Ouster (OUST) 2026 年第三季法說會：隨著 REV8 需求升溫，營收達 5,900 萬美元

重點總覽

• Ouster 公布 2026 財年第三季營收創歷史新高，達 5,900 萬美元，年增約 50%，其中包含約 300 萬美元的授權金收入。
• 該公司出貨量超過 27,000 個感測器，其中包括超過 7,000 個 LiDAR 感測器與 20,000 台攝影機。Stereolabs 攝影機產量季增超過一倍。
• GAAP 毛利率為 43%，與去年同期基本持平。調整後 EBITDA 虧損改善 60% 至 300 萬美元，遠低於前一年同期的 800 萬美元虧損。
• 客戶需求正轉向 REV8。Ouster 目前擁有數千個 REV8 感測器在手訂單，預計出貨量將從第三季的數百個成長至第四季的數千個，並在 2027 財年底達到數萬個。
• 管理層預估 2026 財年第四季營收為 6,100 萬至 6,400 萬美元，其中包含預計 300 萬美元的授權金收入。
• 截至季末的現金總額約為 4.21 億美元。管理層表示，預計無需額外資金即可達成目前的營運目標。

核心財務數據

業務與營運表現

工業領域仍是 Ouster 最大的營收垂直市場，其次是機器人技術。需求受到倉儲自動化、國防、營建、地圖繪製、機器手臂操作、人形機器人以及移動式工業應用的支持。

Stereolabs 出貨量創單季新高，攝影機出貨量超過 20,000 台，產量較上季成長逾一倍。管理層將此成長歸因於產能擴充以及機器人需求的強勁成長。

REV8 自 5 月推出以來採用率持續成長。Ouster 表示已有數百家客戶選擇該產品，多數客戶選擇其原生全彩功能。已公布的應用包括 Forterra 國防車輛、GUSS Automation 自動化果園機具、Seneca 野火應變無人機以及 GOQ 地圖繪製載具。

Ouster 正投資改造生產線、自動化、測試與庫存。其擴大的 Benchmark 合作夥伴關係預計將支援在 2027 年底前達到每年超過 100,000 台的 LiDAR 產能。管理層強調這是產能目標，而非承諾在 2027 年每年出貨 100,000 台。

該公司還推出了基於輝達 (NVIDIA) Jetson Orin 與 Jetson Thor 平台的 ZED Box 2 邊緣 AI 電腦系列。這些系統最多可支援八個感測器，專為即時深度處理、追蹤與多感測器融合而設計。

在軟體方面，重新設計的 Ouster Gemini 流程最多能在單一伺服器上融合 100 個 LiDAR 感測器，比先前的容量提升一倍以上。管理層表示，部分 Gemini 客戶在北美營運超過 500 個據點。BlueCity 也獲得了數個百萬美元等級的專案，以及一項涵蓋額外 100 個路口、價值數百萬美元的擴建計畫。

管理層財務預測

• 2026 財年 Q4 營收： 6,100 萬至 6,400 萬美元，包含預計 300 萬美元的授權金收入。
• 2026 財年全年成長： 若以第四季財測中點計算，管理層預計不含授權金的營收將年增約 48%。
• 全年營業費用： 扣除就業稅退稅收益與一次性收購成本後，預計年增 5%–8%。
• 長期毛利率目標架構： 管理層維持 35%–40% 的目標，同時力求超越此區間上限，並將收益重新投資於產能與成長。

風險與關注事項

REV8 的前景取決於 Ouster 規模化生產、鞏固庫存以及在 2027 年前完成產能投資的能力。近期的現金用途包括擴充製造、營運資金以及持續整合 Stereolabs。

產品組合正從 REV7 轉向 REV8，但根據其更廣泛的產品組合策略，Ouster 仍繼續出貨 REV6 和 REV7。管理層未提供 REV8 與 REV7 之間詳細的毛利率比較。

Ouster 預計將在中國的工業與機器人市場保持活躍，但預計不會實質參與中國的車用市場。管理層也避免將 Stereolabs 第三季的出貨量增長推論為獨立的第四季營收預測。

分析師問答集錦

管理層指出，機器人與智慧基礎設施是工業應用之外的主要成長領域。攝影機與邊緣 AI 電腦是人形機器人平台的核心，而 LiDAR 在無人機、地圖繪製與大型自動駕駛工業車輛中更為突出。

針對 REV8 的競爭優勢，管理層強調原生全彩 3D 資料、測距、解析度、耐用性、功能安全與散熱效能。公司表示，這些特性可提升神經網路訓練、地圖繪製、紅綠燈偵測與標誌辨識的能力。

Ouster 預計透過提升出貨量、改進製造製程以及進一步整合客製化晶片，來降低 LiDAR 成本。隨著生產規模擴大，近期投資仍集中於擴充 REV8 產能與降低單位成本。

關於車用市場機會，管理層表示該公司已準備好以 REV8 及未來產品服務西方自動駕駛計程車 (Robotaxi) 方案，同時指出車用客戶的運作模式仍與機器人客戶不同。

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

Hello, and welcome to Ouster's Third Quarter 2026 Earnings Conference Call. [Operator Instructions] The call today is being recorded, and a replay of the call will be available on the Ouster Investor Relations website an hour after the completion of this call. I'd now like to turn the conference over to Chen Geng, Senior Vice President of Strategic Finance and Treasurer. Please go ahead.

Chen Geng

Good afternoon, everyone. Thank you for joining Ouster's Third Quarter 2026 Earnings Call. On today's call, we have Chief Executive Officer, Angus Pacala; and Chief Financial Officer, Ken Gianella. After the market closed today, Ouster issued its financial news release and Form 8-K, which are available on our Investor Relations website. A webcast replay of today's call will also be available on our website. During this call, we will make certain forward-looking statements. These include all statements about our competitive position, product advantages and growth opportunities, anticipated industry trends, our business and strategic priorities, our operating expense targets, the impact of our recent acquisition, the development and expansion of our products, our products capabilities and performance and our revenue guidance for the fourth quarter of 2026 and long-term financial targets.

Actual results may differ materially from those contemplated by these forward-looking statements. Factors that could cause actual results and trends to differ materially from those contained in or implied by these forward-looking statements are set forth in the third quarter 2026 financial results release and in the quarterly and annual reports we file with the Securities and Exchange Commission. Any forward-looking statements that we make on this call are based on assumptions as of today, and other than as may be required by law, Ouster assumes no obligation to update any forward-looking statements, which speak only as of their respective dates. In today's call, we will discuss both GAAP and non-GAAP financial measures. A reconciliation of GAAP to non-GAAP measures is included in the financial results release. I would now like to turn the call over to Angus.

Charles Pacala

Hello, everyone, and thank you for joining us today. I'll start with a brief recap of the quarter and an update on our strategic priorities before Ken covers our financial results in more detail. We delivered a strong third quarter with $59 million in revenue and over 27,000 sensors shipped, both records for the company. We delivered over 7,000 LiDAR sensors and 20,000 cameras, cementing our evolution into the comprehensive sensing and perception platform that meets the diverse global demands of physical AI. The industrial vertical led third quarter revenue, followed by robotics, supporting customers in warehouse automation, defense, construction and mapping. Our Stereolabs cameras had a record quarter of shipments, unlocked by our investments in production capacity to meet the surging growth in the robotics sector.

As a leader in physical AI, Ouster is enabling our customers to sense, think, act and learn on their autonomy journey. This quarter is a great proof point of the progress we have made across all 3 of our key strategic priorities, revolutionizing our LiDAR camera and AI compute products, extending our leadership in Physical AI solutions and executing towards profitability. Starting off with our product portfolio. REV8 has received exceptional market reception. As the world's first native color LiDAR, REV8 is a paradigm shift that is proving to be an undeniable competitive differentiator. Hundreds of customers have already chosen REV8 for their mission-critical applications, and they have overwhelmingly adopted native color. Customer demand has been robust while we ramp initial production. We have many thousands of REV8 sensors in our backlog, and we expect our delivery cadence to accelerate into 2027.

Our expanded partnership with Benchmark is expected to support an annual LiDAR capacity of over 100,000 units by the end of 2027. We are investing in retooled high-volume production lines, increasing precision automation, scaling testing and accelerating inventory purchases to meet the exceptional demand for REV8. Our customers are selecting REV8 because no other sensor matches its native color, increased accuracy and extreme ruggedness. Whether navigating through smoke and dust or maneuvering through complex cluttered warehouses, Ouster delivers the precision autonomy and enhanced obstacle detection required by operators around the world. Let me now share a few REV8 proof points.

In industrials, we recently announced an expansion of our multiyear strategic partnership with Forterra, the leader in military autonomous ground mission systems. Ouster provides 360-degree situational awareness with a single sensor and industry-leading reliability in the world's most challenging environments. Following the successful deployment of REV7 sensors in Ukraine, Forterra plans to integrate REV8 native color LiDAR into its future defense vehicles. Next, GUSS Automation, a subsidiary of John Deere, plans to integrate our REV8 sensors into its next-generation autonomous orchard machine fleet. REV8 delivers the precision obstacle detection and native color perception needed to scale safe autonomous agriculture worldwide.

In robotics, the superior performance of REV8 is establishing a new benchmark in the aerial mapping and autonomous drone market. We announced a strategic partnership with Seneca to integrate the REV8 into autonomous wildfire response drones. REV8 is built to operate from minus 40 to plus 85 degrees celsius with industry-leading durability, unique capabilities that enable Seneca to navigate smoke, dust and severe weather during critical cargo and firefighting operations. In addition, GOQ, a premier provider of geospatial mapping, is integrating our NDAA and BABA-compliant REV8 sensors into its TrueView drone payloads for utility and infrastructure mapping.

Turning to compute. We recently expanded our AI compute portfolio with the launch of the ZED Box 2 series, our next generation of industrial edge AI supercomputers engineered specifically to power physical AI. This series spans both NVIDIA Jetson platforms, the Jetson Orin for neural depth, tracking and multi-camera fusion in real time and the Jetson Thor for advanced perception and generative AI models. With support for up to 8 sensors, high-speed connectivity and multi-sensor fusion, Ouster is providing customers with a scalable production-ready compute solution that eliminates extra adapters or custom carrier boards. We are also directly addressing some of our customers' largest operational hurdles, compute load, memory bandwidth and system cost.

Through an updated ZED SDK and expanded native support for NVIDIA Isaac ROS across our entire ZED camera family, our customers can run high-performance spatial AI models at the edge with fewer processing bottlenecks, lowering latency and improving compute efficiency. Ouster is a one-stop shop for physical AI. With our latest software and ready-to-deploy vision AI kits, Ouster delivers the platform and performance the world's leading roboticists rely on to accelerate their deployment of physical AI from prototype to the real world. Turning to our physical AI solutions. Both Ouster Gemini, our AI-enabled perception software and BlueCity, our turnkey intelligent transportation platform, have made big strides this year. We are unlocking unprecedented scale with Ouster Gemini through our redesign of our perception pipeline.

By doubling edge processing speeds, Gemini can now seamlessly fuse up to 100 LiDAR sensors on a single server, more than double our prior capacity. This expanded capability is winning enterprise Gemini deployments that require the tracking of thousands of objects across massive city-sized environments in real time. Leading DOTs across North America are rapidly adopting BlueCity advanced detection powered by the REV8 OS1 Max. By delivering continuous 360-degree detection of road users up to 500 feet across lighting extremes and severe weather conditions, Ouster is providing cities with unmatched situational awareness. Advanced detection has led to several million-dollar BlueCity wins in Utah, Washington, Florida and Ontario, Canada. In addition, we recently won an expansion with a leading DOT, securing a multimillion-dollar contract to expand our BlueCity solution across 100 additional intersections.

Advanced detection is enabling our customer to consolidate continuous point cloud data across multiple intersections into a single unified corridor view and unlocking advanced data for improved traffic flow and safety. Complementing these leaps and performance are improvements in our machine learning iteration cycle. By optimizing data labeling, training and verification, we have compressed our end-to-end model training cycle by over 50% year-to-date. Ouster is iterating faster to more rapidly deploy improved perception software across our entire customer base, extending our lead as the foundational partner in physical AI to sense, think, act and learn. Finally, our strong results align with our long-term financial framework. We are making strategic investments in product innovation, manufacturing capacity and customer success as we scale toward profitability. With that, let me now turn the call over to Ken, who will provide more context on our third quarter financial results and full year 2026 outlook.

Kenneth Gianella

Thank you, Angus, and good afternoon, everyone. We delivered another record quarter of revenue while making significant progress on our strategic growth initiatives by expanding our presence in high-growth robotics and physical AI through Stereolabs, increasing our inventory and manufacturing capacity and driving continuous innovation in our solutions. Now turning to our third quarter financial performance. Operating results were again strong with revenue of $59 million. This represents an increase of approximately 50% compared with the third quarter last year. Revenue included approximately $3 million of royalty revenues during the quarter. Industrials led our revenue mix, followed by robotics as the addition of Stereolabs continues to strengthen our market position in robotic manipulation, humanoids and mobile industrial applications.

Our Stereolabs business had a breakout quarter, shipping over 20,000 cameras and more than doubling their production quarter-on-quarter. Our third quarter product mix reflects the operational flexibility provided by our unified sensor platform and the ramp of our REV8 capacity. As Angus mentioned, we are extremely pleased with 8 customer commitments since the launch in May. In Q3, customer demand began to shift towards our REV8 product. We expect production volumes of REV8 to continue to increase in the fourth quarter and grow throughout FY '27. GAAP gross margin was 43%, essentially flat compared with the same quarter last year. Sequentially, our gross margin was 6 points lower, primarily due to a onetime $5 million tariff refund from the prior period.

GAAP operating expense was $45 million, an increase of approximately 9% from the third quarter last year. We continue to anticipate full year 2026 operating expenses to be between 5% and 8% higher year-over-year, excluding the impact of favorable employment tax refund and onetime acquisition costs. The increase in our 2026 operating expense includes approximately 11 months of Stereolabs acquisition, investments in our product road map and the expansion of our global go-to-market capabilities. Adjusted EBITDA was a loss of $3 million, a 60% year-over-year improvement compared with a loss of $8 million in Q3 of last year. We are pleased with our progress, and we remain committed to disciplined investment for growth as we continue to scale towards profitability. Turning to our balance sheet.

We ended the quarter with approximately $421 million in cash balances, which includes the net proceeds of approximately $191 million from the common stock offering we completed in early July. Cash usage for the quarter reflects planned investments in expanding manufacturing capacity, inventory and working capital and the continued integration and scaling of Stereolabs. Our liquidity position remains one of the strongest in the industry and provides us with the strategic and financial flexibility to execute our current long-term model. As we scale towards profitability, we do not expect to need additional capital to achieve our operating objectives. Now turning to our Q4 revenue guidance. Total revenue for the fourth quarter of 2026 is expected to be in the range of $61 million to $64 million, including an estimated $3 million of royalties. For 2026, year-over-year revenue growth at the midpoint of guidance is expected to be approximately 48%, excluding royalties.

In closing, our outlook reflects sustained strength in our unified product portfolio that positions Ouster uniquely within the rapidly growing physical AI market. We remain confident in the strength of our customer demand, the continued innovation of our products and solutions and our ability to execute against our strategic and operational objectives. I'll now turn the call back to Angus for his closing remarks.

Charles Pacala

Third quarter marked the 15th straight quarter of product revenue growth, delivering sense, think, act and learn capabilities to the world's leading physical AI companies. The standout performance of Stereolabs and the overwhelming demand for REV8 are early proof points that our vision of a unified sensing and perception platform is exactly what the market demands. From farms and distribution centers to last-mile delivery, the world's most innovative companies are choosing Ouster. We provide the complete stack of LiDAR, vision, compute and software required to bring physical AI into the real world. With that, I'd like to open up the call for Q&A.

Operator

[Operator Instructions] Our first question comes from Andres Sheppard from Cantor Fitzgerald.

分析師問答

Andres Sheppard-Slinger

Congratulations on the quarter. I guess to start, Angus, it's clear that Ouster by now is much more than a LiDAR company with also cameras, AI compute, sensor fusion and perception software, as you mentioned. So I guess I'm trying to understand which applications do you perhaps expect the most near-term growth, maybe excluding industrials? Is it autonomous vehicles? Is it drones? Is it perhaps humanoids now with cameras? So just looking for an update there. And separately, if maybe you can also share a bit more granularity on the opportunities with the DoD after the certification.

Charles Pacala

Yes. Thanks for the question, Andreas. Well, I guess my answer to your question is yes. That was a good summary of many of the applications that are exploding right now in robotics, autonomous vehicles, I mean, the thesis here is everything that moves is becoming robotic and autonomous. And certainly, there are form factors that are exploding right now like humanoids and drones. And we have a really unique set of offerings to bring to the table for both those. And humanoids, camera-centric platform, but need these powerful AI computers embedded drones are more of a LiDAR-centric platform when we're talking about surveying and kind of precision surveying applications where the REV8 and the OS1 Max are incredibly good fits.

And then that just extends into industrials. So a ton of activity in robotics, and we really are a one-stop shop for anyone that's trying to build the visual cortex of all those applications. I would note that smart infrastructure is another growth area for us. We're seeing a huge appetite for our BlueCity and our Gemini offerings. And again, these are the same fundamental building blocks since think, act, and learn. It just turns out that these aren't moving vehicles. They're fixed infrastructure that still needs the same type of physical intelligence. As to the question on the DoD, I think you're referring to NDAA-compliant offerings for Ouster. And previously, we -- last year, we also had a Blue UAS certification for REV7.

So we're doing great on that front with -- these are really targeted at drone payloads for surveying applications. We're doing great in that space. REV8 is a perfect fit for the drone surveying market, and we highlighted a couple of customers there. And I think there's more to come. We obviously -- we want to get certifications like what we've gotten on REV7 on REV8. So there's definitely more to come there.

Andres Sheppard-Slinger

Wonderful. Very helpful. I appreciate all that detail. Maybe as a quick follow-up. Can you help remind us how big of a moat would you say is the qualification on NVIDIA's platform? What should investors be aware of here? And then maybe lastly, can you remind us what are the key catalysts that we should be aware of maybe in the near or medium term?

Charles Pacala

Yes. In terms of the NVIDIA platform, there are a couple of ways that we work and partner with NVIDIA today. Now one of the things we mentioned on the call is the release of an expanded set of AI computers that are based on NVIDIA Thor chipsets. So there's a moat there just in terms of the investment that we've put into that hardware and the software that runs on that hardware seamlessly. So the unified sensing and perception platform that spans cameras, LiDARs interfacing with our own purpose-built AI computers using NVIDIA chipsets big moat there. We also have a lot of work that we've announced previously with the NVIDIA automotive and Jetson ecosystems. So many different touch places there. I've said before, if NVIDIA is the brain of a robot, then we are the eyes and the visual cortex. And so I think that still holds true.

Operator

Our next question comes from Colin Rusch from Oppenheimer & Co.

Colin Rusch

Can you talk a little bit about some of the unique characteristics of REV8 and what you can do from a software perspective that are really enabling, I think, a little bit deeper relationships with your customers and your ability to just remain stickier with those folks as you go forward?

Charles Pacala

Yes, absolutely, Colin. Thanks for the question. I mean the headline feature that we keep talking about is this native color technology. And it really is a first-of-its-kind complete paradigm shift for the entire LiDAR industry. We're providing fused colorized 3D information out of a single device, never been possible before in LiDAR. And now we have a couple of quarters, many thousands of REV8 LiDARs that have been purchased by customers. and we're seeing overwhelming uptake in the RGB feature. And we're actually offering a backwards compatible non-RGB version of REV8, but overwhelming preference towards adopting the new native color technology.

So I see that as a huge technological moat that took a major investment to get it to a point where it could be adopted across a broad swath of our customer base. So that's a headline feature. It's deeply valuable in training better neural networks with more context, both 3D and color information. You can read road signs instead of just understanding their shape. You can read the state of a traffic light or you can provide amazing survey grade maps that do color and LiDAR. It's just you need to sense the world in both color and 3D to make a robot navigate it successfully. Other than that, I mean, REV8 hits on so many different capabilities simultaneously. And I think that's like just stepping back, why is REV8, why are we so confident it's going to be successful across our -- really our entire LiDAR customer base.

It's because it's got incredible robustness. It's got functional safety. It's got incredible range and resolution with the new OS1 Max. It's got this RGB capability. We actually are -- it's got the point cloud quality itself with accuracy and precision is best-in-class in many cases. Yes, the ruggedization, the thermal profile. So it's just kind of -- if you made a spider plot of it, it would be maxed out on almost all of the features that you would care about in a LiDAR. And to do that all in one platform just means that it's useful on a John Deere GUSS automation vehicle. It's also useful in a drone and it's useful in the BlueCity traffic infrastructure installation. So yes, when you do it all, it's useful everywhere.

Colin Rusch

Okay. And you guys have maintained a very strict level of financial discipline here as you've grown revenue and maintained cost structure here. Can you talk a little bit about the potential to drive material costs out of the sensors and start driving some incremental gross margin and how you might use some of that cost savings to expand the market?

Charles Pacala

Well, I can take that first. I mean, I think there's always opportunities to drive cost out of the LiDAR technology. I mean LiDAR is still not a mature technology from my perspective and we expect to be marching down a cost curve for many years to come. So that is ongoing activity. It happens each and every day at Ouster. There's also a major volume component to this. Our volumes, LiDAR volumes go up every year. We expect them to continue to go up every year. We're talking about 100,000 unit type of capacity at the end of 2027. Volume is still a big lever. So no question, we're really focused on that.

And then just the backdrop here is the whole LiDAR technology is built on a Moore's Law curve of silicon at a very fundamental level. So another way that we drive out cost is by innovating at the chip level and packing more capability into our custom silicon and saving that cost elsewhere in the system as a result. So lots of levers. In terms of gross margins, we've really committed to our long-term financial framework and the 35% to 40% gross margin is still where we're committed to for the time being. Obviously, we've come in a little above those for a while now, but still, that's where we see the future for now.

Kenneth Gianella

Yes. From an investment standpoint, it's definitely taking and looking to beat that top end of the margins and then investing that. Right now, it's about capacity for us, Colin. It's how can we make sure that we can ramp capacity. We did a great job with Stereolabs. You saw that in spades this quarter, taking them the last 2 quarters and building out capacity for them to hit these numbers. We're doing the same thing with REV8 this quarter and next quarter and ramping that into FY '27. So in the near term, it's the more capacity we get, we can lower overall per unit cost and start getting leverage from the rest of the business. So the goal is to continue to drive leverage, both from our operational standpoint, but also from our OpEx.

Operator

Our next question comes from Suji Desilva from ROTH Capital.

Sujeeva De Silva

Congratulations on the progress here. Maybe, Angus, if you could talk about the robotics market and what's the framework for your content opportunity as various robot form factors come out? I know it's primarily camera. Maybe you can help us kind of think about where that can go.

Charles Pacala

Yes, it's a good question because there's a lot of diversity in the way that robotics companies are designing these next-generation systems. So -- but right now, I mean, the AI compute, we focused the product road map on common building blocks or common technology sets, I would say, cameras, LiDAR compute and then focus on building a portfolio or a family within each one of those technology sets that absorbs the diversity of approach for each one of our customers. So if you look at one humanoid robot versus another, there are some humanoid robots that have their primary sensing in their chest. -- versus in the head. Most of them have in the head, but that's a pretty good example of how different some companies are.

Some companies put a palm-mounted camera in the palm of the robot. Others put a wrist-mounted stereo camera for doing grasping and manipulation. So there's a wide diversity of even in cameras, what types of cameras, mono, stereo, where they go on the robot. And that dictates the resolution, the size, the power draw and the cost of the cameras that go into something like a humanoid. Our goal is to sell every flavor of camera that is practical that is going into these robots and to do it in a competitive way so that when a customer has choice, they're actually coming back to Ouster for all those different flavors.

Same is true on LiDAR. That's why we have announced many different SKUs of the REV8 -- and it's also true on the AI compute. There's potentially a one-size-fits-all technology, these edge -- powerful edge GPU devices, but there's not a one-size-fits-all chipset because some things are small drones, other things are big industrial vehicles and you need a family of products. So we're very much -- our product road map is built on these families of products and -- but it's still 3 technologies, AI compute, cameras and LiDAR.

Sujeeva De Silva

That's good insight in your strategy there. And my other question is on the -- you talked about an application that has 100 LiDAR, 1,000 objects you're trying to track maybe an urban environment. It sounds intriguing. I really haven't heard these sort of large content at once opportunities. I'd be curious if you could frame that are those niche applications? Or could that be sort of a unit kind of accelerator for you guys?

Charles Pacala

This is definitely not a niche use case. I mean we have been building and deploying Gemini now for almost 5 years, I think 4 years at this point. And we're achieving a level of maturity with some of our long-term customers that is now scaling to 100 LiDAR sites, thousands of objects being tracked in real time, but it's also maturity on the breadth of deployment. So we have some customers operating over 500 sites in North America right now. So the scale on the Gemini front, this is a part, again, from BlueCity. BlueCity is a whole another story of success. But yes, we're seeing a lot of really positive trends in the uptick of physical AI technology for fixed infrastructure.

Operator

Our next question comes from Kevin Cassidy from Rosenblatt Securities.

Kevin Cassidy

This is Matti calling on behalf of Kevin. Is there any more color you could provide on the LiDAR product mix in the quarter and going forward between REV8 and REV7? And then also just whether the 2 carry similar gross margins?

Kenneth Gianella

I'll start with that. So when you think about the mix, we're at a point where we're just ramping REV8. So REV8 is going from hundreds this quarter to thousands next quarter to tens of thousands by the end of FY '27. So we fully anticipate just on the demand that we're seeing today that, that mix is going to continue. As I said in my part of the script, we're continuing to see that demand rotate more towards REV8. So that capacity, both to deliver it and the demand, we expect that to grow through '27.

Operator

[Operator Instructions] Our next question comes from Tim Savageaux from Northland Capital Markets.

Timothy Savageaux

Congrats on the results. I have a question on the capacity front as well. You mentioned exiting the year with 100,000. I guess that's annual unit volume capacity on the LiDAR side. Obviously, that implies some pretty significant quarterly revenue numbers, probably over $100 million relative to where we are now, more than double that. Is that target? Or as you look at that capacity, where do you expect to be relative to that exiting the year compared to where you are now, I guess? Any color on that would be appreciated.

Charles Pacala

Yes. And Tim, we're talking about the end of 2027, just to be clear. So yes, we're installing -- yes, so this is at the end of 2027, working with Benchmark to have installed capacity of 100,000 annualized units a year for LiDAR. These supply chain management, manufacturing takes time to put all this equipment in place. And we just want to make sure that we are getting ahead of any concerns about us being able to deliver the expanded volumes that we see each and every year and especially as we're ramping REV8 along the way. So there's major investments going in this quarter through the end of 2027 on the CapEx side on just the technology that we're bringing to bear for manufacturing.

We're just upping our game all across the board when it comes to manufacturing and making also the capacity investment at the same time. So -- and yes, to get more of the unit volume question, this is not a commitment to 100,000 unit annualized volume at some point next year. We always want to have more capacity than we are shipping at any given moment because we want to be able to meet demand spikes that happen inevitably in our business. But stepping back, will we hit those types of annualized demands in the company's near future? Absolutely. We grow our unit volumes every year, and we expect to continue to do so.

Kenneth Gianella

Yes. And you've heard the diversity that he talked about on the earlier question. We are really prided ourselves on ramping Stereolabs and what we could ship. They have a very robust portfolio, being able to ship at the increased volumes that they did quarter-over-quarter was a big accomplishment. That's through all of our manufacturing partners. But then the LiDAR side of it as we grow our portfolio, we're still shipping some REV6 and REV7. So there is going to be a continuation of all of our portfolio on this good, better, best approach to the marketplace, being able to offer the diversity of product and pricing to meet the market demand. You need that capacity.

Timothy Savageaux

Okay. Great. Appreciate that. And you mentioned Stereolabs, obviously, a big uptick here in the quarter. I know you had a target for calendar '26 revenue that you might have already hit. I wonder if you expect that type of elevated level of business to continue into Q4 at Stereolabs and whether you have an increased view of where you might be able to come in this year in terms of revenue?

Charles Pacala

Yes. I mean I think we're -- we gave guidance for the quarter for the company, $61 million to $64 million in expected revenue in Q4. And I think I'm incredibly happy with the performance from Stereolabs. I do think that there's just a fundamental shift in the demand for these types of camera products and Stereolabs is a leading brand in stereo camera and monocular camera products for these industries. So there's definitely a step change in demand versus if you look back a year or 2 in what the company was doing. So that holds. But when you get into the details of the ins and outs of each quarter, I think we're not going to go there.

Operator

Our next question comes from Casey Ryan from AmerX.

Unknown Analyst

Terrific quarter and update. I just had a few questions. I'm wondering about the life of sensors, particularly since maybe inception, have they performed and been more durable than you expected maybe initially, maybe these are internal estimates. But I'm curious as maybe the sensors move into more active environments like a robot in a warehouse that maybe blends into things or something, how you think about sort of the like life of the sensor may or may not change, I guess?

Charles Pacala

Yes, that's a great question, Casey. I think durability is one of the key differentiators for Ouster LiDARs. Many of our customers still learning what it takes to field reliable autonomy hardware in their environments. Humanoid robots turns out they fall down and they have incredible shock loads. And so -- and we see our customers kind of learning in real time about their own use cases. So -- but I think the summary here is that Ouster has separated itself on the LiDAR side with extraordinarily durable systems. I hear -- we have the internal data to prove it, but I also hear anecdotes all the time. Actually, I just heard one today.

There was a Beijing Robotics Olympics. And if you've seen any of the clips, the robots were slamming themselves into walls at the end of the 100-meter dash. And I got an anecdote today that the only thing that continued to work on one of the robots was the Ouster LiDAR on the winning robot after it crashed. So we get that kind of thing all the time, and it's we know why because we put in the years of effort to make these things ultra-reliable.

Unknown Analyst

Okay. That's great to hear. And then would that sort of be fair to sort of assign that the Stereolab, that sort of trait -- that sort of Ouster trait to those sensors as well?

Charles Pacala

Yes. I think that's one of the places where there's synergy. We have a lot of learning in those kind of use cases, and we're bringing that to bear into their engineering road map. It will trickle into everything that we do with them for the future.

Unknown Analyst

Okay. All right. Terrific. And then on REV8, I think there's a lot of excitement about the capabilities. Sort of globally, at least from our perspective, right, we've sort of bifurcated the sensor market sort of China and then ROW sort of ex China. But can the capability set start to let you win even if there is some price -- sort of a value price difference when customers start to think about how much value is sort of contained in the REV8 feature set?

Charles Pacala

Yes. I mean the -- yes, on the margin, absolutely. And actually, we have a business in China that we maintain. We have some great customers there today. And just anecdote, we were on the podium for the Beijing Robotics Olympics many, many times without sensors. So I think that -- but it's important to separate out the automotive industry in China from everything else. We really don't expect to be a player in the Chinese automotive market. But on -- in the other verticals, industrials and robotics, especially, we're a player today, and I expect to continue to be.

Kenneth Gianella

And we win head-to-head. I mean I'm answering your other question on the margin. Head-to-head, we have no problem going there because not only all the attributes that Angus listed earlier that people are choosing us for the longevity, the ruggedness, the differentiation with what we bring with the colorized LiDAR to bear, we actually get a premium on that for many of our customers because of the technological differentiation. But now you add in that we're getting the volumes and we can be competitive on a margin standpoint, too. So we have no problem taking the Chinese on head-to-head in Western markets.

Unknown Analyst

Okay. Good. That's actually a terrific answer. Both pieces of that are very encouraging. And then the last thing, a little philosophical, but Angus, you did just separate automotive from, say, robotics and humanoids. But from a selling to a customer and a solution sale, do those markets look that different forever long term in your mind? And maybe there's some connection to like the vision system software piece of this, but do you start to treat them the same? And like I'm thinking about Hyundai bought some Dynamics, they're building humanoids, they're building cars, Tesla is building humanoids building cars. Do those segments always stay separate because of what they want from you? Or do you think of them as being -- someday being the same segment, I guess?

Charles Pacala

It's an interesting question. I'm not sure I have an answer for you. But I would say today, the automotive customers are fundamentally different in how they behave, and we have a lot of them today versus everyone else. And I'm not sure that there's going to be much convergence there. There's 100 years of learnings in auto. that maybe don't all apply to the new industries in robotics.

Kenneth Gianella

But that said, we have a portfolio that is ready and willing to go meet that automotive segment. We have no problems participating within the robotaxi segment, especially with the new REV8 and the products that we have coming out. We are geared and ready when the automotive industry in the Western world is ready to go tackle that problem. We will be there for it.

Operator

[Operator Instructions]

Kenneth Gianella

If there's no additional questions, operator, we can turn it back.

Operator

Okay. Sounds good. I am showing no further questions at this time. I would now like to turn it back to Angus for closing remarks.

Charles Pacala

Alright. Well, thank you all for joining the call. We look forward to speaking with you again during our fourth quarter. Have a good day.

Operator

Thank you for your participation in today's conference. This does conclude the program, and you may now disconnect.