Krish Sankar

Rejected the supply-constraint framing outright and cited the sequential inventory build alongside improved turns.

Gave an explicit near-term gross margin range, one of the more specific forward answers on the call.

Doug highlighted PECVD underfill in advanced packaging as an underappreciated contributor; Tim added NAND backside stress management via Vector DT as another growth driver.

Doug reiterated the WFE upgrade was broad-based across all customer segments, driven by incrementally available clean room space rather than any single driver.

Doug listed global factories across Oregon, California, Ohio, Malaysia, Taiwan, Korea, and Austria, noting flexibility to adjust but feeling good about the current setup.

Doug acknowledged quarter-to-quarter variability in NAND but said overall trajectory toward the $40B opportunity has strengthened from a quarter ago.

Doug confirmed global multinationals in China grew over 90% Q/Q but declined to attribute a specific reason.

Doug declined to detail manufacturing specifics but noted the ability to adjust tool placement with enough lead time depending on the rules.

Doug quantified the impact at roughly $700 million of forecasted 2025 revenue, second-half weighted, with lead times unchanged.

Management pointed to NAND technology upgrades plus advanced packaging, backside power, gate-all-around and dry photoresist as combined outperformance drivers.

Sanjay said 2027 stays tight and tightness continues beyond 2027, citing long capacity lead times and structurally strong AI demand.

Sanjay reiterated floor-price gross margins will be well beyond past-cycle peaks but declined to confirm specific per-gigabyte pricing.

Sanjay confirmed SCAs are multiyear (vs typical one-year LTAs), designed to bring stability and visibility across tight and loose industry environments, with robust provisions for both Micron and customers, but declined to give specifics on pricing or cancellation terms.

Mark declined to guide Q4 gross margin specifically but said tight market conditions beyond 2026 and structural supply constraints support sustained margin strength, with the caveat that incremental price gains have less effect at these high margin levels.

Mark committed to gross margin expansion beyond fiscal Q2 for both DRAM and NAND, while noting the pace will be more gradual than the Q1/Q2 jumps due to mathematical compression at high margin levels.

Sanjay declined to specify Micron's HBM share target in the $100B 2028 TAM, reiterating the company will manage HBM/non-HBM mix around portfolio and customer strategy in the current tight-supply environment.

Sanjay clarified that HBM4 uses Micron's in-house base die while HBM4e (a 2027 product with TSMC partnership) will offer both standard and customized base logic dies, with customized skus expected to carry higher gross margins.

Sanjay declined to quantify 2026 HBM supply but said Micron has front-end fungibility on one-beta wafers plus assembly-and-test capacity headroom, giving flexibility to manage HBM vs non-HBM mix with ROI in mind.

Sanjay said HBM4 trade ratio is greater than 3 (above HBM3E's ~3), HBM4E will be even higher, and declined to break out GPU vs ASIC HBM margin differences.

Sanjay pointed to tight leading-edge DRAM supply, HBM trade ratio and consumer inventory normalization as real drivers, without quantifying tariff pull-in versus underlying demand.

Charlie confirmed NAND prices significantly affect the business; Rob declined to speculate on individual customers or future design-win timing.

CEO said most enterprise AI purchasing is still in the cloud with little on-prem native buildout; sovereign clouds and titans do more on-prem, plus some high-end banking, automotive and pharma, but otherwise it is very low.

CFO said none of the RPO growth is from hyperscalers, hyperscaler revenue is recognized in product revenue (not ratable) and excluded from RPO.

CEO said NAND-price volatility isn't a big driver of the hyperscaler model shift; different hyperscalers have different purchasing preferences, producing a more mixed/nuanced model with varying COGS.

CTO said other-hyperscaler engagements are progressing well (co-engineering, not traditional sales) with testing and multiple POCs underway but still early stage; some phases may accelerate given Meta learnings.

CFO attributed the RPO deceleration to lower Evergreen//One TCV this year, noting a record Q4 Evergreen//One performance (+20%) and growth built into FY26 guidance.

Management said double-digit exabytes in CY2026 reflects customer coordination though sales are TBD; licensing is tied to shipped capacity.

Management said the 150-millimeter Fremont fab has ample capacity; upgrades are about capability, not capacity.

Management said error correction is done in the control-systems layer in ambient conditions, not on the quantum chip itself.

Management said Rigetti does not need a new fab for quantum advantage and questioned peers' CMOS-foundry acquisitions.

Asked by an associate (Steven, labelled 'Kinney Chin' in the transcript) on behalf of Krish Sankar. Management said both Novera orders are complete systems including fridges and controls, with future upgrades adding revenue.

Steven (associate, on for Krish Sankar) asked about current quantum-asset valuations and whether M&A is important. Management said it sees no M&A targets that would accelerate the roadmap and views itself in the technology leadership camp.

Asked by an associate (Steven Silverman) on behalf of Krish Sankar. The CFO attributed elevated Q1 OpEx to FICA and spot bonuses and expected a modest reduction, without giving a run-rate target.

Asked by an associate (Steven Silverman) on behalf of Krish Sankar. Management laid out the chiplet tiling roadmap: 36-qubit by mid-2025, then 100-plus qubits by year-end.

Philip Carter confirmed data center under $100M and auto ~$250M annually. Data center growing faster than auto. Strong design win traction and good bookings momentum in both segments.

Philip Carter confirmed China total revenue less than $200M annually with handset less than $20M. Minimal exposure reflecting deliberate strategy to focus on premium and profitable segments.

Phil confirmed strong visibility across both iOS and Android. Book-to-bill above 1, low channel inventories. No material difference in visibility between the two platforms.

Phil committed to disciplined OpEx management. Past increases were targeted R&D investments. Any modest future increases will be focused on core R&D. Expects nominal OpEx growth rates going forward.

Raji quantified auto at ~$60M/quarter, up significantly YoY. Highlighted wins at BYD, Nissan, Ford across 5G telematics, infotainment, power isolation. Long design cycles provide durable revenue visibility.

Liam listed design wins (Motorola, Logitech, Charter, Mercedes-Benz) but did not directly answer how long the upgrade cycle typically lasts historically.

Confirmed normal seasonality: mobile down in March, broad markets expected to continue modest sequential growth. Only guides one quarter at a time.

CFO said roughly 70% of the Windows 11 refresh is complete and still driving SMB demand, increasingly augmented by rising demand for AI PCs, edge AI and agentic workloads.

Interim CEO said the target refers to AI PCs excluding workstations, with attach rates similar to the rest of the PC business, plus incremental solution opportunities; PS head noted AI PCs were 46% of mix and are expected to reach 60%-70% in 2027.

Karen confirmed double-digit unit decline without further specifics. Ketan noted revenue growth still expected through Win 11 momentum, AIPC mix shift (35% and rising), and pricing actions driving revenue TAM expansion despite unit decline.

Karen declined to quantify net memory impact but noted memory costs up ~100% sequentially, BOM share rising from 15-18% to roughly 35% for year. Ketan Patel (PS head) outlined mitigation playbook across supply management, cost actions, and demand shaping.

Enrique explained IDC definition (NPU built-in) for the 15% figure, HP's next-gen AI PCs require 40+ TOPS NPU, AI companion, enhanced security.

Enrique noted commodity pressure through H1 FY25, mitigation from repricing, lower-cost component qualification, and SKU consolidation in H2.

Mosley emphasized aerial density focus over unit volume growth. Adding more capacity per spindle nets more exabytes than adding units. Romano pointed to Mozaic 5 at 50TB qualifying late calendar 2027.

Romano declined to break out individual contributions but confirmed pricing, mix, and cost are all contributing positively and interrelated. HAMR mix rising but specific percentage not disclosed.

Mosley indicated pricing will adjust slowly over time as demand develops. Romano noted only the first customer received a slightly lower price for helping with initial qualification.

Romano confirmed confidence in continued sequential gross margin improvement driven by mix, pricing strategy, cost efficiency, and HAMR ramp in second half of calendar 2025.

Romano disclosed surveillance accounted for roughly 4-5 exabytes of nearline shipments per quarter.

Pressed on the market-share decline; Dave reframed share as an outcome of the build-to-order strategy, and Gianluca said the pricing environment continues to improve each quarter.

Rene reframed from chip count to core count. GPUs are reticle-limited while CPU core counts could double or quadruple. Dedicated CPU racks for Agentic orchestration will shift ratios significantly.

Rene described a multi-year evolution where CPU workloads expand at the expense of GPUs as inference goes agentic. Custom CPU-based solutions for prefill and decode will proliferate. AI compute will also migrate to smaller form factors in physical AI and edge devices, all of which are Arm-based.

Rene described the training-to-inference shift as an enormous trend for Arm. Meta partnership is enabling scalable software between cloud and edge. Battery-level AI compute is Arm's unique domain.

Jason explained ACV has a ~7% annual escalator baseline from ATA deals, with AI/CSS upside pushing it to mid-teens recently. Majority of future growth will come from royalties, not licensing.

Rene cited strong local-brand growth (Xiaomi, Oppo, Vivo) and said the premium/flagship market in China is all v9, with the transition moving into the midrange.

Clarke said traditional servers are x86 today, GPU systems skew toward Arm, and he is optimistic about Vera for future advanced workloads.

David declined to share specific breakdown but Jeff noted over 4,000 customers, record enterprise revenue quarter, and enterprise as fastest-growing portion of five-quarter pipeline.

Jeff confirmed DLC bias in backlog for large deployments. Yvonne explained the $5B AI revenue raise offset by ~$1B reduction in CSG/traditional server/storage expectations.

Jeff emphasized training and inference as separate opportunities, cited enterprise use cases like their digital service assistant, and highlighted Project Lightning parallel file system.

Jeff did not quantify attach rates but gave extensive strategic answer on storage opportunity including three-tier architecture, PowerScale platform, and Project Lightning parallel file system for AI.

C.C. Wei said GPU or ASIC makes no difference to TSMC since both use leading-edge technologies and TSMC supports all types.

C.C. Wei cited a customer's ~$50 billion per gigawatt investment but declined to quantify TSMC's wafer content, noting it varies by project and involves many chips.

Wendell gave direct figures: the Japan fab ramps to 40,000 capacity, with 2025 revenue contribution not significant relative to the whole company.

Wendell deferred second-half color as too early given tariff uncertainty; C.C. Wei said N2 demand is very strong with tape-outs exceeding expectations.

Steven stood in for Krish Sankar. Baratz said only annealing delivers commercial production value today and that a per-network 'operations' count does not apply to annealing.

Markovich said the backlog is roughly 50-50 commercial versus research, including the $20M FAU system sale and the enterprise QCaaS deal.

Baratz said dual-rail is not made at SkyWater, and D-Wave fabricates its own Josephson junctions; he is skeptical of the IonQ-SkyWater deal and pursuing alternate fab.

CEO confirmed high 300s exabyte estimate for data center in 2026, building on 130% YoY growth in 2024 and high teens in 2025, with mid-40s percent growth in 2026.

CEO confirmed data center revenue growth is majority AI-driven, though did not provide a specific percentage split.

Irving said LTA repricing occurs through the year as contracts kick in, with more rapid increases in non-nearline.

Asked about Flash ASP implications of the Q2 guide; management pointed to flat blended pricing and a modest cost headwind, direct.