Harsh Kumar

Steve confirmed strategic shift to include advanced-node, high-performance products alongside traditional MCU/analog. Data Center Solutions, AI business unit, and FPGA initiatives will pull overall CAGR higher. Series of new products planned across all three areas.

Steve noted MCU and analog growth alternate quarter to quarter. June quarter analog led; September quarter MCU led with 9.7% growth. With thousands of customers, contributions are never perfectly linear.

Steve declined to break out specific end market performance, instead emphasizing broad-based growth across all product lines and geographies.

Eric confirmed $49.3 million sell-through gap versus $103 million prior quarter. Steve noted inventory declined from 266 to 251 to 214 days with target of 195-200 by September end.

Steve confirmed demand signals predated tariff concerns and cited A&D growing to 17-18% of revenue as a standout market. Semiconductors currently exempt from tariffs.

Steve confirmed the plan assumes some recovery driven by inventory depletion and new design wins, but details deferred to March 3.

Steve cited 2.5-3 months of sell-through as the historical norm for distributor inventory and sees no reason it wouldn't return to that level.

Ganesh candidly said green shoots have lost some intensity. Expedites and pull-ins continue rising but customer confidence to place backlog remains lacking.

Ganesh acknowledged pressure from purchasing managers but said pricing is not the gross margin driver; underutilization is the real issue.

Murphy explained multi-year visibility comes from customer planning cycles requiring 6-8 quarter forward coordination on R&D, capacity, and ramps. Called FY2027 and FY2028 forecasts 'base case, not dream the dream.'

Murphy guided to model 20% as safe base case, noting it's tied to handful of programs. Upside possible but being conservative given historical noise. Second half and exit rate much higher than first half, building momentum into FY2028.

Koopmans confirmed several programs already in production since late 2024, with new programs moving to production every quarter. Murphy cited strong design win momentum since AI Day across hyperscalers and emerging customers.

Murphy noted on-prem enterprise is small and relatively stable, not a meaningful drag going forward. Spending has shifted so dramatically to AI that on-prem is no longer a major contributor.

Murphy contextualized that Q1 followed a 20%+ sequential Q3-Q4 step-up. Year-over-year data center growing 70%+ at scale. AI is fastest growing portion and tracking toward majority of total company.

Murphy outlined barriers to entry: technology leadership, manufacturing scale, supply chain relationships, business model flexibility. Noted only Marvell and one competitor can service the market.

Murphy declined to give more precision beyond 'above 10%' for Q1 cloud+AI sequential growth, deflecting the request for teens-level detail.

El-Khoury said auto grew 70% over 5 years on flat SAAR from content growth. Sticking with high single-digit auto growth above SAAR. Trent said ~25% higher revenue needed for full utilization from consistent mix.

El-Khoury said don't read into QoQ, look at H2 vs H1. Auto bottomed in Q2 as guided. Growing closer to demand but no restocking. Visibility improved but not where they'd like. Managing with discipline.

Trent confirmed 25-30bps per utilization point math still holds. Utilization will drive margin into 2026. El-Khoury added auto content grew 70% over 5 years on flat SAAR, sticking with high single-digit growth model.

El-Khoury said AI data center is the primary driver of Other segment growth. Client computing is a small piece. May provide more detail as AI DC becomes sizable.

El-Khoury said no inventory write-off benefit in recovery. 900bps is purely utilization-driven. Base inventory at 121 days is healthy within target. Strategic inventory will burn through.

El-Khoury said guidance is purely ON-specific data-driven, no padding. Highlighted internal and disti inventory discipline, disti down $55M in Q4.

El-Khoury said exact under-shipping magnitude is unknown because demand itself hasn't stabilized. Managing to visible data rather than forecasting a snap-back.

Harsh Kumar now at BMO (previously Piper Sandler). Baratz said QCI integration is going well, remains on track for the 17-qubit system this year, with a mixed annealing/gate customer pipeline.

Baratz said dual-rail plus cryogenic control needs far fewer physical qubits per logical qubit and fewer control lines, giving a scaling and timing advantage.

Baratz said D-Wave seeks to solve government problems today rather than R&D funding, and agencies are starting to realize its systems can deliver value.

Baratz said prebuilt libraries are possible but D-Wave is currently building applications on customer data with Japan Tobacco and Triumph rather than templated offerings.

Baratz said Davidson is a separate partnership focused on US defense optimization applications, not one of the previously referenced system-sale opportunities.

Baratz said this is the first tight integration with an exascale supercomputer, with joint AI-workflow research expected at Jülich.

Phil highlighted complementary technology portfolios, particularly in intensified products. Combination expected to reduce volatility, increase RF scale, bring valuable engineers, and create broad market synthesis.

Phil described broadest RF portfolio in the industry from antenna to pads. AI workloads moving to edge increase transmit capability requirements. Combination also provides platform for WiFi and other wireless innovation.

Phil moved product marketing functions from a centralized group back into business units for tighter engineering alignment. Sales team now focused on revenue generation and customer acquisition. New sales executive bringing fresh perspective.

Phil attributed outperformance to both stronger units and favorable phone mix toward models with higher Skyworks content. Noted these factors are difficult to predict 3-4 quarters out. Forward guidance comprehends previously disclosed content decline.

Phil pointed to secular trends driving RF content growth: more transmit power, new frequency bands, MIMO. Baseband choice may create incremental opportunities. Would not confirm specifics.

Hock disclosed backlog mix: ~$53B XPUs, ~$20B networking/DSPs/optics. Said silicon photonics will be deployed 'not anytime soon', copper and pluggable optics still hold in scale-up.

Hock confirmed XPU share will grow multi-generationally at each customer as software stacks stabilize and models improve, progressive share gains across generations.

Hock said hyperscalers choose proven networking technology above all, Broadcom's 10+ year track record in switching and routing plus doubled Tomahawk 5 radix for a specific hyperscaler drives preference.

Hock explained networking content expands near-exponentially as clusters scale, growing from 5-10% to 15-20% of AI silicon. Sovereign data centers stick to merchant silicon due to lacking custom software stacks.