Teradyne's Q3 2025: Contradictions Emerge on Mobile SoC Demand, Compute Growth, Tariff Impacts, and Robotics Outlook

Date of Call: October 29, 2025

Financials Results

• Revenue: $769M, up 18% sequentially (Q3)
• EPS: $0.85 non-GAAP, up 49% (management cited improvement in Q3)
• Gross Margin: 58.5% non-GAAP, above guidance (Q3)
• Operating Margin: 20.4% non-GAAP operating profit (Q3)

Guidance:

• Q4 revenue expected $920M to $1.0B.
• Q4 non-GAAP gross margin 57% to 58% (midpoint 57.5%).
• Q4 OpEx expected 31% to 33% of sales (non-GAAP operating profit midpoint ~25.5%).
• Q4 non-GAAP EPS $1.20 to $1.46 (GAAP EPS $1.12 to $1.39) on ~157M diluted shares.
• Guidance assumes continued AI-driven demand and includes some one-time supply costs to meet accelerated deliveries.

Business Commentary:

* Revenue and Earnings Growth: - Teradyne reported 18% sequential growth and 49% increase in non-GAAP EPS for Q3 2025. - The growth was driven by strong demand for AI applications in semiconductor tests, particularly in compute and networking devices.

• AI-Driven Demand and Market Expansion:
• Compute-related revenue was more than 50% higher than expected in the second half of 2025.

• This was due to the rapid advancements in AI and the resulting increase in demand for high-performance processors and networking devices.

• Memory Segment Performance:

• Memory test sales more than doubled from Q2 to $128 million in Q3, driven by AI applications and HBM-related demand.

• Despite a declining total memory TAM for 2025, Teradyne expects its memory revenue to sustain at 2024 levels due to strategic market positioning.

• Robotics Segment Developments:

• In Q3, 8% of robotics sales were for AI-related products, up from 6% in Q2.
• This increase aligns with the company's strategy to establish UR cobots as the preferred platform for AI-driven work cell applications and expand its OEM channels.

Sentiment Analysis:

Overall Tone: Positive

• Management: "AI is the dominant driver of our business" and noted Q3 "grew sequential revenue 18% and non-GAAP EPS by 49%." CFO provided strong Q4 guidance ($920M–$1B revenue, 57%–58% gross margin) and described ramping capacity to meet accelerated AI-related demand.

Q&A:

• Question from Christopher Muse (Cantor Fitzgerald & Co.): How much of the roughly $150M upside for December is driven by HBM, VIP, networking, SLT or other segments?
  Response: Upside is primarily in compute and memory: roughly ~2/3 compute (VIP/networking) and ~1/3 memory, with HBM strongly represented.

• Question from Christopher Muse (Cantor Fitzgerald & Co.): How are you thinking about compute intensity, increased test insertions and test time as compute drives leading edge?
  Response: Test intensity will rise as die sizes, chiplet complexity and data‑center reliability demands increase, expanding TAM and favoring suppliers that can support dual‑sourcing and complex test flows.

• Question from Mehdi Hosseini (Susquehanna Financial Group): How will design wins in wafer‑level test manifest and will burn‑in be part of those design wins?
  Response: SLT (system-level/test and burn‑in) is critical and contiguous with burn‑in; new module formats (e.g., CoWoP) increase system test and burn‑in needs.

• Question from Mehdi Hosseini (Susquehanna Financial Group): Were the recent design wins factored into your prior 2028 EPS target and how should we view that target now?
  Response: Design wins cited are not in current guide; management expects the long‑term destination remains achievable but the composition will shift much more toward AI‑driven segments; update to model in January.

• Question from Timothy Arcuri (UBS Investment Bank): Is the Semi Test uplift in the guidance roughly evenly split between memory and SSD for calendar Q4?
  Response: Not even split; it's ~2/3 compute/networking and ~1/3 memory driving the uplift.

• Question from Timothy Arcuri (UBS Investment Bank): How should we think about revenue shaping into next year — should Q1 be down following heavy Q4 shipments?
  Response: Projects are lumpy but overall demand is accelerating; seasonal shaping is changing as business shifts from mobile to compute‑driven ramps; management will provide a detailed update in January.

• Question from Unknown Analyst (on behalf of Krish Sankar, TD Cowen): For VIP and merchant GPU opportunities, how much expansion in customer base do you see and are relationships direct or via foundries?
  Response: VIP demand is highly concentrated (driven by ~2 customers); hyperscalers increasingly control supply chains; merchant GPU test decisions are made by the specifiers (not foundries), so go‑to‑market targets specifiers directly.

• Question from Unknown Analyst (on behalf of Krish Sankar, TD Cowen): For HDD strength, do you expect strong double‑digit Q4 growth in System Test or will impact be greater in 2026?
  Response: Orders are increasing but HDD capacity additions take time; meaningful effect expected more in 2026 than in 2025.

• Question from Shane Brett (Morgan Stanley): Do you expect SoC test strength from the December quarter to carry into H1 or are you baking in conservatism for March?
  Response: Projects are accelerating and Q4 was a new high watermark, but timing remains lumpy; management expects robust demand into 2026 but cautions against linear extrapolation quarter‑to‑quarter.

• Question from Shane Brett (Morgan Stanley): How low has NAND been this year and what growth do you expect for NAND going forward?
  Response: NAND is at a very low point; its recovery depends on mobile unit demand and potential SSD growth for AI; DRAM/HBM is currently the stronger segment.

• Question from James Schneider (Goldman Sachs): What do you expect for mobile SoC heading into next year and how much test content increase with N2?
  Response: Mobile SoC volumes have been low; 2026 could improve but magnitude is uncertain — N2 and advanced packaging should raise test intensity by double digits per part; unit volumes are the key X factor.

• Question from James Schneider (Goldman Sachs): From a modeling perspective, how much OpEx should increase relative to revenue growth?
  Response: Target is OpEx growth at roughly 50% of revenue growth (OpEx increases about half as fast as revenue over time).

• Question from Samik Chatterjee (JPMorgan): How much of the memory improvement is market share versus better industry purchasing and which end markets should gain share into 2026?
  Response: Teradyne has high share in final/performance memory test (HBM/DRAM final test); wafer‑sort share is lower — 2026 should see memory expansion, but wafer‑sort SAM may grow meaning the composition may not yield large share gains even as revenue rises.

• Question from Samik Chatterjee (JPMorgan): What are the drivers of the Q4 gross margin guide versus Q3?
  Response: Volume is a tailwind, but margin is offset by factory expansion investments and some one‑time elevated supply costs to meet accelerated customer deliveries.

• Question from Brian Chin (Stifel): How much of AI accelerator revenue was weighted to the second half and was compute revenue above expectations for the year?
  Response: Compute (VIP and networking) materially exceeded early‑year expectations and drove the recovery; mobile and auto were weaker than anticipated.

• Question from Brian Chin (Stifel): For Robotics, do you expect seasonal Q4 improvement and does Titan SLT win incumbent multiple insertions?
  Response: Robotics should see seasonal uplift in Q4 but forecasting is limited; Titan SLT offers upgradability and an incumbency advantage that supports follow‑on insertions.

• Question from Vedvati Shrotre (Evercore ISI): Has the singulated‑stack HBM test increased TAM and will all suppliers adopt it?
  Response: Today only one major HBM manufacturer routinely uses singulated stack testing; broader adoption would depend on demonstrated downstream yield improvements.

• Question from Vedvati Shrotre (Evercore ISI): How much of SoC in H2'25 is compute and does the 2028 mix (1/3 mobility, 1/3 compute, 1/3 auto) change?
  Response: Compute is a significant component of H2'25 (AI‑driven share rose from ~50% in Q3 to ~60% in Q4); management expects the long‑term mix to shift materially toward compute/AI and will update the model in January.

• Question from David Duley (Steelhead Securities): Have you won any stand‑alone GPU business included in guidance or revenue?
  Response: No — no standalone GPU merchant wins are included in guidance and there was no revenue from that in Q3.

• Question from David Duley (Steelhead Securities): Is the HBM ramp into Q4 driven by HBM4, new insertions, or other factors?
  Response: HBM ramp is driven roughly half by new test insertions (e.g., singulated stack tests) and half by additional capacity needs, and is driven by HBM4 demand.

Contradiction Point 1

Outlook for Mobile SoC Demand in 2026

It involves differing expectations for the recovery of mobile SoC demand in 2026, which could impact revenue projections and investor expectations.

Can you provide insights into next year's mobile SoC demand? - James Schneider (Goldman Sachs Group, Inc., Research Division)

2025Q3: Mobile SoC is at low levels. We're optimistic about an improvement, but we're unsure of its magnitude. The key will be unit volume. - Gregory Smith(CEO)

When will mobile SOC trends recover, and what impact will they have on gross margins? - James Schneider (Goldman Sachs)

2025Q2: The mobile recovery is more likely in the second half of 2026. The impact on gross margins should be neutral as we scale volume. - Sanjay Mehta(CFO)

Contradiction Point 2

Compute Demand and Market Composition

It involves changes in expectations regarding the demand dynamics for the compute segment, which is a key driver for the company's revenue.

What was the compute segment's weight for Q2, and is AI revenue above expectations? - Brian Chin (Stifel, Nicolaus & Company, Incorporated, Research Division)

2025Q3: Compute revenue was significantly above expectations. AI-driven segments, including compute and networking, were strong. - Gregory Smith(CEO)

What are your observations on AI accelerators? Is the 2-nanometer mobility win exclusive to your large existing customer? Have you expanded design wins in that area? - Christopher Muse (Cantor Fitzgerald)

2025Q1: In general, we're bullish about the AI market. We're also bullish about compute. We don't see material pushout, but there is some uncertainty around the second half of this year. - Gregory Smith(CEO)

Contradiction Point 3

Impact of Tariffs on Supply Chain

It involves the impact of tariffs on the supply chain, which could affect the company's ability to deliver products and maintain its competitive position.

Were recent design wins factored into the 2028 EPS target? - Mehdi Hosseini (Susquehanna Financial Group, LLLP, Research Division)

2025Q3: We have shifted a significant portion of our onshore capacity from North America to Asia. Therefore, our factory cost and utilization have been negatively impacted. - Sanjay Mehta(CFO)

Greg, you mentioned tariff-related pushouts at your Analyst Day 7 weeks ago, but now some OSAT customers are experiencing pull-ins. What dynamics are driving this shift? Are the pushouts primarily from auto, analog, or industrial sectors, or are they also affecting mobile and HPC? - Sreekrishnan Sankararayanan (TD Cowen)

2025Q1: The effects of pull-ins for end orders like people hurrying to get chips are mostly affecting capacity that's already in place. It is not the factor we haven't seen significant pull-ins into Q1 or into Q2 to provide additional capital equipment to support that. - Gregory Smith(CEO)

Contradiction Point 4

Gross Margin and Operational Expenses

It involves changes in expectations regarding operational expenses and gross margin, which are critical indicators for investors.

What OpEx leverage should we expect in 2026? - James Schneider (Goldman Sachs Group, Inc., Research Division)

2025Q3: We anticipate OpEx growth of approximately half the rate of revenue growth. This is our long-term target. - Sanjay Mehta(CFO)

What is the expected full-year gross margin range without guiding for the second half due to visibility challenges, and do you have an updated perspective on OpEx? - Christopher Muse (Cantor Fitzgerald)

2025Q1: From an OpEx perspective, the story is consistent with a variable compensation model perspective. We'll have a favorable impact if revenue comes down, lower spend, and if it goes up, we'll have more spend, but the narrative is consistent. - Sanjay Mehta(CFO)

Contradiction Point 5

Impact of Large Robotics Win in 2026

It involves differing expectations regarding the impact of a large robotics win in 2026, which could influence revenue projections and strategic planning.

How is the growth in Semi Test split between memory and SSD? - Timothy Arcuri (UBS Investment Bank, Research Division)

2025Q3: The opportunity is primarily driven by manufacturing in the U.S., with little impact in 2025. The demand in 2026 will represent a sizable fraction of the UR business, but exact numbers are not provided. - Gregory Smith(CEO)

Was the 9% growth in robotics primarily due to restructuring, and how will that impact sales? - Robert Stephen Barger (KeyBanc Capital Markets)

2025Q2: The upside in Q4 is mainly from compute and memory. About 67% of this upside is from compute, and 33% is from memory. The majority of the memory growth is in HBM. - Gregory Smith(CEO)