Onsemi Semiconductor (ON.US) Q3 conference call: CEO denies downplaying automotive business and expects AI data center revenue to double next year
The Zhitong Finance App learned that after the US stock market on November 2 (Monday), Ansemi Semiconductor announced the results for the third quarter of 2026. Revenue for the quarter was US$1,722 million, up 11% year on year and 7% month on month, higher than the median value of the company's guidance, and about US$1.71 billion higher than analysts' expectations; non-GAAP gross margin of 41.2%, up 190 basis points from month to month, for the fifth consecutive quarter of expansion; non-GAAP operating margin of 23.2%; non-GAAP diluted earnings per share of $0.90, up 43% year over year. The company expects fourth-quarter revenue of $1,715 billion to $1,815 million, non-GAAP earnings of $0.95 to $1.07, with median values of US$1,765 billion and US$1.01, respectively, higher than analysts' expectations of US$1.74 billion and US$0.96; in median terms, revenue growth of 2.5% month-on-month is higher than seasonal. The stock price failed to receive a boost after the results were announced: the stock closed down 2.18% to $70.11 during the regular trading session on Monday, and fell further by about 5% after the market, to $66.56 at 16:42 EST.
The issue that received the most attention during this call was not the extent to which revenue and earnings per share exceeded expectations, but rather the contraction of the automobile business. Automobile revenue for the quarter was US$777 million, down 1% month-on-month; CFO Thad Trent further anticipated a “high single-digit percentage” decline in the fourth quarter, while total revenue for the same period was expected to increase 2.5% month-on-month. The company attributed the reason to supply redistribution — while supply was still limited in some technical areas, the third and fourth quarters “intended to allocate supply to markets with stronger terminal demand, such as AI data centers and industries,” while automotive customers “supported the delivery of all terminal requirements, but not orders for inventory recovery.” Five of the nine questioning analysts questioned this trade-off. The core question was whether long-term shares would be damaged and where would the incremental supply needed to resume growth next year come from. El-Khoury clearly denied this: “We are not downplaying the automobile market”, “this is short-term, supply-driven, not strategy-driven”, and “this just shows that there is no loss of share; it is simply a matter of replenishing inventory, not terminal demand.” He said the company will resume supply in the first quarter, and the automotive business will return to the long-term growth target of “higher than SAAR” in 2027.
Three business segments: PSG increased 21% year over year
By business segment, Power Solutions (PSG) revenue was US$890 million, up 7% month-on-month and 21% year-on-year, accounting for about 52% of total revenue; analog and mixed signal business (AMG) revenue of US$597 million, up 9% month-on-month and 2% year-on-year; and Intelligent Sensing Business (ISG) revenue of US$235 million, up 3% month-on-month and 2% year-on-year. PSG's year-on-year growth rate was significantly higher than the remaining two divisions, in line with the direction of concentrated release of AI data centers and industrial power demand.
AI data center: about US$620 million for the whole year, doubling to a higher base next year
Starting this quarter, the company removed the AI data center from the “other” terminal market and disclosed it separately as the fourth largest terminal market. The business's revenue for the quarter was US$180 million, up 39% month-on-month and 175% year-on-year, accounting for about 10% of total revenue. Management expects a month-on-month increase of more than ten percentage points (mid-teens) in the fourth quarter, with a year-on-year increase of about US$620 million for the full year of 2026, an increase of 140%; in 2027, it is still expected to double the higher base in 2026.
El-Khoury said the company began bulk shipments to several iconic platforms in the third quarter, including Google, which simultaneously supplies silicon carbide and silicon MOSFETs for power supply units (PSUs), as well as 54 volt to 12 volt applications. He emphasized that the company's revenue covers the complete power link “from the power supply unit to the low-voltage intelligent power level next to the processor” and that the move to an 800-volt architecture will present the next round of content opportunities: “Our low to high voltage power product portfolio spans silicon, silicon carbide, and vertical gallium nitride, which allows us to meet the challenge of direct conversion from 800 volts to 54 volts, 12 volts, and 6 volts within the AI rack.” He also said that at the OCP summit last month, the company showcased an 800 volt to 6 volt solution based on vertical gallium nitride, combined with gate drives and MOSFETs to achieve leading efficiency and power density in the most challenging conversion process of AI racks.
Industry and the “AI halo”: double-digit growth continued in the fourth quarter
Industrial revenue was US$467 million, up 10% month-on-month and 9% year-on-year. The company expects double-digit month-on-month growth in the fourth quarter. El-Khoury attributed the main reason for exceeding expectations to the so-called “AI halo” — that is, the demand for power grids, energy storage, and energy infrastructure outside of AI data centers. “Everything outside the AI data center fence is very strong, and it came better than we expected”; in addition, PMI improvements have driven industrial demand. The CFO added that the general industry accounts for about 40% of the total industrial market and is also strengthening.
Other increases in the industrial sector come from robotics and physical AI: the company says it has cooperated with more than 10 leading drone manufacturers to increase shipments of 2 million and 8 megapixel Hyperlux image sensors; in the field of humanoid robots, it has obtained new design wins in connection and perception with many leading US manufacturers; its indirect time of flight (iTOF) technology can increase depth resolution to 1 mm, up to 10 times better than competitive products. On the energy side, the company has expanded orders for hybrid EliteSiC and IGBT modules, and customers include solar power supplies.
Acquisition of Ferric raises AI data center target to over $3 billion by 2030
The company announced the acquisition of Ferric on the same day. The consideration is 600 million US dollars in cash, plus up to 175 million US dollars in contingent consideration linked to revenue milestones. The settlement is expected to be completed in the fourth quarter. Ferric focuses on integrated voltage regulators (IVRs), integrating controllers, power switches, capacitors, and magnetic components in a single package and placed close to the GPU.
Management said the acquisition will raise the 2030 AI data center revenue target by US$500 million, or 20%, to more than US$3 billion. When asked if the target increase was all due to the acquisition, El-Khoury answered, “This is related to this acquisition.” He emphasized that this is a “multi-year opportunity far beyond 2030”. It is expected to increase growth and gross margin targets, and increase EPS after scaling up, but there is no real financial impact in the near future. He also said that Ferric has entered into contact with all hyperscale cloud vendors at the design and platform architecture level. “Electrically, the transient response is superior to any solution on the market, and the mechanical Z axis height is significantly superior to the current industry level of about 2.5 mm,” while Ansemi brings scale, manufacturability, and customer coverage.
Gross margin expanded for the fifth consecutive quarter, pointing to a long-term target of 53%
The non-GAAP gross profit margin was 41.2%, up 190 basis points from month to month, and the capacity utilization rate remained flat at 83%. Trent explained that the improvement in gross margin during the quarter reflects the lagging effect of 77% utilization levels two quarters ago. The formula “every 1 percentage point increase in utilization corresponds to an improvement in gross margin of 25 to 30 basis points” still holds true, and the same applies in 2027.
He presented a breakdown of the 53% long-term gross margin target: utilization rate improvement of about 450 basis points (corresponding increase from 77% to 92% — 93%), 225 basis points of Fab Right (production line optimization), 400 basis points of improvement in the new product structure and combination, and some of the benefits brought about by price adjustments. The fourth quarter non-GAAP gross margin guideline was 41.5% to 43.5%, and the median value corresponded to an additional 130 basis points month-on-month expansion. Inventory increased by $22 million month-on-month, but the number of days fell 6 days to 186 days, of which strategic inventory fell 7 days; channel inventory fell from 10.1 weeks to 9.9 weeks.
Capital allocation: Priority deleveraging, repurchases are restricted by transactions
Operating cash flow for the quarter was US$516 million, up 23% year on year; free cash flow was US$489 million, up 31% year on year, accounting for 28% of revenue; capital expenditure of US$28 million, accounting for 1.6% of revenue. Capital expenditure for the whole year is expected to account for about 2% of revenue. Cash and short-term investments at the end of the quarter were approximately US$4.4 billion, with total liquidity of US$5.9 billion (including US$1.5 billion in undrawn revolving credit). The third-quarter repurchase was only $20 million, and Trent said the buyback activity was limited by the transaction window related to the Synaptics deal and renegotiation of its terms.
Regarding cash use, Trent said that Ferric will pay in cash on the balance sheet; the first task after the completion of the Synaptics transaction (about 5.7 billion US dollars in full cash, 123 US dollars per share, replacing the original exchange plan of about 7 billion US dollars) is deleveraging. “This is clearly the highest priority after investing in the business. We will take the opportunity to buy back, but the intention is to quickly reduce leverage.” He also said that capital expenditure intensity will return to about 5% in 2027, which is consistent with the long-term model, and “based on the current outlook or even some upward space, 2027 will not hit the upper limit of production capacity.”
Fourth quarter guidance: automotive decline, industrial and data center takeover
The company expects revenue growth of 2.5% month-on-month (median value) in the fourth quarter. Structurally, automobile revenue will drop by a high single-digit percentage, reflecting the timing and pace of continuous supply redistribution and customer delivery. It also said that the first quarter will return to normal levels as supply goes online; industrial growth is low by double digits; AI data centers grow by more than ten percentage points; and the “other” terminal market is growing by more than ten percentage points, mainly due to the seasonal nature of the wireless business and a major customer project. This “other” project is manufactured on an external contract basis, so it does not use internal production capacity. At the same time, management said that based on current visibility, momentum will continue until 2027, revenue will continue to grow, and gross margin will continue to expand.
Here are the details of Ansemi Semiconductor's third quarter conference call.
Conference: Onsemi Semiconductor (Onsemi) 2026 Third Quarter Results Conference Call
Time: November 2, 2026 at 5:00 p.m. EST
Participating management
· Parag Agarwal, Vice President of Corporate Development and Investor Relations
· Hassane El-Khoury, President and CEO
· Thad Trent, Executive Vice President and Chief Financial Officer
ASK AN ANALYST
· Vivek Arya, Bank of America Securities Research Division
· Blayne Curtis, Jefferies Research Group
· Joseph Quatrochi, Wells Fargo Research Division
· Thomas O'Malley, Barclays Research Division
· Gaurav Saini, StoneX (asked on behalf of Gary Mobley)
· Christopher Caso, Wolfe Research Research Division
· James Schneider, Goldman Sachs Research Group
· Tyler Bomba, Robert W. Baird Research Group (asked on behalf of Tristan Gerra)
· Tore Svanberg, Stifel Research Group
Moderator: Hello, thank you for waiting. Welcome to Ansemi Semiconductor's 2026 third quarter results conference call. (Operator Reminder) Please note that this session is being recorded. I now leave the meeting to today's speaker, Parag Agarwal, VP of Corporate Development and Investor Relations. Please get started.
I. Executive speeches
Parag Agarwal (VP of Corporate Development and Investor Relations): Thank you Daniel. Good afternoon. Thank you all for attending Ansemi's third quarter results conference call. With me today are President and CEO Hassane El-Khoury and CFO Thad Trent. The conference call is being webcast on the investor relations section of the company's website at www.onsemi.com. The replay of the webcast and the third quarter results announcement will be made available on the company's website about 1 hour after the conference. The recorded content will be accessible for about 30 days after the conference.
More information has been posted in the Investor Relations section of the company's website. Our results announcement and this presentation include a number of non-GAAP financial measures. The reconciliation between these non-GAAP measures and the most directly comparable GAAP indicators, as well as discussions on some limitations when using non-GAAP measures, are included in our results announcement, which has been posted separately in the investor relations section of the company's website.
In this call, we will make predictions or other forward-looking statements about future events or the company's future financial performance. We caution that such statements are subject to risk and uncertainty, and may cause actual events or results to differ materially from predictions. Significant factors that may affect our business, including factors that may cause actual results to differ materially from forward-looking statements, are described in the company's recently submitted Form 10-K, Form 10-Q, and other documents submitted to the U.S. Securities and Exchange Commission, as well as in the third quarter results announcement. Our estimates or other forward-looking statements are subject to change, and we are under no obligation to update these forward-looking statements to reflect actual results, changes in assumptions, or other events, except as required by law. Now I'll give my time to Hassane. Hassane?
Hassane El-Khoury (President and CEO): Thank you Parag and thank you all for participating this afternoon. Our third-quarter results reflect the continued strength of our business and the momentum we are building up in strategic growth markets — growth is higher than seasonal trends, AI data centers continue to accelerate, and industrial demand improves. Where we see particularly strong demand is where power density is a key differentiator: from AI data centers and the grid infrastructure they support, to energy storage and next-generation vehicle architectures. These are the strategic markets we have concentrated our investment in, and this strategy is now translating into revenue growth, increased gross margin, increased share, and new customer orders.
AI data centers continue to be our fastest-growing market, expanding faster than we anticipated at the beginning of the year, and new orders are already climbing. In the automotive sector, we continue to increase bike content and target new platform orders in various regions; industrial demand continues to improve, driven by power infrastructure and energy applications.
In the third quarter, we achieved revenue of $1,722 million, non-GAAP gross margin of approximately 41.2%, and non-GAAP diluted earnings of $0.90 per share, all above the guideline median value. The combination of stronger demand, structural improvements, and strict implementation is driving significant operating leverage in our model, where profit growth has greatly exceeded revenue growth.
Some of our technology segments are still limited in supply, so in the third quarter and into the fourth quarter, we made a short-term decision: allocating supply to markets with strong demand for terminals, such as AI data centers and industrial—these two markets performed far better than our original plans. In the automotive sector, we are supporting the delivery of all terminal requirements, but we are not supporting backlog orders for inventory reimbursement. As supply and demand are rebalanced, we expect automotive growth in 2027 to be in line with long-term goals.
In AI data centers, our revenue covers the complete power link, from the power supply unit (PSU) to the low-voltage intelligent power level close to the processor — increasing computing power is bringing higher power density requirements and higher semiconductor content. In the third quarter, our AI data center revenue increased 39% month-on-month to US$180 million. We now expect the business to achieve revenue of approximately US$620 million in 2026, an increase of 140% year over year. For 2027, we still expect AI data center revenue to double from 2026, as new products continue to be shipped.
In the third quarter, we began bulk shipments to several iconic platforms, including Google — supplying both silicon carbide and silicon MOSFETs for power supply units as well as 54 volt to 12 volt applications. The move to an 800-volt architecture has created the next content opportunity for Ansemi to bring AI data center power supply directly into our field of strength. Our complete power product portfolio, covering low to high voltage, spans silicon, silicon carbide, and vertical gallium nitride, which remains our competitive advantage, enabling us to meet the demand for direct conversion from 800 volts to 54 volts, 12 volts, and 6 volts within AI racks.
As power systems become more complex, customers increasingly need solutions that go beyond a single device. At the OCP summit last month, we showcased an 800 volt to 6 volt solution based on differentiated vertical gallium nitride technology. Coupled with gate drives and MOSFETs, we achieved industry-leading efficiency and world-leading power density in the most challenging conversion process of AI racks.
The shift to 800 volt HVDC has also accelerated the shift from mechanical to solid state power switches for fast shutdowns — from solid state transformers and circuit breakers to hot plug applications. Our silicon carbide JFET product portfolio has excellent thermal performance, has been designed and is contributing to data center revenue.
On Investor Day, we talked about investing in building differentiation capabilities before these market inflection points hit. As power density becomes more critical in AI data centers, automotive, and industrial sectors, we are continuing to do so. One example is our EPP, an embedded power platform, which we launched on Investor Day. EPP uses the silicon wafer itself as a package while addressing electrical, thermal, and mechanical constraints. Since its launch, we have been encouraged by customer interest in the platform, further strengthening our judgment: power density is becoming an increasingly important differentiator in automotive, AI data centers, and industrial applications. EPP has sent samples to the first (alpha) customers in these markets and is manufactured using our 12-inch silicon process in East Fishkill.
Today, we are also pleased to announce the acquisition of Ferric, which will further extend our capabilities to integrated voltage regulators (IVR) and put us in a leading position to power the next generation of AI processors. As processor power requirements continue to rise, IVR integrates a complete power regulator — including the controller, power switch, capacitor, and magnetic components — in a package and places it closer to the GPU, thereby reducing power loss and solving the power density problem at the core. Ferric has the most competitive IVR solution in the industry. Its efficiency and transient response are industry-leading, the package surface is compact, and the Z-axis height is the best in its class. We expect Ferric to increase our market share in the AI data center power link core stage (core stage). Although this is a multi-year opportunity for Ansemi that far exceeds 2030, we are still raising our 2030 AI data center revenue target by $500 million to over $3 billion.
Next, let's talk about industrial business. The spillover effects of AI are generating momentum outside of data centers. We have expanded orders for hybrid EliteSiC and IGBT modules to include leading customers such as Sunlight Power, thanks to the 99.3% efficiency platform we launched last quarter. Also in the industrial business, we are seeing increased momentum in the fields of robotics and physical AI — an area where our power and perception capabilities complement Synaptics highly. Demand for first-person view (FPV) drones is driving increased shipments of our 2 million and 8 megapixel Hyperlux image sensors, and we have partnered with more than 10 leading drone manufacturers. We continue to diversify the robot market and obtain new design orders with many leading American humanoid robot manufacturers in terms of connectivity and perception. In terms of perception, we are advancing indirect time of flight (iTOF) technology to achieve world-class depth resolution of 1 mm, up to 10 times better than competitive solutions, making it possible for robots to accurately sense steps — which is critical for robots to interact with the physical world. As physical AI expands, these applications bring together many of our core capabilities in power and perception, creating significant new content opportunities for Ansemi.
Turn to the car. As the industry evolves towards electrification, higher levels of autonomous driving, and software-defined vehicles, we continue to win new customer designs in all regions. Our investment in new products continues to expand the semiconductor content for bicycles, up to $2,000, as the architecture evolves. Even with SAAR being relatively flat, these shifts have created multiple content growth vectors for Ansemi. We continue to increase our share of the electric vehicle market in North America, making progress in traction inverter design and new on-board charger platform orders, and the Treo-based 10Base-T1S that we have already launched is climbing the slope. On the European OEM side, we expanded design orders across our power product portfolio as they began their transition to 48-volt and regional architectures. In China, we continue to expand our leading position in high-end OEM 900-volt electric vehicles, winning orders for silicon carbide traction inverters for NIO's flagship sedan and Geely's rapidly expanding export models.
Looking ahead, we are more and more confident about our business trajectory. AI data centers continue to expand rapidly, automotive and industrial demand improves, and content growth and new platform orders form multiple growth vectors. All of these trends are converging on Ansemi's differentiated fields — intelligent power and perception — and our strict execution has enabled us to turn this growth into stronger profit and cash flow generation. I would like to thank our employees around the world for their continued focus and execution. It's their work that enables us to serve our customers and create value for our shareholders. Now I'll leave the time to Thad and ask him to elaborate on the results and fourth quarter guidance.
Thad Trent (CFO): Thanks Hassane. We believe the recovery is gathering momentum, and we are now fundamentally stronger than in previous quarters. We have more content, broader market opportunities, and significantly stronger profitability. With most of the heavy investment over, we have the ability to turn revenue growth into profit and free cash flow growth.
In the third quarter, we continued to see this benefit in financial results. Third-quarter revenue of US$1,722 million was higher than the guideline median value, up 7% from month to month. Non-GAAP gross margin increased 190 basis points month-on-month to 41.2%, and non-GAAP earnings per share were $0.90. The year-over-year growth rate was four times the revenue growth rate.
On Investor Day, we talked about the company continuing to invest before major market inflection points, build endogenous capabilities, and supplement it through strategic mergers and acquisitions to speed up time to market. AI data centers are a clear example. As the architecture evolves, we continue to enhance our capabilities in all aspects of the power link, enabling Ansemi to seize more opportunities.
Today, we announced the acquisition of Ferric for $600 million, plus up to $175 million in contingent consideration linked to revenue milestones. As Hassane said, Ferric puts us in the lead in powering the next generation of AI processors. We expect Ferric to benefit our growth and gross margin targets, increase EPS after scaling up, and have no real financial impact in the near future. This is a multi-year opportunity to raise the 2030 AI data center revenue target by 20% to over $3 billion, and continue to accelerate thereafter. We expect the deal to close in the fourth quarter, subject to customary delivery conditions. More details about the acquisition are available on the company's investor relations website.
Back to Q3 earnings. Revenue of US$1,722 million, up 7% month-on-month and 11% year-on-year, was US$22 million higher than the median value of guidance. The increase came from industrial and AI data centers. Vehicle revenue was $777 million, down 1% month-on-month and year-over-year. The month-on-month decline was mainly due to the redistribution of supply to the AI data center and industrial terminal markets. We expect automotive revenue to drop by a high single-digit percentage in the fourth quarter, reflecting the timing and pace of continued supply redistribution and customer deliveries. As supply comes online, it is expected to return to normal levels in the first quarter. Industrial revenue was US$467 million, up 10% month-on-month and 9% year-on-year. We're seeing strength in energy infrastructure, factory automation, and the broader industrial market. We expect a low double-digit percentage increase in industrial revenue in the fourth quarter. AI data center revenue was US$180 million, up 39% month-on-month and 175% year-over-year. Demand continues to accelerate. We now expect AI data center revenue to reach about US$620 million in 2026, an increase of more than ten percentage points in the fourth quarter. In 2027, we expect to double the higher base.
Note that starting this quarter, we are increasing disclosure transparency by disclosing AI data centers separately as the fourth end market to better reflect this important growth driver. You will find it in supplementary materials, on the Investor Relations website, and in the 10-Q form.
Revenue from other terminal markets (now excluding AI data centers) was US$299 million, up 11% month-on-month and 10% year-on-year, mainly due to the normal seasonality of wireless services. We expect revenue from other businesses to increase by more than ten percentage points in the fourth quarter compared to the previous quarter.
Look at the division of business units. The Power Solutions Business (PSG) revenue was $890 million, up 7% month-on-month and 21% year-over-year. The analog and mixed signal business (AMG) revenue was US$597 million, up 9% month-on-month and 2% year-over-year. Intelligent Sensing Business (ISG) revenue was US$235 million, up 3% month-on-month and 2% year-over-year.
Talk about gross profit margin. GAAP gross margin was 41.4%, and non-GAAP gross margin expanded to 41.2% for the fifth consecutive quarter due to improved utilization since the beginning of the year and the favorable structure brought about by new products. The utilization rate is flat at 83%, and we expect the fourth quarter to be at a similar level and increase as demand signals improve after entering 2027.
Our Fab Right (Production Line Optimization) strategy remains an important contributor to long-term gross margin opportunities. In the third quarter, we completed the sale and delivery of a previously announced back-end factory, and we are still planning to complete the sale of the Mountain Top plant by the end of 2027. We are operating with a more focused and efficient capacity layout, creating additional opportunities for long-term expansion of gross margins.
GAAP operating expenses were US$335 million, and non-GAAP operating expenses were approximately US$310 million. GAAP operating profit of USD 378 million, accounting for 22% of revenue; non-GAAP operating profit of USD 400 million, accounting for 23% of revenue. GAAP net profit was US$314 million, and non-GAAP net profit was approximately US$354 million. GAAP tax rate 21%, non-GAAP rate 15%. GAAP shares are approximately 397 million shares, corresponding to GAAP earnings of $0.79 per share. The number of non-GAAP shares is approximately 392 million shares, which corresponds to non-GAAP earnings of $0.90 per share.
Go to the balance sheet. Cash and short-term investments at the end of the quarter were approximately US$4.4 billion. Total liquidity was $5.9 billion, including $1.5 billion in undrawn revolving credit. Operating cash flow was $516 million and free cash flow was $489 million, accounting for 28% of revenue. Capital expenditure of US$28 million represents 1.6% of revenue. We expect capital expenditure to account for about 2% of revenue for the full year. Inventory increased by $22 million month-on-month, but declined by 6 to 186 days in terms of days. Strategic stocks continued to be digested in the third quarter, falling for 7 days; basic inventory increased by 1 day as we continued to support higher demand. Distribution channel inventory weeks fell from 10.1 weeks in the second quarter to 9.9 weeks, and sales continued to support healthy channel and mass market growth. We repurchased $20 million of shares in the third quarter, and repurchase activity was limited by transaction restrictions related to the Synaptics acquisition and renegotiation.
Looking ahead, let me explain the key elements of the fourth quarter non-GAAP guidance. As a reminder, today's results announcement includes a table detailing GAAP and non-GAAP guidance. We expect fourth-quarter revenue to be between $1,715 billion and $1,815 million, up 2.5% month-on-month in median terms, which is a higher than seasonal increase. We expect non-GAAP gross margin of between 41.5% and 43.5%, which includes approximately $8 million in equity incentive expenses. Calculated at the median value, this meant an additional 130 basis points of expansion over the previous month. Non-GAAP operating expenses are estimated to be between $303 million and $318 million, including approximately $34 million in equity incentives. We expect other non-GAAP earnings of approximately $27 million. We expect the non-GAAP tax rate to remain at 15%, and the number of non-GAAP diluted shares will be around 392 million. From this, non-GAAP earnings per share are between $0.95 and $1.07, and capital expenditure is expected to be between $40 million and $50 million.
The fourth-quarter outlook reflects continued demand momentum and another step forward in our model. Based on current visibility, we expect this momentum to continue into 2027, with revenue continuing to grow and gross margin to expand as we scale our differentiated products and achieve the benefits of a more efficient operating model. Looking ahead, the opportunity is not just a recovery in demand. We have built a stronger business with significantly greater profit and cash flow potential. Our product portfolio is more focused, our manufacturing layout is more efficient, and our growth path is more capital-efficient. As we bid farewell to this year, our priorities are clear: execute on immediate needs, scale up our differentiated platform, and seize the huge market opportunities ahead. We believe the company is in a good position to achieve continuous growth and gross margin expansion to create long-term value for shareholders. Now I'm returning the meeting to Daniel, and I'm open to questions.
II. Q & A session
Moderator: Our first question came from Bank of America Securities's Vivek Arya.
Vivek Arya (Bank of America Securities Research): Hassane, you seem to have correctly downweighed the automotive business in the second and third quarters. I think the fourth quarter was also the case; judging from what you just said, it was a high single-digit decline. But then you said that next year cars will return to their long-term target growth trajectory. I understand it is about a single digit or 10%. What I'm curious about is, in the long run, will insufficient supply in that market have an impact on market share? Also, the market will grow again next year. Where will the additional supply come from?
Hassane El-Khoury: Yes. First, I'd like to correct one statement you used. We're not reducing the weight of the car market. Automobiles are still strategically important as an end market. Even in the five-year outlook we emphasized on Investor Day, automobiles are still the largest market, and this will not change. What you're seeing this quarter and next, and what we're implying, is short-term, supply-driven, not strategy-driven. I mean, we're not losing our share because we're mostly working with OEMs, and as I mentioned in my preparation statement, we're supporting all of their terminal needs; what we're suppressing — also after communicating and aligning with them — is inventory recovery that they would have hoped to do within a few quarters. We'll resume shipping in the first quarter, and as supply goes online — supply is always online or increasing due to our ever-limited technology — but you've seen results from industrial and AI data centers have even surpassed our guidelines and expectations. So we decided to adjust the structure a little bit in the short term, but that's not a strategic decision. Our strategy is in line with the emphasis on Investor Day. This is why we are reaffirming a return to growth in 2027, which in itself indicates that no share has been lost. This is simply inventory replenishment, not a terminal demand issue, and is short-term in nature.
Vivek Arya: Understood. My question is, can Thad take us through the addition and subtraction of gross margin volume, structure, and utilization ratio from the third quarter to the fourth quarter. Then conceptually speaking, if automobiles and industries grow at the long-term growth rate you set next year, and data centers are likely to double, how should we consider the potential for gross margin expansion in 2027?
Thad Trent: OK, Vivek, our gross margin derivation framework hasn't changed. The biggest driver of gross margin expansion in the fourth quarter was improved utilization. Note that this is a reflection of a round of improved utilization two quarters ago. So you'll start to see its effects. You'll also begin to see the effects of some price adjustments — previously rising investment costs had headwinds and were reflected in profit and loss, but now we are beginning to see the positive impact of price adjustments. However, the formula for each 1 percentage point increase in utilization corresponds to an improvement in gross margin of 25 to 30 basis points still holds true. It was also established in 2027. So as I mentioned in my preparation statement, we expect continued growth, and we expect utilization rates to continue to improve after entering 2027, which will continue to expand gross margins throughout the year. Obviously, revenue growth will help, too. Another factor that's really starting to work is the new product. As a result, we obtained a favorable structure. As already seen in the third quarter, new products began to climb with favorable gross margins. So this will be a nice welcome start into next year. In addition, there is Fab Right. The announced back-end factory sales will be reflected in 2027, and the Mountain Top factory will be reflected more in 2028. This is also a good trend moving forward.
Moderator: The next question comes from Jefferies' Blayne Curtis.
Blayne Curtis (Jefferies Research): I'd like to continue with the previous question. I'm assuming cars would have grown in the December quarter, but now it's falling by a high single digit. I'm trying to figure out how many restocking may have occurred in the December quarter. Also, I'm curious if you see these trends increasing.
Hassane El-Khoury: Yes. The trend is strengthening. So there will be — you can understand that there will be a catch up, because this part of inventory replenishment is not really a terminal requirement. We are supporting terminal requirements, and this has not changed. So on top of that, you can assume some seasonality. However, from the perspective of inventory replenishment, demand will not disappear. So you'll see it resume in the first quarter as supply goes live. It's just a matter of time — you can understand it as a question of the quarter's timing. We have no guidance for the first quarter. But once you get to the first quarter, you'll see that some of this should have happened in the fourth quarter, if normalized. I've always said that in automobiles, and even in the many markets we serve, I don't look at the quarter-on-quarter comparison, nor do I interpret it too much. I look at trends and see what decisions we're making. So I had zero concerns about the decisions we made in the car. These are the right decisions for the company. To service the growing demand for industrial and AI data centers that we have delivered in the third quarter and will continue to be delivered in the fourth quarter, then we will resume normal supply and demand from 2027, and you will see us achieve our long-term goals for the rest of 2027.
Blayne Curtis: The data center business will double again in 2027, which is quite impressive. I was wondering if you could give some rough drivers.
Hassane El-Khoury: Yes. The general context is that across the entire power link — I've been saying for a few quarters that we get design orders for the entire power link from high voltage to the core — these have already begun to land and generate revenue. We've gone from design to design orders and now to revenue. In terms of customer breadth, we've emphasized some customers in the past. Some customers can roughly guess who they are; moreover, they span the power link, not only through our direct cooperation with computing power vendors and design at the board level, but also through the supply chain, such as the power supply unit side. We've mentioned it on several phone calls in the past. All of these are maturing. We started climbing more projects in the third quarter, as hinted at on Investor Day. So these have all paid off. This is really execution, execution, and re-execution. But more importantly, the new products we've launched in the past two years are now being delivered. This climb will continue in 2027, and we will drive some new stuff to revenue in 2027, and now we are getting design orders. To emphasize, the 2027 doubling is on top of the higher base we just talked about today. So 2027 is even higher than what we expected 90 days ago.
Moderator: The next question comes from Wells Fargo's Joseph Quatrochi.
Joseph Quatrochi (Wells Fargo Research): I'd like to ask about this. You raised your 2030 revenue target by $500 million. Is this all due to this acquisition? Or does it also include an increase in the base figure itself 90 days ago?
Hassane El-Khoury: This relates to this acquisition.
Joseph Quatrochi: Okay. One question, considering the current cash expenses of this acquisition, and Synaptics, you are also borrowing some money for this. How should we consider the use of free cash flow and return on capital?
Thad Trent: We'll use cash on our balance sheet to buy Ferric. Once the Synaptics transaction is complete, we will focus on deleveraging. That will be our top priority. Obviously, this was after investing in the business. But we'll focus on that. We will take the opportunity to buy back our shares. But the intention here is to quickly reduce leverage.
Moderator: The next question comes from Barclays Tom O'Malley.
Thomas O'Malley (Barclays Research): My first question is about industry. The September quarter looked significantly more than seasonal, and then the December quarter looked like very strong results. Are there any vertical fields in industry that can be named to perform better? The numbers for the second half of the year were quite impressive. I'd like to start here.
Hassane El-Khoury: Okay. What I want to name is what we call the AI aura, from energy — understood as power grids to AI data centers. So everything outside of the AI data center fence is very strong. It's better than we expected. PMI is also improving, so general industry is good too. However, if you ask me the reason for exceeding expectations and improving our outlook, the main reason is that the AI aura we emphasized on Investor Day is being realized as expected, and even higher in the short term. This is also the reason I mentioned that we should prioritize serving this market and rank ahead of other markets.
Thad Trent: Yes. The general industry accounts for about 40% of the total industrial market, and we see strength there as well. I think the PMI is improving as you can see.
Thomas O'Malley: Understood. Thad, second question, you mentioned that the weak car revenue is due to supply redistribution and the timing and pace of customer delivery. However, the utilization rate appears to be flat. Are you lacking external components? Because if you don't get more supply, I thought the utilization rate would continue to rise, and you said it would be until next year. Can you describe this dynamic and where exactly are the restrictions?
Thad Trent: Okay. Keep in mind that what you see from the profit statement today is the utilization rate two quarters ago, when we were running at 77%. We were running at 83% last quarter and still 83% in the fourth quarter. So this is where we're catching up with that part of the supply because the cycle time from the fab to the back end needs to run out. This is where the supply is online. Supply constraints are concentrated on specific production lines in specific regions. So it's not comprehensive; otherwise, we would obviously raise the utilization rate even higher. However, when we look at next year and see that demand in various terminal markets continues to strengthen, we will raise the utilization rate.
MODERATOR: The next question comes from Gary Mobley from STONEX.
Gaurav Saini (StoneX, asked on behalf of Gary Mobley): I am Gaurav Saini, Gary Mobley from STONEX. I would like to ask specifically about utilization leverage in gross margin estimation: Of the 53% gross margin target path you set, about half comes from utilization. So, considering changes in restructuring and depreciation, how should we consider the incremental gross profit margin per dollar of revenue from now on?
Thad Trent: Looking at it this way, the third-quarter results included about 450 basis points of underutilization of production capacity. Again, this goes back to 77% utilization in the first quarter. If you calculate this, it rises to about 92% or 93%, which is roughly the peak for us, and you use the 25-30 basis point formula to get 450 basis points. So to deduce for you that from now to 53%, you have about 450 basis points of utilization improvement, about 225 basis points of Fab Right (I talked about the steps we took there), and about 400 basis points from structural improvements and new product volume — I've already said that we're starting to see this reflected in profit and loss, and we'll see more next year. The final component is price increases, which are not fully reflected in profit and loss. So there was a bit of a smooth breeze there too. Taken together, this is roughly close to the long-term goal of 53%.
Gaurav Saini: Understood. You guys just talked about the 2027 new product climb. I'd like to ask more about the contributions we can see in the next fiscal year. How should we consider the contributions of vertical gallium nitride, EPP, and TreO revenue layers that didn't exist before?
Hassane El-Khoury: What you'll see next is obviously Treo continuing to climb the hill. Treo is already contributing revenue this year. We've already talked about vertical gallium nitride contributing revenue in 2027. Then add EPP on top of that, a few later ones. With the exception of Treo, which started climbing last year, several other projects started from a very small base at the time, and we have more than doubled our growth as we move forward. The rest will start climbing and following a similar trajectory. Therefore, you can understand the 2027 base as: new products that already exist in 2025 and 2026 will begin to make substantial contributions, and our newly launched EPP and vertical gallium nitride will begin in 2027 and continue to move forward, eventually forming a more mature new product structure on a timeline close to 2030, depending on whether it is automotive or industrial.
Moderator: The next question comes from Chris Caso of Wolfe Research.
Christopher Caso (Wolfe Research): The first question is about Ferric's acquisition. Can you talk about what stage this acquisition is currently at in terms of revenue and design orders? I know you used it to raise your 2030 goals. But looking at 2027 and 2028, what can we expect from this business?
Hassane El-Khoury: No, I'm not giving this acquisition or revenue guidance for 2027 right now. But I want to tell you that in our due diligence, it's clear that we know this market ourselves, and everyone knows the evolution trajectory of IVR — it's a solution after processors are getting more and more powerful. This is the first point. We know the criticality of integration, power density, and Z-axis height, and we also know that Ferric is involved with all hyperscale cloud vendors at the design level and platform architecture level. This comes from their skills and abilities. What we bring is scale, manufacturability, and our breadth of coverage across hyperscale cloud vendors and computing power platforms. This combination makes it strategic. Judging from their contacts, they are well-positioned to work with all hyperscale cloud vendors, based on detailed adjustments made by our team. Of course, this also brings a highly qualified and talented team that complements our internal work, which is a win-win situation.
Christopher Caso: Understood. I'm asking a question about capital expenditure and next year's production capacity. If I understand correctly, part of the supply constraints you are facing now is only because they didn't start wafer casting earlier at the beginning of the year, and now they are catching up. If so, where are you in terms of production capacity as we enter next year? Do you think next year's production capacity will be enough to meet the market conditions you are seeing? So what kind of capital expenditure should we expect next year?
Thad Trent: Yes, Chris, I think this description is accurate; this is a catch up, which is why Hassane said it was temporary. We'll catch up in the first quarter. So based on our outlook today, and even some upside above the outlook, we have the production capacity we need. So I think we're far from hitting the upper limit of production capacity as we enter 2027. Our capital expenditure intensity is about 2% of revenue this year, and I expect it to be around 5% next year, which is in line with our long-term model. This will allow us not only to invest in maintenance capital, but also to increase production capacity where needed and on specific production lines that can obtain higher output. However, we believe that for a market that will grow in 2027 and beyond, we are in a very good position in terms of our online production capacity.
Hassane El-Khoury: Yes. And the production capacity we talked about, which will help catch up in the first quarter, can be understood as production capacity is already behind, and investment is already behind. It's just going to go live and be delivered and then implemented until 2027. So it's not something we want to look forward to, it's what we've done; we're just reaping the benefits from a time lag perspective.
Moderator: The next question comes from Jim Schneider of Goldman Sachs.
James Schneider (Goldman Sachs Research): Hassane, I'd like to ask you to describe the behavior of OEM customers when they are willing to restock their inventory. I remember you mentioned in your preparation statement that automotive customers only support an increase in terminal demand, not inventory replenishment. Can you roughly describe how these customers and the broader OEM customer base are putting more inventory on their balance sheets?
Hassane El-Khoury: Okay. First, let me explain that this is specific to OEMs. Not all OEMs act the same. I can only talk about the customers we need and are in touch with. They can clearly see that AI and industry are recovering, so they are trying to lock in demand ahead of time to avoid “traffic jams” in terms of production capacity and capacity constraints, because what they see is not only power semiconductors, but also simulation, computing power, and memory. So they're looking at the 2027 plan and hope for something called a BCP (Business Continuity Plan) in our industry. These are strategic plans we have drawn up with strategic OEMs. We sat down to clearly calculate their production capacity and delivery pace, and we prepare the goods in advance based on this. This is the normal standard process for replenishing inventory prior to climbing a hill. That's why we have these communications — we're not going to build them all in the 3rd and 4th quarters; it's our decision. We will continue to ship according to terminal demand. We know that their backlog orders are now on account, and we will fulfill them, which is why we have built a new production capacity model based on this. So it's temporary, short-term, and we'll catch up in 2027.
James Schneider: One question. Your remaining subcategories seem to be growing very well, even after listing data centers separately. Can you talk about what drives this “other” category? Can you continue this growth rate until 2027, and why can you or can't?
Thad Trent: In this “other” category, we participated on an opportunistic basis. In preparation for the presentation, we mentioned that the third quarter was seasonal for the wireless terminal market in this category. Looking ahead to the fourth quarter, we have a major customer climbing the hill for a project. So it fluctuates a bit. I think this is the reason why the fourth quarter's increase reached a high level of about ten percent. Entering next year, no, I don't think it will grow at that rate. I think you'll see AI data centers, automobiles, and industries grow at a much faster rate, and this one will be closer to normal historical levels because we'll be focusing on other markets. Incidentally, the fourth quarter hill climb I mentioned was manufactured on an external OEM basis. So you won't see the impact of [internal production capacity]. That's why we're able to support it.
Moderator: The next question comes from Baird's Tristan Gerra.
Tyler Bomba (Robert W. Baird Research Group, asked on behalf of Tristan Gerra): I'm Tyler Bomba, asking questions on behalf of Tristan Gerra. Is inventory recovery included in the strength of the industry? You said cars are not allowed to be restocked. I'm curious about the state of the industry?
Hassane El-Khoury: No. The strength of the industry really comes from the AI aura I mentioned in my previous answer; what we saw was that terminal products were shipped on the market. So we can see sales — not only what we sell to our customers, but also what our customers sell to their end customers. If you look at the large industrial companies that do “grid-to-data centers”, all of which have already announced results, have performed very well in the quarter, we can also see this and see what they are building. That's why when we configure the supply, we say that this is a real requirement, something that is being deployed at the site. So this isn't inventory reimbursement; it's simply an increase in demand, and it's being installed on the site as quickly as possible.
Tyler Bomba: Okay. One question, how should we consider the geographical structure of silicon carbide growth in 2027, and the respective contributions of new platforms and content growth to the recovery of electric vehicle production itself?
Hassane El-Khoury: I think it's pretty broad. Of course, from the perspective of electric vehicles, China is still a strong driving force for silicon carbide. However, a major increase in new silicon carbide products also came from North America and AI data centers. Of course, the JFET I specifically mentioned in my preparation speech mainly enters AI data centers. So I would like to say that as we diversify regions and end customers in terms of electric vehicles, we have now added a very broad and diverse AI market, whether in terms of applications, customers, or regions.
Moderator: The next question comes from Stifel's Tore Svanberg.
Tore Svanberg (Stifel Research): I'd like to go back to the business continuity plan you mentioned. We all know that AI data center production capacity has been tight for a long time. I'm curious what makes car OEMs willing to restock stocks now? Does that mean their needs are really starting to accelerate, or am I misunderstanding it?
Hassane El-Khoury: No, I can't say they're speeding up, but I think there's definitely a need for them to move forward. Keep in mind that for us it's not just SAAR, it's also bike content. As you have heard in my prepared statement, Europe is moving towards a 48-volt regional architecture. This creates higher demand opportunities in terms of content, not necessarily the number of complete vehicles. The structural share of electric vehicles is increasing, which means more content. So you can understand what I'm talking about as content. When we talk about returning to the long-term model, that is, higher single-digit growth than SAAR, I mean this. This is what the OEM ordered. What they are seeing is an increase in content on the new platform, not necessarily SAAR or the number of vehicles. So I want to be very clear: for us, inventory corresponds to the content in the car.
Tore Svanberg: Very helpful. A question about Ferric. Many companies are now deploying IVR, and I hope you can share some of the uniqueness that Ferric brings. I know you mentioned better magnetic components, smaller packages, etc. in the demo. But is there anything else to share? Why is their technology the winning formula compared to other companies' solutions?
Hassane El-Khoury: Yes. If you consider what is required for this solution to work, there are three aspects: thermal, mechanical, and electrical from an IVR perspective. Electrically, we know Ferric's transient response is superior to any other solution on the market. This covers the electrical dimension. This is important because high-end processors carry a dynamic load, have a large number of transients, and the voltage must be adjusted or responded to. This provides superior electrical performance compared to peers or competitors. Then there is the mechanical dimension, that is, power density. There are other solutions on the market that are not necessarily magnetic, and there are also other innovative solutions, but Ferric's solution that integrates electrical and magnetic components with a controller is the most compact in terms of area and the most compact in terms of Z axis height. You can think that currently everyone's Z axis height is at the level of 2.5 mm, and Ferric far exceeds that. This is extremely important when you're talking about a truly integrated IVR. Finally, heat. With the efficiency brought by the first two, heat is much less of a problem for GPU or computing power vendors. Of these three aspects, if you draw a Venn diagram, Ferric is right in the middle. That's why we love it, and why we want this asset instead of comparing it to other solutions.
Moderator: This concludes the questioning session. I will now return the meeting to President and CEO Hassane El-Khoury to make a closing statement.
Hassane El-Khoury: Thanks for the questions. We focus on areas where challenges are getting harder and where our differentiation is most important. Power density has become critical in AI data centers, automotive, and industry, and our investment in high-voltage technology, Treo-based solutions, and EPP enables us to address these challenges at the device, platform, and system levels. Combined with the leverage in our operating model, we believe this will enable us to drive profitable growth, stronger cash flow generation, and long-term shareholder value. Thank you all for participating today.
Moderator: This concludes this call. Thank you all for participating, but you can hang up now.