Tore Svanberg

Ranked the segments directly, leading with enterprise data and comms and naming consumer and notebook storage as the cautious areas.

Answered the device technology question but explicitly declined to time revenue, tying it to when the data center transition happens.

Michael outlined three advantages: monolithic integration vs competitors' multi-chip approach, module development/testing automation since 2016, and BCD process advancing from 60nm to 40nm.

Michael confirmed DDR5 interface opportunity is real and expands SAM alongside existing PMIC position. Tony clarified not a 2026 revenue contributor but a strategic footprint expansion.

Michael confirmed aggressive capacity expansion across silicon, silicon carbide, gallium nitride, and module components. Heavy quality audits maintain high standards.

Michael noted minimal DDR4 presence but growing DDR5 business. Tony confirmed SSD/HDD uptick driven more by enterprise than consumer, with storage being data center driven.

Michael revealed near-term goal of 10 million modules per month with 100% automated testing including reliability test, using proprietary robotic systems.

Michael quantified module revenue at way less than 1/3 of enterprise data, noting growth in the last half year. Tony added 800V solutions are 2027-2028 revenue opportunities.

Bernie gave unusually granular Q3 segment guidance: enterprise data +20-30% QoQ, consumer seasonal uplift, other lines up high single digits, storage/compute some caution.

Michael confirmed engagement with multiple large and small customers, calling MPS a winner. Declined to revise the $4B SAM upward but said they will get there.

Michael detailed a decade-long diversification strategy including post-pandemic capacity expansion outside China, now positioning MPS well for both U.S. and China supply.

Michael confirmed increasing confidence about the second half but stopped short of forecasting magnitude, noting visibility has improved.

Bernie provided clear segment-level direction for Q1, noting automotive and communications as growth drivers while enterprise data will be down.

Michael gave a broad directional view of flattish enterprise data for 2025 but acknowledged limited visibility into timing of customer ramps.

Bill gave broad color on the DSP, ZF Optics, and CPX opportunity and the long-term TAM, but did not break out the specific component mix within the $600M.

CEO declined to rank the three but said each optical category grows >80% YoY; later clarified ZeroFlap will be the largest optical contributor.

CEO said ALC can ramp fast like AEC/ZF optics; Weaver content per GPU can be $2,000-$3,000, so it can also ramp quickly with inference GPUs.

Bill cited reliability and power efficiency across intra-rack and rack-to-rack links up to 7 meters, with a stronger value proposition and ASP uplift at 200 gig per lane.

Bill said ALCs and ZF optics are optical while OmniConnect is initially copper with near-package optics later, all delivered as system-level solutions.

Bill said the purpose-built 112-gig VSR SerDes fits 1,200 on a max reticle die for 120 Tbps at 10-inch reach, and near-package optics will leverage the microLED work.

Asked whether the largest customer is pausing or expanding in fiscal 2026; Fleming said that customer is expected to remain the largest by year-end, and by far.

Asked for color on AEC use cases in row-scale networks; Brennan said rack densification is expanding the AEC opportunity, driven by reliability and power efficiency.

Brennan framed AEC demand across front-end, back-end scale-out and disaggregated chassis, with the 50-to-100 gig transition broadening toward the end of fiscal '26.

Bill confirmed Credo will pursue the same system-level model for PCIe AECs as for Ethernet, including a full-rack AEC demonstration at GTC.

Bill said Credo has good visibility into deployments and consumption; a high consumption rate and little logistics-partner inventory suggest real sell-through.

Bill declined to link the trend to any single GPU supplier's architecture, saying customer-specific network optimization is what drives Credo's success.

Roche said double-digit compounded analog growth for several years is possible, citing data center SAM of $1B-$1.5B per gigawatt.

Puccio confirmed capacity to support $20B 2030 vision through internal and external expansion, with no trouble getting needed external capacity to date.

Roche gave a thorough answer noting calm customer environment, doubled internal capacity, and increased external optionality, with some stress points but overall good shape.

Vince identified power management as the primary area for doubling down, alongside digital niches in low-power/low-latency compute and ML algorithms for base stations and consumer. Customers demanding more complete solutions.

Vince outlined a target to double the multi-hundred-million-dollar automation business by 2030, driven by sensing modality growth, demographic pressures, and manufacturing localization trends.

Vince outlined content growth from hundreds of dollars in fixed arm robots to potentially thousands in autonomous/humanoid systems, with ADI positioned for sensing, edge computing, and connectivity.

Vince explained ADI has desensitized geographic concentration, invested in US and European fabs, and secured dual sourcing with partners for 95% of products by end of 2026.

Vince confirmed strong new product conversion rates in ATE, automotive, and data center, with new generations capturing higher ASPs and legacy products pulled along by new anchor products.

Vince detailed CodeFusion Studio and ADI Assure platforms as software investments that complement analog franchise. Positioned software as enabling easier customer adoption rather than a standalone business.

Jitendra tied exceptional new-customer traction to in-network compute and HyperCast in COSMOS, now part of the orchestration layer, deepening stickiness across generations.

CFO said supply is in place through year-end with a diversified back-end supply chain despite industry supply pockets.

CFO said Aries keeps growing strongly on PCIe 6 while Scorpio remains the largest and fastest-growing business.

Management said software-defined fabrics let one silicon serve many customized scale-up topologies, and the influx of opportunities justifies R&D.

Management framed Xscale as adding fiber chip coupling for optically enabled Scorpio switches, with material optical revenue in 2028-2029.

Management walked through five growth steps and said X-Series carries much higher dollar content, a new baseline through 2026, without confirming a doubling.

Management explained the gearbox bridges PCIe Gen 5 and Gen 6 while retiming, adding higher ASP to the Aries retimer TAM.

Management confirmed both P and X Series contribute in 2025, with P first and X later in the year.

Steve said lead times are broadly expanding and within a quarter or so nothing may be available in four to six weeks as die inventory falls toward the model.

Steve clarified the High-Performance Compute pillar comprises the microprocessor, FPGA, and timing businesses.

Steve outlined two hurdles before buybacks: repay ~$350 million borrowed to cover dividends during the downturn, then reduce leverage from 4.2x to the 1.5x target. No buybacks expected near term.

Steve confirmed utilization improvement begins well before inventory reaches 130-150 day target, with substantial step function improvement in December quarter as wafer starts increase.

Eric confirmed most restructuring savings are already reflected. March quarter OpEx came in $7-8M below guidance due to early March RIF implementation.

Steve identified a gap in low-end 32-bit products left unaddressed during 8-bit to 32-bit transition. First products from combined group launching in early January.

Ganesh said the shift depends on when customers feel supply risk or regain business confidence, noting these reversals can happen quickly.

Eric confirmed theoretical capacity to double revenue at current mix without major CapEx, though future mix could require some investment.

Hassane said onsemi is comfortable with external supply and still does over 60% internal manufacturing, framing constraints as a short-term catch-up not a cap.

El-Khoury said ON has the toolbox of IP and technology needed. Deploying products quickly. Doubled Treo products YoY. Same mindset across AI data center and automotive.

El-Khoury said vGaN started from IP acquisition in 2024. Fab turned on, first products yielding, lead customers evaluating first-gen samples. Already working on second gen. Revenue expected in 2027.

Trent said mostly maintenance CapEx now. Mid-single-digit percent of revenue for the full year. Lower capital intensity supports 100% FCF buyback.

El-Khoury said Treo wins are broad across auto, AI, industrial, medical. On track to double products YoY. First production revenue shipping. 60-70% margin profile confirmed.

El-Khoury will share milestones like product introductions and margin confirmation but won't provide quarterly or annual Treo-specific revenue breakouts.

Trent said 20-25bps per point from current levels to low-60s utilization, then revert to 15-20bps at scale. Based on current capacity before Fabrite changes.