, including the MI350-to-MI400 transition and customer mix, and also inquired about the planned use of proceeds from the ZT Systems divestiture.

Answer

Chair and CEO Lisa Su outlined that the MI355 will ramp through H1 2026, with the MI400 and Helios platform expected to be significant revenue contributors in 2026, initially driven by hyperscalers. EVP, CFO & Treasurer Jean Hu reiterated that proceeds from the ZT sale will follow their capital allocation policy: reinvesting in the business, particularly AI, and returning cash to shareholders via buybacks.

CJ Muse from Cantor Fitzgerald requested an overview of the Instinct accelerator ramp into 2026, asking about the MI350-to-MI400 transition timing and the expected customer mix between hyperscalers and sovereign AI. He also asked about the planned use of proceeds from the $3 billion sale of the ZT Systems manufacturing business.

Answer

CEO Lisa Su outlined that the focus through the first half of 2026 is the MI355 ramp, with the MI400-based Helios platform expected to be a significant revenue contributor in 2026. She anticipates hyperscalers will drive the initial ramp, with sovereign opportunities following. CFO Jean Hu reiterated AMD's capital allocation strategy for the ZT proceeds: reinvesting in the business, particularly AI, and continuing to return cash to shareholders via buybacks.

CJ Muse focused on gross margins, asking about the 50 basis point guided decline and whether it was primarily due to product mix. He also inquired if KLA expects to be at the higher end of its 40%-50% incremental operating margin target if WFE growth continues in double digits over the next couple of years, given greater contributions from higher-margin silicon.

Answer

CFO Bren Higgins confirmed the gross margin decline was mostly due to mix adjustments, noting a consistent tariff impact of 50-100 basis points. Regarding operating margins, he stated that KLA aims for the middle of the 40%-50% target range for trendline growth, and could outperform if growth levels exceed the high single-digit trendline.

CJ Muse asked about the 50 basis point decline in gross margins, assuming it's due to product mix, and sought clarification on KLA's 40-50% incremental operating margin target, specifically if the company expects to be at the higher end of that range given greater contributions from higher-margin silicon.

Answer

CFO Bren Higgins confirmed the gross margin decline is primarily due to mix adjustments, noting a consistent tariff impact of 50-100 basis points. He explained that the 40-50% incremental operating margin target is a long-standing goal, with performance typically in the middle of the range for trendline growth, and outperformance for above-trendline growth.

CJ Muse of Cantor Fitzgerald inquired about the drivers behind management's constructive view on growth in 2026 and asked for more detail on the outlook for domestic China WFE.

Answer

EVP & CFO Bren Higgins explained that while it's early to quantify, constructive 2026 discussions are driven by high-performance compute, strong DRAM demand fueled by HBM, and a bottoming in legacy nodes. Regarding China, Higgins noted that after elevated investment in 2023-2024, they expect headwinds in 2025 and likely 2026 as spending normalizes. President & CEO Rick Wallace added that sustained investment in advanced logic and packaging further supports this positive momentum.

CJ Muse of Cantor Fitzgerald inquired about the drivers behind management's constructive outlook for 2026, asking what factors are needed for firm conviction, and followed up on the domestic China investment outlook for 2026.

Answer

EVP & CFO Bren Higgins stated it's early to quantify 2026 but noted encouraging discussions driven by high-performance compute, HBM-fueled DRAM, and a bottoming in legacy markets, while expecting headwinds in China. President & CEO Rick Wallace added that sustained investment in advanced logic provides momentum. Regarding China, Wallace commented it's too early to tell the magnitude of the decline, but their focus is shifting to stronger demand from other customers.

CJ Muse asked about the drivers behind the approximately $150 million upside in Teradyne's December quarter guidance compared to consensus, specifically inquiring about the contributions from HBM, VIP networking, and System Level Test (SLT). He also asked about the longer-term outlook for high-performance compute, including test intensity, increased test insertions, and test time, particularly with NVIDIA as a lead customer.

Answer

CEO Greg Smith attributed the Q4 upside primarily to compute and memory, with compute accounting for roughly two-thirds and memory (strongly represented by HBM) for one-third. For the longer term, Mr. Smith expressed bullishness on compute due to increasing die sizes, performance requirements, chiplet-based designs driving upstream testing, low tolerance for latent defects in data centers, and the growing trend of dual sourcing in the supply chain.

CJ Muse inquired about the drivers behind the Q4 2025 revenue upside compared to prior expectations, specifically asking about the contributions from HBM, VIP networking, and System Level Test (SLT). He also asked a longer-term question regarding high-performance compute, focusing on test intensity, increased test insertions, and test time in a leading-edge environment.

Answer

CEO Greg Smith explained that the Q4 upside was primarily driven by compute and memory, with compute accounting for roughly two-thirds and memory (especially HBM) for one-third. For the longer term, Mr. Smith expressed bullishness on compute test intensity due to larger die sizes, performance demands, chiplet-based designs, and the low tolerance for latent defects in data centers. He also noted that dual sourcing strategies in this segment are beneficial for Teradyne.

CJ Muse of Cantor Fitzgerald asked about the drivers for Teradyne's more positive outlook, questioning if it's due to a current business pickup or new design wins, and sought a framework for 2026 AI contributions.

Answer

President & CEO Greg Smith confirmed the confidence stems from an uptick in AI compute demand in both SoC and memory, driven by wins from 2024 and early 2025. He noted that while the timing of Q4/Q1 ramps is an 'x factor,' these wins will have a more significant impact in 2026. For 2026, he expects a similar DRAM-dominated memory mix and continued strength in VIP and networking compute, with potential upside from merchant suppliers.

CJ Muse inquired about the progression of customer conversations regarding AI infrastructure spending, seeking details on expedited meetings and actual orders, and how recent announcements are translating into tangible business visibility for Lam Research.

Answer

President and CEO Tim Archer clarified that recent AI infrastructure announcements represent future demand, while current equipment needs are driven by enterprise SSDs and NAND upgrades. He emphasized Lam Research's product relevance for leading-edge nodes, HBM, and advanced packaging. Regarding Lam's outperformance relative to WFE, Tim Archer explained that short-term comparisons are complex due to timing, but over the long term, Lam's etch and deposition markets are expected to outgrow WFE due to increasing technology intensity in 3D devices and advanced packaging. EVP and CFO Doug Bettinger concurred.

C.J. Muse asked about the translation of recent AI infrastructure spending announcements into tangible customer conversations, expedited meetings, and actual orders for Lam Research, and how this impacts business visibility. He also inquired about Lam's relative outperformance to WFE and the critical drivers for continued outperformance in 2026.

Answer

Tim Archer, President and CEO, clarified that recent announcements indicate future demand, while current equipment needs are driven by enterprise SSDs and high-bandwidth memory (HBM). He noted robust demand for 2026, emphasizing that Lam's core markets (etch and deposition) are expected to outgrow WFE long-term due to technology trends like 3D devices and advanced packaging. Doug Bettinger, Executive Vice President and CFO, concurred.

CJ Muse asked for a framework outlining the key drivers of Lam Research's expected outperformance versus WFE in 2025 and 2026, and questioned the sustainability of high gross margins given the favorable China mix.

Answer

President and CEO Timothy Archer attributed the outperformance to technology inflections like Molybdenum (Moly) adoption, gate-all-around, backside power, and advanced packaging, which increase etch and deposition intensity. EVP and CFO Douglas Bittinger confirmed the current gross margin benefits from a favorable mix but guided for a normalization to around 48% in the December quarter due to a less favorable mix and rising tariffs.

CJ Muse of Cantor Fitzgerald asked for the key drivers of Lam's expected outperformance relative to WFE in 2025 and 2026, and questioned the gross margin outlook, including the impact of China and a potential new normalized level.

Answer

President and CEO Timothy Archer attributed outperformance to technology inflections like Moly, gate-all-around, selective etch, backside power, and advanced packaging, noting that etch and deposition intensity is rising faster than WFE. EVP and CFO Doug Bittinger explained that while the current gross margin benefits from a favorable mix, he expects it to normalize to around 48% in the December quarter due to less favorable mix and rising tariffs, declining to provide a longer-term normalized view yet.

CJ Muse inquired about the recent inflection in DRAM demand, particularly from inference hyperscalers, asking about its breadth and sustainability, and whether the typically slower February quarter would exhibit normal seasonality given current supply constraints. He also sought details on the projected fiscal 2026 CapEx of $18 billion, specifically how it partitions between front-end equipment and clean room space, and the implied gross CapEx figure.

Answer

Sanjay Mehrotra, Chairman, President & CEO, affirmed strong and broadening AI trends across data centers, AI-enabled smartphones, and PCs, driving robust demand into 2026, coupled with tight supply. Mark Murphy, EVP & CFO, clarified that the fiscal 2026 CapEx would be predominantly for DRAM, covering construction, facilities, tools for node transitions, and new greenfield installations, and explained that the $18 billion figure represents net CapEx, with gross CapEx offset by government incentives, similar to the fiscal 2025 breakdown.

CJ Muse inquired about the recent inflection in DRAM demand, particularly from inference hyperscalers, its breadth and sustainability, and whether the seasonally slower February quarter would see normal seasonality or be supported by limited supply. He also asked for a breakdown of the fiscal 2026 CapEx between equipment and clean room space, and the implied gross CapEx.

Answer

CEO Sanjay Mehrotra confirmed strong and broadening AI trends across data centers, AI-enabled smartphones, and PCs, driving robust demand into 2026. He reiterated tight supply factors, lean supplier inventories, and very tight Micron DRAM supply, leading to a healthy demand-supply environment. CFO Mark Murphy stated that the vast majority of 2026 CapEx would be for DRAM, encompassing construction, facilities, and tools for node transitions, clarifying that the $18 billion figure is net CapEx and providing 2025 gross and net CapEx figures.

CJ Muse observed an inflection in DRAM demand, particularly from inference hyperscalers, and asked about the breadth and sustainability of this demand. He also questioned whether the typically slower February quarter would experience normal seasonality or if limited supply would lead to better-than-normal performance, given expectations of tightness into fiscal 2026. Additionally, he asked Mark Murphy about the implied net CapEx of $18 billion for fiscal 2026, seeking a breakdown between equipment and clean room space, and the implied gross CapEx.

Answer

Sanjay Mehrotra, CEO, confirmed strong AI trends across training and inference, broadening demand vectors for all DRAM in data centers, AI-enabled smartphones, and AI PCs (driven by higher content and Windows 10 end-of-life). He expects strong demand throughout 2026, supported by a healthy demand-supply environment and lean inventories, particularly tight Micron DRAM supply. Mark Murphy, CFO, stated that the majority of the fiscal 2026 CapEx, guided around $18 billion net, would be for DRAM, encompassing construction, facilities, tools for node transitions, and new greenfield installations. He clarified that net CapEx includes government incentives, noting that fiscal 2025 had $13.8 billion net CapEx from $15.8 billion gross CapEx, with $2 billion in government incentives from the U.S., Singapore, and Japan.

CJ Muse asked where Micron is seeing tariff-related pull-ins across its DRAM and NAND businesses and what was driving the uptick in the company's DRAM bit outlook for the year.

Answer

EVP and Chief Business Officer Sumit Sadana stated that any tariff-related pull-in impact was 'fairly modest' and that aggregate customer demand for the rest of the calendar year remains healthy. He reiterated that the improved 2025 DRAM outlook is driven by robust AI-related data center demand and a firming growth trajectory in the industrial and broad distribution markets, not by tariff effects.

CJ Muse of Cantor Fitzgerald asked for details on where Micron is observing tariff-related pull-ins across its DRAM and NAND segments and consumer versus non-consumer markets. He also asked for the drivers behind the increased DRAM bit outlook for the year.

Answer

EVP and Chief Business Officer Sumit Sadana stated that any tariff-related pull-in impact was 'fairly modest' in Q3 and that aggregate customer demand for the rest of the calendar year remains healthy. He reiterated that the improved 2025 DRAM outlook is driven by robust AI data center demand and a recovery in industrial markets, not tariff effects.

CJ Muse from Cantor Fitzgerald sought clarification on HBM market share, noting that the timeline to reach parity with overall DRAM share seems to have been pulled forward to the second half of the year, and asked about the expected revenue impact and the drivers of HBM growth in calendar 2026 (bits vs. ASP).

Answer

Chairman, President & CEO Sanjay Mehrotra confirmed that based on FQ3 results, Micron is already at a greater than $6 billion HBM revenue run rate. He stated it is possible the company will achieve its HBM market share goal earlier than the end of the calendar year due to very strong execution on yield and output, noting the 12-high yield ramp is proceeding faster than the 8-high ramp. He highlighted that this execution and a strong roadmap position Micron well for customer requirements in 2026.

CJ Muse requested more detail on the Q3 gross margin outlook, asking about factory utilization plans and whether the margin pause was solely due to depreciation. He also asked about the financial impact of the increased Investment Tax Credit (ITC) on future CapEx and depreciation.

Answer

CFO Rafael Lizardi stated that Q3 gross margin will be flat as higher revenue is offset by increased depreciation, with factory loadings remaining similar to Q2. Regarding the new tax law, Lizardi said the company was pleased, expecting significantly lower cash taxes for several years, which will benefit cash flow. He noted that CapEx plans remain consistent with prior guidance.

CJ Muse requested more detail on automotive growth drivers, particularly the trends in China versus non-China, and asked about near-term gross margin drivers beyond utilization and mix.

Answer

CEO Kurt Sievers noted that NXP's Q3 guided auto revenue is only 4% below its prior peak, with growth driven by the end of inventory burn at Western Tier-1s, while China remains strong. CFO Bill Betz identified other near-term gross margin levers, including the opportunity to refill the channel to its 11-week target, ongoing cost improvements, and the full-year effect of annual price adjustments.

CJ Muse from Cantor Fitzgerald asked for a deeper dive into the automotive recovery, specifically requesting color on growth drivers by geography, and inquired about near-term gross margin drivers beyond utilization and mix.

Answer

CEO Kurt Sievers noted that NXP's auto revenue is only 4% below its prior peak and that while China continues to grow, the significant change is the end of inventory burn at Tier 1s in Europe and the US, which allows shipments to track closer to end demand. CFO Bill Betz identified refilling the channel back to the 11-week target and improved costs from annual price adjustments as key near-term gross margin drivers.

C.J. Muse inquired about the strategic rationale for NXP's recent acquisitions (Kinara, Aviva, TTTech), questioning if they were offensive or defensive moves, and asked for an update on the automotive inventory correction and geographic demand trends.

Answer

CEO Kurt Sievers characterized the acquisitions as clearly offensive, designed to add differentiation to NXP's CoreRide platform for software-defined vehicles globally, not just in China. Regarding automotive, he noted Q2 2025 marks a turning point with year-over-year revenue expected to be flat for the first time in five quarters, driven by stabilizing orders and a pickup in Asia, even as some Western Tier 1s continue to digest inventory. He also reaffirmed a low single-digit price erosion for the full year.

CJ Muse from Cantor Fitzgerald revisited the 2025 EUV revenue outlook, questioning the change from a perceived 50% growth to 30%. He also pressed for more detail on the increased uncertainty for 2026, asking if it was purely a top-down macro view or if bottom-up customer build plans had changed.

Answer

EVP & CFO Roger Dassen corrected the premise, stating the initial 2025 EUV growth forecast was 40%, not 50%, and the revision to 30% is fully explained by a shift of revenue to the Installed Base upgrade business. Regarding 2026, he reiterated that the uncertainty stems from customers' heightened concerns over tariffs compared to three months prior, causing them to delay confirming their demand, rather than a change in fundamental build plans.

CJ Muse asked for a breakdown of the total custom silicon business between AI and non-AI applications, and for insight into the growth prospects for the non-AI portion.

Answer

CEO Matt Murphy explained that for the current and next fiscal year, the custom silicon business is "largely driven by AI," as the upside in AI programs has outpaced other opportunities. While non-AI programs are contributing, AI represents the vast majority of the business. He declined to break out specific growth rates but noted that the custom segment would naturally grow faster off a smaller base than the more established interconnect business.