FormFactor (FORM) 2026 年第三季法說會：營收創 2.805 億美元新高

重點總覽

• FormFactor 公布連續第四個季度營收創歷史新高，達 2.805 億美元，季增 9%、年增將近 40%，超越其 2.6 億至 2.8 億美元財測展望的高標。
• Non-GAAP 毛利率達 56.5%，較第二季上升 320 個基點。其中約 1,050 萬美元（即 380 個基點）主要來自關稅退稅及有利的 Quantum 子元件成本，管理層預期這些因素不會常態重演。
• Non-GAAP 稀釋每股盈餘從第二季的 0.82 美元增至 1.04 美元。第三季每股盈餘中約 0.09 美元歸因於關稅退稅。
• 在 CPU 及更廣泛的高效能運算需求支撐下，晶圓代工與邏輯業務營收季增 10%。受惠於 Triton 光子系統銷售增加，系統業務營收成長 23%，達到創紀錄的 5,950 萬美元。
• 管理層預估第四季營收為 2.85 億美元（上下浮動 1,000 萬美元），Non-GAAP 毛利率為 54%（上下浮動 150 個基點），Non-GAAP 稀釋每股盈餘為 0.95 美元（上下浮動 0.09 美元）。
• Farmers Branch 廠已出貨首個生產單位，預計到 2027 年底年產能將達到約 4 億美元的探針卡生產能力。管理層強調，這代表的是產能規模，而非保證營收。

關鍵財務數據

管理層目前認為在第三季的營收水準下，基準 Non-GAAP 毛利率約為 53%。這項改善反映了週期時間、良率、人力表現、製造支出與生產效率的提升。

業務與營運表現

晶圓代工與邏輯

晶圓代工與邏輯業務營收季增 10%，管理層預期第四季將進一步成長。需求廣泛分布於個人電腦 (PC)、行動裝置、資料中心 CPU 及網路領域，而 GPU 和超大型雲端業者客製化 ASIC 正成為新的成長驅動力。

在第三季，FormFactor 向全球領先的晶圓代工廠出貨了用於單一 GPU 設計的探針卡，金額達數百萬美元。管理層表示，在 Farmers Branch 廠提供額外產能的前提下，公司已具備在 2027 年擴大 GPU 及客製化 ASIC 設計市占率的優勢地位。

資料中心 CPU 需求是另一個重要貢獻來源。管理層提到公司的業務覆蓋了長期合作的客戶關係、x86 無廠半導體供應商及 ARM CPU 應用。

DRAM 與 HBM

DRAM 探針卡營收維持在歷史高點，與上季基本持平。隨著客戶調整晶圓投片以獲取更高的 DDR 市場價格，產品組合轉向 DDR5 並減少了 HBM。

儘管產品組合較不理想，探針卡毛利率仍上升約 90 個基點至 55.3%。管理層將此改善主要歸因於產能利用率、良率、支出紀律及製造成本控制，而非普遍提價。

公司繼續將 HBM 視為貫穿 2026 年至 2027 年的主要成長驅動力。管理層表示正致力於早期 HBM4E 與 HBM5 設計，目前未看到 HBM 去堆疊帶來重大成長阻力。

系統與光子學

在共封裝光學 (CPO) 對 Triton 需求加速的推動下，系統業務創下季度營收歷史新高。FormFactor 目前預計 2026 年 Triton 營收將超過 3,000 萬美元，高於最初預期的 1,000 萬至 2,000 萬美元。Triton 在 2025 年未產生任何營收。

公司也在佈局近封裝光學 (NPO) 與可插拔應用。管理層表示，Triton 採用較多外包的製造模式，應可避免產能限制預期的 2027 年成長。

Farmers Branch 廠擴建

FormFactor 已為其近 30 萬平方英尺的 Farmers Branch 工廠通過組裝與測試部分的認證，並比原先預期更早出貨首個生產單位。公司計劃在 2027 年期間新增用於晶圓代工與邏輯產品的 MEMS 前段製程。

預計產能將在 2027 年期間呈大致線性爬升，在年底達到每年約 4 億美元的探針卡生產能力。將晶圓代工與邏輯生產轉移至 Farmers Branch，預計也將釋放 Livermore 廠區的 DRAM 產能。

FormFactor 將 Farmers Branch 廠 2026 年預期現金資本支出從 1.4 億至 1.7 億美元調降至 1 億至 1.1 億美元。管理層表示，這反映的是付款時間點，而非計畫範疇或爬升速度的改變，現金支出將遞延至 2027 年。

公司還揭露了來自泰瑞達 (Teradyne) 與擴大合作相關的 3 億美元投資。管理層表示，該項投資強化了資產負債表，但並未改變 FormFactor 的資本配置優先順序。

管理層財務展望

第四季毛利率預計將受惠於約 200 萬美元的剩餘關稅退稅、產品組合改善及出貨量提升。在收齊退稅後，管理層預計基準毛利率將不再享有關稅退稅效益。

第四季營收展望的中位數將使 FormFactor 有望在 2026 年首次實現全年營收突破 10 億美元。公司重申其長期目標模型：營收達 16 億美元、毛利率達 55% 以及 Non-GAAP 每股盈餘達 5 美元。

風險與關注焦點

• FormFactor 現有製造廠區的出貨產能已接近極限，這使得 Farmers Branch 廠產能爬升的時間點與執行狀況對後續成長至關重要。
• 新廠效率較低（包括招聘與啟動支出）可能會在 2027 年對毛利率造成壓力。管理層預期更高的出貨量與營運效率將抵消這些影響，使毛利率維持在目前基準附近。
• 第三季毛利率中約 1,050 萬美元的效益來自管理層預期不會常態重演的項目，包括關稅退稅與有利的 Quantum 子元件成本。
• 在位元供給吃緊的 DRAM 市場狀況持續期間，DRAM 探針卡產品組合可能會繼續在 DDR 和 HBM 之間動態調整。
• 共封裝光學的需求目前仍集中於單一生態系，不過管理層觀察到正逐步擴展至近封裝光學與可插拔應用。
• 在產能擴充期間，公司持續應對內部與外部的供應限制，同時維持產品品質。

分析師問答集錦

• Farmers Branch 產能爬升：管理層描述了 2027 年產能呈大致線性爬升的情況，從年初近乎零增加到年末每年約 4 億美元的產能規模。實際營收利用率將取決於市場需求。
• 穩態毛利率：管理層表示，Farmers Branch 廠在初始爬升期過後，應有助於提升目前的毛利率水準。長期經濟效益已反映在公司的目標模型中。
• GPU 與客製化 ASIC 商機：FormFactor 已在 GPU 與客製化 ASIC 計畫中出貨並通過認證，但管理層表示目前的成長受限於生產產能。Farmers Branch 廠旨在支援更大幅度的量產出貨。
• HBM 去堆疊：管理層預期這不會對 HBM 成長造成重大影響，並強調 FormFactor 在最終堆疊測試與高速效能方面的差異化優勢。
• 客戶認證策略：Farmers Branch 廠初期將服務晶圓代工與邏輯客戶。具備切實可行且可加速認證計畫的專案將獲得優先權，以避免新產能閒置。

完整法說會逐字稿

---

完整財報電話會議逐字稿

管理層陳述

Operator

Thank you and welcome, everyone, to FormFactor's Third Quarter 2026 Earnings Conference Call. On today's call are Chief Executive Officer, Mike Slessor; and Chief Financial Officer, Aric McKinnis. Before we begin, Stan Finkelstein, the company's VP of Investor Relations, will remind you of some important information.

Stan Finkelstein

Thank you. Today's a company will be discussing the P&L results and some important non-GAAP results intended to supplement your understanding of the company's financials. Reconciliations of GAAP to non-GAAP measures and other financial information are available in the press release issued today by the company and on the Investor Relations section of our website.

Today's discussion contains forward-looking statements within the meaning of the federal securities laws. Examples of such forward-looking statements include those with respect to the projections of financial and business performance, future macroeconomic and geopolitical conditions the benefits of acquisitions and investments, including the ramp-up of manufacturing facilities, anticipated industry trends, potential disruptions in our supply chain the impacts of regulatory changes, including tariffs and changes in export controls, the anticipated volatility in demand for products; our ability to develop, produce and sell products and the assumptions upon which such statements are based.

These statements are subject to known and unknown risks and uncertainties that could cause actual results to differ materially from those expressed during this call. Information on risk factors and uncertainties is contained in our most recent filing on Form 10-K with the SEC for the fiscal year ended December 27, 2025, and in our other SEC filings, which are available on the SEC's website at www.sec.gov and in our press release issued today. Forward-looking statements are made as of today, October 28, 2026, and we assume no obligation to update them.

With that, we will now turn the call over to FormFactor's CEO, Mike Slessor.

Mike Slessor

Thanks for joining us today. During the third quarter, FormFactor again grew revenue, gross margin and earnings per share, with each metric exceeding the high end of our outlook range. These results continue a theme you've seen from us for the past year, namely broad-based growth and demand from our diversified leadership position at the intersection of high-performance compute and advanced packaging and our ability to serve this demand by producing more output from our existing manufacturing footprint.

On a year-over-year basis, we grew quarterly revenue close to 40% within essentially the same manufacturing footprint. Our fourth quarter outlook extends this trend. Although we are now reaching the limits of our ability to extract more output from our existing manufacturing footprint, we're also bringing our new Farmers Branch factory online. We recently qualified the assembly and test portion of this new facility and only a few days ago, shipped the first production unit to a leading customer.

We celebrated these milestones last week, together with almost 200 customers, suppliers, partners, employees and government officials at a ribbon-cutting event that formally launched or nearly 300,000 square foot facility. Over the course of 2027, our top priority is the ramp-up of Farmers Branch. We'll ramp as quickly as we can while ensuring we both sustain the efficiency improvements we've demonstrated in the last year and maintain the market-leading product quality, it is one of FormFactor's foundational competitive advantages. Aric will provide more details on the status and plans for Farmers Branch later in the call.

Earlier this month, we announced the expansion of our long-standing partnership with Teradyne, one of the world's leading automated test equipment, or ATE, suppliers. Together with our multiyear partnership with Advantest, this announcement formalizes the collaboration and partnership between FormFactor and both market-leading ATE suppliers. The goal of these collaborations is to speed up the development, validation and high-volume production of next-generation semiconductor applications, enabling high-performance compute, like high-bandwidth memory and co-package optics. These partnerships also reinforce and solidify the open ecosystem market structure that provides our mutual customers with full probe card ATE interoperability, enabling customers to select the optimum probe card and ATE combination for their specific chip architectures and designs.

Turning now to segment and market details. Foundry and logic was one of the key drivers of our third quarter growth, increasing 10% sequentially on the expected broad-based strength across multiple applications. We expect continued demand growth and breadth in the current fourth quarter. In this market, we're building on our established positions in PCs, mobile, data center CPUs and networking, while adding new growth vectors like GPUs and hyperscaler custom ASICs. Specifically, during the third quarter, we shipped probe cards to the world's leading foundry for a single GPU design in multimillion dollar volumes. Building on this successful qualification and initial ramp, we're well positioned to meaningfully grow share in this important part of the foundry and logic market in 2027, with Farmers Branch providing the required capacity.

Looking ahead in foundry and logic, accelerating trends of higher-speed interfaces, greater power density and increasingly complex advanced packaging chiplet architectures for next-generation GPUs and custom ASICs or increasing probe card performance requirements. These higher performance requirements make FormFactor's advanced MEMS probe technology, a requirement for cost-effective production test, and we're building the foundation to achieve share gains in 2027 across a diverse set of GPU and custom ASIC designs. Our exposure to multiple customers and applications provides us with several ways to participate in the growth of high-performance compute and shows the benefits of our diversification strategy, reducing reliance on any single customer program or architecture to drive our revenue and growth.

Turning to DRAM probe cards. Third quarter revenue was as expected, essentially unchanged from the record levels of the second quarter, but the mix shifted significantly. This mix shift was also expected with APM revenue declining and DDR5 revenue increasing as our customers move our wafer start mix to more DDR designs to capture the profit opportunity provided by significant DDR market price increases. Since probe cards are specific to each customer chip design, our DRAM probe card mix tracks these dynamic customer product shifts, and we expect these quarter-to-quarter shifts between DDR and HBM to continue so long as the current bit constrained DRAM market conditions persist.

Another important point related to this mix shift is that at the same time, we achieved record DRAM probe card revenues with a DDR5-heavy and HBM-light product mix, we delivered record gross margin performance. Compared to the form factor of years past, this is evidence of a very different profitability profile, demonstrating the impact of the successful execution of several DRAM probe card manufacturing cost reduction programs. The net result is improved DRAM probe card gross margins, which has turned less in the dependence of our company-level gross margins on product mix.

Turning now to our Systems segment. Systems delivered a record quarter, growing 20% sequentially on accelerating demand for our Triton system for co-packaged optics or CPO applications. This accelerating demand has been evident in the steady increase in our expectations for this business in 2026. We started the year expecting between $10 million and $20 million of Triton revenue and now expect upwards of $30 million in 2026. In tandem with the growing CPO business, we're also serving demand for additional photonics applications, such as CPO's close cousin near-packaged optics or NPO as well as pluggable devices. This beyond CPO application coverage is the result of our intentional focus on developing the world's leading optical probing technologies applicable across a wide range of photonics applications.

In this case, we've prioritized building leadership in test insertion 1, which ensures known good die or KGD, of the photonic integrated circuit or PIC. Both CPO and NPO architectures require KGD PICs before they are combined with the electronic ICs to form the optical module and scale-up and scale-out network switches. As a result, form factor benefits from broader photonics adoption, whether it be CPO or NPO. Part of our leadership position in Photonics probing stems from the acquisition of Keystone Photonics a year ago.

As we've discussed, Keystone's innovative and differentiated optical probe technology enables high-efficiency optical coupling to our customers' photonic devices in much the same way our advanced MEMS probes provide best-in-class electrical connectivity to our customers' chips. To maximize our competitive advantage from this acquisition, we've made the decision to more tightly integrate the Keystone team and technologies into FormFactor, with an exclusive focus on enhancing FormFactor's development and delivery of the world's leading photonics probing solutions.

We believe the exciting growth in CPO, NPO and pluggables represents the very early stages of widespread adoption of photonics in the broader semiconductor industry. Traditional copper interconnect is reaching physical limits in speed, heat and energy consumption and Photonics provides a fundamentally more efficient way to transmit data within and between data centers using light instead of electricity. FormFactor is ideally positioned to help lead test for this new area as our lab to fab strategy has produced a decade-long first-mover advantage paired with key customer and partner relationships.

Before turning the call over to Aric, I want to once again thank the global FormFactor team for continuing to demonstrate remarkable agility and perseverance in navigating the current challenging supply environment. quickly resolving internal and external constraints to again grow in the third quarter. With our recent qualifications and Farmers Branch coming online, we're well positioned to accelerate our growth as test intensity and complexity continue to rise at the intersection of high-performance compute and advanced packaging. The midpoint of our fourth quarter outlook puts us on track to exceed $1 billion of revenue in 2026, and we're excited to cross this important milestone for FormFactor on the way to the $1.6 billion revenue level and $5 of non-GAAP earnings per share of our target model.

Aric, you're up.

Aric McKinnis

Thank you, Mike, and good afternoon. Q3 was another strong quarter for FormFactor. We delivered non-GAAP gross margin of 56.5% and non-GAAP earnings per diluted share of $1.04 on our fourth consecutive quarter of record revenues, revenue levels close to 40% higher than the same quarter a year ago. These results continue the operating story we discussed last quarter, addressing broad-based demand by driving higher output from our existing manufacturing footprint through improved cycle times, yields and operating efficiency. Revenue increased 9% sequentially, while non-GAAP operating expenses fell by approximately 250 basis points as a percentage of revenue.

Combined with record revenue, these improvements produced strong operating leverage across the P&L. Non-GAAP gross margin increased 320 basis points sequentially to 56.5%. Approximately $10.5 million or about 380 basis points of benefit resulted primarily from tariff refunds and favorable subcomponent costs in our Quantum business that are not expected to recur. Excluding these items, non-GAAP gross margins would have been between 52.5% and 58%, about 150 basis points above the baseline described last quarter. This durable improvement in baseline non-GAAP gross margins is the result of continued improvements in labor performance, manufacturing spending, yields and other production costs. We now view around 53% as the gross margin baseline at Q3 volumes.

Over the course of the Farmers Branch ramp in 2027, we expect leverage from higher volumes and increased operational efficiency to effectively offset the temporary inefficiencies typically associated with new factory ramps. Accordingly, we expect 2027 gross margins to continue at levels similar to the current baseline. Stepping through our results in a bit more detail. Q3 revenues of $280.5 million came in $10.5 million, above the midpoint of our Q3 outlook range of $260 million to $280 million and were up $22.3 million or 9% from Q2. Systems segment revenues grew on higher Triton photonic system revenues, reaching a new record of $59.5 million in Q3, up $11 million or 23% from Q2.

GAAP gross margin for the third quarter was 54.7%, up from 50.7% in Q2. Cost of revenues included $5.1 million of GAAP to non-GAAP reconciling items, primarily related to stock-based compensation, amortization of intangibles and restructuring charges. Details of the GAAP to non-GAAP reconciling items are outlined in our press release issued today and in the reconciliation table available Relations section of our website. On a non-GAAP basis, third quarter gross margin was 56.5%, 320 basis points higher than Q2 and 100 basis points above the high end of our outlook range. The favorable impact of tariff refunds on gross margins in the third quarter was about 300 basis points or $8.6 million.

In Q4, we expect to receive back the remaining roughly $2 million of tariffs paid under IEEPA or around 70 basis points at the midpoint of the outlook range. Once these refunds are received, we do not expect any ongoing benefit to baseline gross margins from tariff refund. Probe card gross margin increased by about 90 basis points to 55.3% as higher utilization, improved yields, manufacturing spending discipline and tariff benefits more than offset the less favorable mix in DRAM. Systems gross margin increased by more than 10 percentage points to 61%. About half of the Systems gross margin improvement from Q2 was driven by record revenue volumes that created significant operating leverage on fixed costs, primarily due to increased sales of our Triton photonic systems.

The remaining improvement came from items we don't expect to recur as described earlier. Our GAAP operating expenses were $71.8 million for the third quarter, down from Q2 and down as a percentage of revenue. On a non-GAAP basis, operating expenses were $64.4 million or 23% of revenue and also down from the second quarter at $65.7 million or 25.4% of revenue. This approximately 250 basis point sequential improvement in non-GAAP operating expenses demonstrates continued operating leverage despite including increased Farmers Branch ramp costs. Included in Q3 operating expenses were $6.4 million of preproduction ramp costs.

We expect these expenses to transition to the cost of revenues as capacity expansion phases are released to production. GAAP net income for the third quarter was $73.8 million or $0.93 per fully diluted share, up from GAAP net income of $56.2 million or $0.71 per fully diluted share in the previous quarter. The increase was driven primarily by higher revenue, higher gross margin and lower restructuring-related costs. Third quarter non-GAAP net income was $82.3 million or $1.04 per fully diluted share, up from $65 million or $0.82 per fully diluted share in Q2. About $0.09 of EPS was attributable to tariff refunds in the third quarter. The GAAP effective tax rate for the third quarter was 15.5% and the non-GAAP effective tax rate for the third quarter was 17.6%.

Moving to the balance sheet and cash flows. We delivered free cash flow in the third quarter of $64.8 million compared to $52.6 million in Q2. This increase in free cash flow was driven primarily by higher cash flows from operations, partially offset by higher cash spent on CapEx for Farmers Branch. Cash flows from operations were $104.3 million in Q3, up $42.5 million from Q2, and driven primarily by higher net income, partially offset by working capital investments to support the higher operating levels. At quarter end, cash -- total cash and investments were up $50 million to $399 million.

Our capital allocation framework remains unchanged. Cash will be used to reinvest in the business, pursue targeted acquisitions to expand our served markets and to execute share repurchases with the objective of offsetting dilution from stock-based compensation. During the third quarter, we did not repurchase any shares. And at quarter end, $70.9 million remained available under the $75 million repurchase program authorized in April 2025. As Mike mentioned at the beginning of the current fourth quarter, we announced a collaboration with Teradyne. This included a $300 million investment in FormFactor in order to formalize our partnership and reinforce the open ecosystem for probe card and ATE solutions.

The areas of partnership closely align with the growth areas in our target model and are intended to accelerate development of solutions for our mutual customers. This investment strengthens our balance sheet, but does not change our capital allocation priorities. With respect to investing in the business, the Farmers Branch ramp to initial target capacity remains on track, if not slightly ahead of our original schedule due primarily to optimization of timing of capital spending and favorable payment terms. We now expect 2026 cash capital expenditures for Farmers Branch to be between $100 million and $110 million, below our prior estimate of $140 million to $170 million.

Importantly, this simply reflects changes in timing of cash rather than a change in scope or velocity of the ramp, as project commitments and total project costs associated with the initial target capacity remained aligned with our original plan. We expect the associated cash spending to shift into 2027 as we continue executing the ramp and the expansion of the site. And we expect capital investment in Farmers Branch to remain at similar levels in the next year. Exiting 2027, we expect Farmers Branch will have the capability to produce about $400 million in annual probe card revenue, and we expect that capability to ramp over the course of 2027.

Preproduction ramp costs are currently recorded as a component of G&A, and we continue to expect $25 million to $30 million of Farmers Branch preproduction expense in 2026. As we begin production, costs will move from operating expenses into cost of revenues. We expect to see this begin in Q4 with about $5 million of start-up costs planned to move from OpEx to cost of revenues in the quarter and another $5 million remaining in OpEx, the latter related to capacity expansion phases not yet released to production.

Turning to the fourth quarter non-GAAP outlook. We expect revenue of approximately $285 million, plus or minus $10 million. At the midpoint, we expect non-GAAP gross margin of 54%, plus or minus 150 basis points. The expected Q4 non-GAAP gross margins are better than the baseline gross margin of 53% described earlier, primarily on tariff refunds, better mix and higher volumes. At the midpoint, we expect Q4 non-GAAP operating expense of approximately $69 million, plus or minus $2 million, including Farmers Branch ramp costs. Our Q4 non-GAAP effective tax rate is expected to be within the range of 15% to 19%.

Non-GAAP earnings per fully diluted share for Q4 are expected to be $0.95 plus or minus $0.09. A reconciliation of our GAAP to non-GAAP Q4 outlook is available on the Investor Relations section of the website and in our press release issued today. Q3's results and our Q4 outlook demonstrate the operating leverage available when strong demand is combined with an efficient cost structure and disciplined spending. We are continuing to produce more from our existing footprint through better cycle times, higher yields and improved resource efficiency, even while investing in Farmers Branch to create the capacity and structural economics required for the next phase of growth.

Our position at the intersection of high-performance compute in advanced packaging provides broad exposure across the AI compute architecture from HBM paired with advanced GPUs, CPUs and custom ASICs to emerging co-packaged optics and other photonics applications. As we have demonstrated, the underlying operating model is substantially stronger than it was 1.5 years ago and positions us well to deliver against our target model of $1.6 billion in revenues at 55% gross margins and $5 of EPS.

With that, let's open the call for questions. Operator?

Operator

[Operator Instructions] Our first question comes from the line of Krish Sankar of TD Cowen.

分析師問答

Sreekrishnan Sankarnarayanan

I told them Michael, Aric. First one, I think, Mike, you mentioned that alteration gross margin similar to the current baseline, which is 53%. I'm just wondering, I understand your Farmers Branch ramping up. Are there any other headwinds to think about? Or why would it margins get better as volumes ramp next year?

Aric McKinnis

Yes. Thanks for your question, Krish. I'll take this one. This is Aric. So we stated that we expect that current operating levels that our baseline gross margins are now at 53% and that's up from the 51% to 51.5% we discussed last quarter. Of course, with any ramp of a production facility, you're going to have some inefficiencies, and that's associated with head count additions and other spending that's required to support the ramp as you're ramping up capacity. What we expect to be able to do and one of the planned potential successes of our strategy is that we expect to be able to effectively offset that investment with associated increase in volume as well as efficiencies similar to the ones that we've been driving over the last 1.5 years. So while you would typically see a degradation in gross margins as a result of a ramp, we think we can largely offset those potential inefficiencies that are temporary in nature and sustain the current operating levels.

Sreekrishnan Sankarnarayanan

Got it. And then a quick follow-up for Mike, I think you had like pretty good traction in the past on HBM. Just kind of a two-part question on it. One is how do you think about like all these chatter of HBM destacking, how would that impact you? And the part 2 version of that is, if we do more -- lower the signal requirements of the pin speed requirements for HBM, how would it impact probe card intensity?

Mike Slessor

Yes. So as you said, Krish, 2 parts to that, that HBM continues to be a major driver for our business, both here through 2026 and into 2027. On the destack, we're really not seeing any significant impact. to orient everybody, we do see third quarter HBM revenue down, but that's primarily a result of customers shifting mix. And with the HBM4E and HBM5 design funnel we have and the forecast we have from customers, we expect continued strong growth I'll remind people that one of the places where we're really strongly differentiated is in the final stack test. And of course, this HBM destack involves more completed stacks with fewer die at each stack. We're moving from for example, 12 high to 8 high. And so this potentially hard to account for all the moving parts, but we don't see any significant headwinds associated with HBM growth associated with the destack.

On the speed question, if you look at the road map, the move from HBM4 to, as I said, we're working right now on initial design for 4E and 5, that continues to be an advancing road map with higher speeds. And that's also been a place where FormFactor provides strong differentiation for our customers. high-speed test performance. So they're guaranteeing to their customers, they're meeting their test specs with high productivity and high parallels from our smart matrix DRAM architecture.

Operator

Our next question comes from the line of Craig Ellis of B. Riley.

Craig Ellis

Mike, I wanted to follow up on some of the comments around foundry logic given the strength we saw in the business. First, can you go into a little bit more detail on some of the things that we're driving the sequential growth in the quarter. And I think you indicated that you expected the business to grow again in the fourth quarter. Is it similar drivers? Is it new drivers and as you commented, can you help us understand the degree to which a traditional x86 server strength is benefiting the business in light of the fact that Micron was out there less than a month ago, talking about high teens growth this year and next.

Mike Slessor

Yes. So foundry and logic, as you know, up in Q3, right around 10%. And as I said in my prepared remarks, we expect foundry and logic to grow again in Q4. One of the key features of this growth -- key drivers of this growth is CPUs for presumably a genetic AI use cases, right? This has been an interesting development in 2026. And this is an example where we have really strong exposure, right, both through historical long-term customer relationship where we have high native share that customer, almost a 10% customer in Q3 based on this data center CPU strength, and we would expect them probably to appear as a 10% customer in Q4, but also exposure with x86 fabless and ARM CPU features. So as that business continues to grow, nice exposure, it is one of the key drivers for foundry and logic right now.

Craig Ellis

That's really helpful. And then the follow-up is on the systems business. You've been real clear this year we should expect strength in Triton. It seems like it's coming in even stronger than we expected 6 months ago, maybe even 3 months ago, as you talk to customers, what are they signaling about the durability of the strength we're seeing today as we look through the fourth quarter of this year and into 2027?

Mike Slessor

Yes. Triton and Photonics has been one of the real highlights for us in 2026. So it's been a set of investments we've been making for the better part of a decade. And as you noted, we've accelerated our expectations now to north of $30 million in Triton revenue for 2026 after starting the year at between $10 million and $20 million. Customers continue to be very bullish is still really dominated by one ecosystem, at least on the CPO side. But as I mentioned in the prepared remarks, what are the other things we're seeing is some diversification where we're seeing adoption in NPL applications and some pluggables applications as well. In pluggables, not necessarily the Triton system but a derivative technology. There's a focus area where we expect continued growth, right? Photonics is such a compelling solution for the power and energy challenges that the industry face. It's really a question of when and not if, and we're seeing the early innings of that when really take place here in '26 into '27.

Operator

Our next question comes from the line of George Brown of Deutsche Bank.

George Brown

Just 2 from my side. So just firstly, on the GPUs and maybe the ASIC side, I think you said in your statements that you ship products for the GPU in this quarter. But I think most of the step-up in your sales to TSMC which is again a sort of over 10% customer. It sounds like it's mainly from Triton or the CPO stuff rather than maybe less so the GPU. So how should we think about the GPU sort of contribution in Q3 and into Q4? And also as we go into 2027 with more capacity coming online. And I guess in a similar vein, on the ASIC side, sort of are there any updates here in terms of qualifications. I know last quarter, you talked about the Google TPU likely requiring advanced MEMS at some point. So any update there as well? And then I have a second.

Mike Slessor

Okay. Let me see if I could parse, there was a lot in that question. Let's start with the GPU. We did say in the prepared remarks that we shipped probe cards for a single GPU design to the world's leading foundry in multimillion dollar quantities. And this is the very early part of us becoming a key part of those projects anyway. We're well positioned to grow that business. And start to take significant share in 2027, and that's a key feature of our growth. Similar for custom ASICs, right, where we qualified shipped multiple units but we're really constrained essentially by our ability to produce in volume. And that's why Farmers Branch, getting that ramped up, as Aric described, is the #1 priority for the company. Get ourselves very well positioned, whether it be GPUs, x86 fabless CPUs, custom ASICs, customers need our technology. This advanced MEMS technology is a key enabler for high-performance compute getting the capacity online to go deliver that top priority for the company, and we're executing really well on that as fast as we can.

George Brown

Perfect. And just following on from what you said in terms of capacity. We're hearing more from your peers about accelerating capacity additions really sort of doubling down here to ramp as fast as possible. I think Advantest said this morning, they're accelerating their plans to double their capacity on the logic test equipment side and hinted at doubling against over the midterm. So just wondering how much flexibility here you guys have in terms of accelerating the ramp at Farmers Branch into '27, maybe into '28 because I think there's multiple phases you have, I think, Phase 1 to Phase 3. And obviously, the injection from cash from Teradyne, I wonder if that helps, I guess, is my question.

Aric McKinnis

Yes. I think, as Mike shared, this is our #1 priority. As a company right now is making sure that we bring up the capacity at Farmers Branch as fast as possible, but within the context of making sure we preserve our quality and we do it the right way. I provided an expectation of where we think we'd be able to exit 2027 at about $400 million of capability, that's framed on a variety of potential scenarios and that's one that we think balances the risks with our execution ability. We feel really good about the team that we have working on this. It's a team that has done this before. They've demonstrated their ability to deliver over the past 1.5 years on the cost improvements that we've been working on. So we feel really good about where we're at.

Maybe the additional point I'll add here is that Farmers Branch is really a platform for us to build on going forward. And so by definition, it is flexible. That's one of the basic tenets around that side is we want to make sure that we can add capacity in a modular fashion as we move forward and as the demand environment dictates. In the ramp phase, there's -- we've got plenty to do over the next 12 months. So I think it's a good balanced view of what we put out there in terms of our potential exit in the year and then moving beyond that in the coming years.

Operator

Our next question comes from the line of Matt Prisco of Cantor.

Matthew Prisco

I guess to start off, just a follow-up on Farmers Branch there. How should we think about the linearity of that ramp throughout 2027? And do you expect to have the demand to fill that fab completely as it kind of ramps to that $400 million annual run rate capability exiting '27?

Aric McKinnis

Yes. I think -- so I'll just reiterate, we expect to be able to exit the year at this $400 million annual run rate capability. That's a capability exit, not necessarily the revenue, the revenue is going to be dictated by the timing of demand as you state. Now the current demand environment is clearly very strong. So I think our are thinking right now is that capacity is the limiting factor, and that is why it's the #1 company priority that we're focused on.

Matthew Prisco

And then that linearity through the year, should we expect that to be even...

Aric McKinnis

Yes, sorry. And then, yes, I think -- I think the linearity is it's a fair assumption to assume it roughly linear. Now there could be puts and takes as things kind of shift around in the schedule, but I think a mental model, that's a fair way to think about it, starting at something close to 0 and ramping up over the course of 2027 to the $400 million worth of annual capacity.

Matthew Prisco

Got you. And then as my follow-up, just going back to gross margins. Can you give a little more color on what's driving that 2-point increase in the baseline? And then maybe the headwind in Farmers Branch at an early ramp you expect? Can you quantify that headwind at all through '27, so we could start thinking through once we're at that full run rate, what kind of tailwind we can expect from that facility?

Aric McKinnis

Yes. So the increase in the baseline gross margin is driven really by the same things that we've been seeing over the past 1.5 years. It's cycle times of yield, just improving our efficiency in our existing footprint. We're getting more out with the same amount of resources, and that's driving up our gross margins because it's driving down our unit cost. That's really what's driving that. Once we exclude those nonrecurring items, as it relates to 2027, we haven't quantified the pluses and minuses over that time frame, mainly because it's all sense of precision. But we are committed to when we exit the year after we have completed the initial ramp, that it will be accretive to the gross margin levels we see today once we're at that initial capacity.

Operator

Our next question comes from the line of Elizabeth Sun of Citi.

Yiling Sun

I guess my first question is also on the Farmers Branch. I understand you're trying to ramp as fast as possible. throughout next year. I'm just curious, for the near term, it seems like capacity has been coming on more of a constraint on the growth it seems like Q4 is kind of constrained on that as well. So I was just curious, as we go into Q1 next year, how much longer does Farmers -- how does the capacity continue to be a headwind or your constraints? And it's just how fast you can ramp in Q1 so that like it doesn't -- for performance rent so that it doesn't become a constraint?

Aric McKinnis

Yes. I agree that we're starting to see more and more tightness in our ability to produce. We're operating kind of at the limits of what our current footprint can do. And as you correctly noted, and maybe just we've managed to squeeze out significantly more production over the course of this year from the beginning of this year to the end of this year. That's driven through these process improvements, these efficiencies that we've driven really the same footprint. We've squeezed that orange multiple times now, and it's getting harder and harder to get more out of that.

You can see that in our outlook for Q4, where incremental guide up a little bit less than maybe the prior quarter, what we were able to achieve. So you're correct in your statement that we are starting to see those limits. Farmers branches, of course, the relief fall for that. And I'll just go back to kind of what I said previously is that we expect that capacity to come online effectively linearly over the year. to that $400 million ending exiting capability.

Yiling Sun

Okay. Got it. And then on the gross margin side, I think Aric talked about or Mike talked about like there is structurally increasing gross margin in DRAM side. There's like cost programs you are executing. So I'm just curious right now and how is there any pricing element in that gross margin improvement for DRAM overall? And also how far is DRAM gross margin away from foundry logic right now?

Mike Slessor

Yes, Elizabeth, this is Mike. I'll take that. What we did note and just to orient everyone, we delivered again record DRAM revenue and at the same time, delivered record or near-record company gross margins. That's something you would have not seen from the form factor in the past. And it goes to these DRAM cost reductions. If I look at the DRAM gross margins in parse them, there's very little contribution for price increases. As we've shared with people in the past, our business model is really about delivering value to our customers and then getting compensated for that value. So we're not running a commodity business here. There are situations where an expedited delivery or a capacity reservation is valuable for our customers, and they are compensating into that. But it's not the prominent theme associated with DRAM gross margin improvement.

Operator

[Operator Instructions] Our next question comes from the line of Charles Shi of Needham.

Yu Shi

Maybe I want to push you guys a little more on the Farmers Branch gross margin. understand that during the ramp, there's always offsetting factors. But in a steady state, let's say, once Farmers Branch is fully ramped, you're in a steady state there's no grand cost-related headwinds left. I'm sure you know the cost structure of that new facility and what would be the steady state gross margin accretion to the overall form factor, if you can shed some light into that. That would be great.

Aric McKinnis

Yes. I think over the longer term, what we're seeing in terms of what we've been able to achieve so far this year as an example, and as we head into the ramp of Farmers Branch is very close alignment to the target model that we communicated earlier this year in terms of operating leverage and the sort of gains that we're able to drive as volumes increase. I think that continues to hold as we exit the Farmers Brands ramp. And the target model did include the investments required to build out effectively the capacity to support the $1.6 billion revenue level. So I think looking at -- the target model profitability is a good guide for you beyond next year.

Yu Shi

Okay. Maybe a question for Mike. Mike, the pricing system revenue, if I understand correctly, from last year to this year based on what you just said, north of $30 million, it's approximately 4x of last year's level. Obviously, this is probably still a small part of the FormFactor business, 3% or 4% of the total. But -- but let's say you've had the franchises in the past that it can grow multiple times in a shorter amount of time like HBM franchise, right? So do you do you have the capacity in place, let's say, that the Triton system demand can do another 3 or 4x or even higher going into next year. And with all the attention on Farmers Branch that seems like a proof card facility, anything you're doing on the system side.

Mike Slessor

Yes. So Triton really has been a great growth story for us in 2026. But as I mentioned earlier, really the result of many, many years of investment in partnership with some of the driver customers. In 2025, it was 0 revenue. So it's gone from 0 to $30-plus million here in 2026. And when you look at how we manufacture the Triton system, it's a little less internally focused and more externally focused on different subcomponents. We're basically integrating and assembling some subcomponents. And so it's not quite the capacity constraints that we have in what's a pretty vertically integrated business in probe cards. All of our growth forecasts, even at the high end, remember, we talked about this being a $400 million served market in 2030, showing good progress this year to start to move on the way there. We expect the same in 2027, and we do not expect capacity to be a constraint on the growth of that business.

Operator

Our next question comes from the line of Brian Chin of Stifel.

Brian Chin

Mike, so in the past, when TSMC as Bob is head or 10%, now that's, I think, 2 quarters in a row, maybe the business was $100 million below the run rate you're at now, so call like a $190 million or $180 million versus like a $280 million now. And historically, that composition of that business is -- there's been certainly a bias towards things like mobile processors, et cetera. Is that that kind of historic larger part of the TSMC foundry business, is that kind of a similar size and kind of all the incremental growth from 10% at $180 million versus 10% at $280 million. Is that all coming mainly from like custom ASICs or -- how would you describe how the composition of that TSMC business has evolved?

Mike Slessor

Yes. It's actually a microcosm of how we've driven our larger growth strategy at the company. there is multiple components to it. As you note, historically, that business was anchored in essentially mobile application processor ramps broadened out a little bit as high performance compute started brought it out a lot this year with some more high-performance compute in different projects. And we also got the Triton system there now as well. So a broader based set of opportunities with that customer, again, an almost perfect parallel to the way we're driving growth and diversification for the overall company, where we've got exposure to multiple different opportunities, making sure that when those opportunities hit, we've got exposure to them. I think the business with customer, again, a perfect microcosm of how we approach driving the overall growth strategy for FormFactor.

Brian Chin

Okay. That's helpful. And then, in terms of the initial waves or batches of products that you start to bring up in Texas and the new probe card facility. Is there a bias towards just new customer design straight into Texas? Or is there some sort of balance there between products and maybe you're transferring out of liver more and moving there? What's kind of the thought process on what kind of fleet of?

Mike Slessor

Yes. So that gets pretty at the weed, but let me try and give you some of the themes for it. We are initially ramping assembly and test for our foundry and logic products, and we'll be adding the MEMS front end for foundry and logic products in 2027. That frees up DRAM capacity at Livermore. And so the set of customers served at a Farmers Branch primarily -- well, not primarily, exclusively from the and logic customers. Within that set, we're obviously optimizing things for speed to ramp. If customers want a lengthy, detailed qualification process, well, maybe they're going to be closer to the end. We want to make sure we're utilizing that capacity as aggressively as we possibly can. One of the earlier questions about us being capacity constrained, we don't want that capacity sitting idle while customers decide whether they're going to fall or not. So there's very much a bias whether it be new customers or existing customers who have a practical but aggressive qual plan to get them into Farmers Branch first.

Brian Chin

Okay. That's pretty helpful. And then maybe in terms of the comment that things are going a little bit ahead of plan, is that in terms of your initial view of how that engagement is going with your customers? Is that kind of the learning curve on training new employees and hiring them? Is that the availability of the equipment sets?

Aric McKinnis

Yes. It's really a reflection on the expected revenue shipment timing. So we initially expected the first revenue out to be at the very, very end of this year to beginning of 2027. And as Mike mentioned in his prepared remarks, we shipped our first card in the last couple of weeks. So that's the acceleration that we're seeing. That said, the other areas that you mentioned, such as workforce, et cetera, also seem to be going well. Again, we've got a really good team working on this. I think we're continuing to tick through the milestones at a good pace.

Brian Chin

All right. Sounds good. Just be careful lifting those cards in the factory.

Operator

Thank you. I would now like to turn the conference back to Mike Slessor for closing remarks.

Mike Slessor

Thanks, everybody, for joining us today. An exciting time for FormFactor. As I said in the prepared remarks, our guidance supply is going for $1 billion of revenue this year for the first time in company history. We've got a bright future in front of us with Farmers Branch coming online. Thanks for joining.

Operator

This concludes today's conference call. Thank you for participating. You may now disconnect.