Tuya Inc - ADR Stock

Good morning, and good evening, ladies and gentlemen. Thank you for standing by, and welcome to Tuya Inc.'s Second Quarter 2026 Earnings Conference Call. [Operator Instructions] Please be informed that today's conference is being recorded.

I'll now turn the call over to your first speaker today, Ms. Regina Wang, Investor Relations Associate Director of Tuya. Please go ahead.

Thank you, operator. Hello, everyone. Welcome to our second quarter 2026 earnings conference call. Joining us today is our Founder and CEO, Mr. Jerry Wang; and our Co-Founder and CFO, Mr. Alex Yang. Our results and webcast of the conference call are available at ir.tuya.com. A replay of this call will also be available on our IR website in a few hours.

Before we continue, I'd like to refer you to our safe harbor statements in our earnings press release, which applies to this call as we will make forward-looking statements.

With that, I will now turn the call over to our Founder and CEO, Mr. Jerry Wang. Jerry, please?

Hello, everyone, and thank you for joining Tuya's earnings conference call for the second quarter of 2026. Tuya maintained a solid growth momentum during the quarter, despite the continued complexity of the global operating environment. Our total revenue reached USD 92.9 million, a year-over-year increase of 16%, with growth accelerating from the first quarter. Within this, revenue from our core PaaS business increased 16.9% year-over-year. These results reflect the ongoing rising smart product penetration, including steady demand home appliances, increased adoption of differentiated solutions such as smart door locks and growing demand for emerging AI-enabled product categories and also underscore the resilience of our platform business across different regions and product categories.

In terms of strategic execution, we continue to advance our AI-driven development strategy extending our AI capabilities beyond foundation models and stand-alone features towards platformization, productization and scenario-based deployment.

In the second quarter, shipment volumes of AI companion product solutions continues to expand and consumer acceptance of new forms of AI hardware began to be validated. Meanwhile, we launched the Tuya CoBuilder, which applies coding to AI hardware development, enabling developers to cover the core development process from product concept to physical device validation using natural language, further shortening AI hardware development cycles. This development further reinforce AI evolution from a mere conversational tool into a technology that operates in real physical environment and participate in sensing, understanding and execution.

Looking ahead, we will deepen our focus on the following 3 key areas: First, we will continue to advance AI-native application and product innovation, centering on high-potential scenarios such as AI home, AI energy and AI robots. We will drive the large-scale adoption of AI across a broader range of physical devices.

Second, we will continue to enhance AI development tools, such as AI coding, agent orchestration and cloud-edge-device collaboration, further shortening the cycle from ideation and development to deployment on physical devices for AI hardware.

Third, we will advance the global expansion of proven solutions while further strengthening our development ecosystem and industry partnerships to jointly explore long-term opportunities in the AI application market.

Now, let me turn the call over to our Co-Founder and CFO, Alex Yang, who will share more details about our financial performance and business progress.

Hello, everyone. This is Alex. I will now provide a brief overview of our second quarter results. Please note that unless otherwise stated, all figures are in U.S. dollars and all comparisons are on a year-over-year basis. In the second quarter of 2026, we generated total revenue of approximately USD 92.9 million, up 16% year-over-year and accelerating from the 8.3% growth recorded in the first quarter. Our PaaS business maintained strong growth where revenue from the smart home and robot product segment has also increased by double digits.

Of our total revenue, the PaaS business generated revenue of about USD 67.9 million, a year-over-year increase of 16.9%, serving as the important growth drivers for the quarter.

At the end of the second quarter, the number of PaaS premium customers for the trailing 12 months reached 318, contributing approximately 89.5% of the PaaS revenue with our core customer base remaining stable. The AI application and others segment generated revenue of about USD 11.5 million, a year-over-year increase of 3.9%, primarily driven by growth in cloud-based storage revenue such as video cloud storage. We continued to advance the value-added services, including video and AI-driven energy saving, among others, with AI-enabled applications capabilities while gradually strengthening our new and recurring service capability.

Smart home and robot products revenue was about USD 13.5 million, a year-over-year increase of 23.2%, primarily driven by growing customer demand from smart security, energy and other differentiated smart products. We will continue to increase the contribution of high value-added products and strengthen their integration with software and value-added services.

Looking at the specific driver of PaaS growth. Home appliances, smart door locks, electronics and energy products and AI companion product solutions performed relatively well during the quarter. Growth in the home appliances segment was mainly driven by customers' rollout of the smart-enabled models, the expansion of their geographic reach, a higher contribution from smart-enabled products and the migration of certain overseas brand projects from customers' legacy solutions into Tuya's. Growth in smart door locks was primarily driven by increased adoption of audio-video and low-power Wi-Fi solutions.

By comparison, demand recovery in categories such as traditional lighting and IP cameras has been relatively slow, reflecting continued divergency in performance across products and regions. In AI companion products, shipment volumes of the devices powered by our solutions continue to expand. During the June 18 Shopping Festival in China, Fuzozo, built on Tuya's solutions, ranked first in the AI toy categories on Tmall, while a number of other ecosystem products also delivered strong ranking and sales performance across major e-commerce platforms.

This provided early validation of both consumer acceptance and the commercialization potential of the new form of AI device. Beyond basic voice interactions, we have building out capabilities in multimodal perception, persona and memory, content services and user engagement, helping customers accelerate the development and mass production of the AI-native consumer hardware.

In the energy sector, solutions, including EV chargers, smart power distribution, metering and home energy management maintained solid growth. We are expanding our AI energy capabilities from electricity consumption, analytics, abnormal alerts and personalized recommendations towards dynamic electricity tariff management and user authorized automated device coordination.

Within the smart home ecosystem, customers' adoption of Matter-based solutions continue to increase across categories such as electronic products, lighting and climate control. In parallel, we enhanced local control, multiprotocol interoperabilities and third-party ecosystem compatibilities.

On margin side, our blended gross margin for this quarter was 46.3%. By segment, gross margin for PaaS was 46.8%, gross margin for AI application and others were 72% and gross margin for smart home and robot products was 21.9%. Gross margin fluctuations were mainly driven by the volatilities in upstream semiconductors cost and changes in business mix in line of the expectation. Despite this, gross profit increased by 11.1% year-over-year to approximately USD 43 million.

On expenses, we maintained disciplined expense management while continuing to invest on AI R&D and platform capability. GAAP operating expenses for this quarter were approximately USD 33.7 million, down 10.4% year-over-year, primarily due to the lower share-based compensation expenses. In terms of profitability, we recorded GAAP profit from operations of approximately USD 9.3 million with a GAAP operating margin of 10%. Non-GAAP profit from operations was approximately USD 9.6 million, a year-over-year increase by 11.7%, while non-GAAP operating margin remained in the double digit at 10.3%. While delivering revenue growth, we maintained relatively stable core operating profitability.

Net profit for the quarter was approximately $18.6 million, while non-GAAP net profit was approximately $18.9 million. The year-over-year decline in non-GAAP net profit was primarily due to the lower financial income and foreign exchange losses, while core operating profit continued to grow. On cash flow side, net cash generated from operating activities was $6.2 million during the quarter and remained positive.

At the end of the second quarter, the company's total liquid assets, including cash and cash equivalents, term deposits and treasury securities amounted to approximately USD 976 million, continuing to provide ample resources to support development of AI capability, global business expansion and our ability to navigate external uncertainties and a long-term strategy investment.

Next, I'll briefly walk you through our progress in the AI developer ecosystem. At the end of the second quarter of 2026, the number of registered developers on our platform exceeded 2.09 million. Launched during the second quarter, Tuya CoBuilder served as an AI developer gateway to the Tuya developer platform, applying AI coding to AI hardware development. By describing their requirements in natural language, developers can complete product definition, add user interface, embedded firmware, AI agents, and workflow development in one place, and then proceed directly to the device flashing and debugging. This covers the core development process from product concept to physical devices validation and help shorten the AI hardware development cycles.

In just over a month since launch, Tuya CoBuilder's AI-powered panel generation capabilities has expanded to cover 30 product categories, with average generation time for a single panel reduced to approximately 190 seconds only. This progress demonstrates that we are advancing our developer tools beyond development assistance towards end-to-end delivery capability, spanning product definition, software generation and deployment on physical devices. At the application layer, we continue to enhance Hey Tuya's device task execution capabilities, control reliability and response efficiency while exploring subscription-based and value-added services across scenarios such as AI-driven energy saving, pet care and video understanding.

Certain scenarios has already begun to generate early payment and renewals. We'll continue to focus on high-frequency use cases and long-term user value. From a broader perspective, AI capabilities are gradually expanding beyond single model integrations and in compact device sensing, contextual understanding, memory, agent orchestration and device-side execution. We'll continue to leverage the strength of our platform, device ecosystem and global developer base to translate AI capability into a scalable commercial value across a broader range of the real-world scenarios.

In summary, our revenue growth accelerated in the second quarter of 2026 with the PaaS business continuing to serve as a primary growth engine. Meanwhile, our AI capabilities are being commercialized in parallel across multiple paths, including PaaS, smart products and AI applications. Despite the impact on gross margin from semiconductor supply chain price fluctuation and business mix change, we maintained stable operating profitability and ample financial resources. Looking ahead, we'll remain focused on AI-native applications, physical AI scenarios and developer platform capability and continue to advance the transformation of AI technologies from tool-level capabilities into tangible and scalable commercial value. Thank you, all. Operator, right now we can begin the Q&A.

[Operator Instructions] We will now take our first question from the line of Yang Liu of Morgan Stanley.

Congratulations on the solid earnings. My question is about the future demand outlook. Based on your discussion with key customers, in current environment, what is the growth or demand outlook going into the second half of 2026? If you can provide a little bit more breakdown by geographic that will be even better, like what's the demand profile in U.S. or in Europe and ASEAN, et cetera.

Okay. Thank you, Liu. So right now, we see that the end demand and internal momentum is still within our expectation. So as we speak in the beginning of this year that the entire customers and the consumer side, they're looking forward to -- still to consuming more and transfer more legacy devices and solutions into the new AI one that we provide. So this momentum continues. So what we see that we have the accelerated type of rebalancing on the demand side. So this will be the overall view. So we see that the recovery will not come overnight. So it's gradually climbing. What we found here is the momentum still continue, especially based on this kind of positive -- very positive sell-through feedback from the end user side. That's the first one.

If I break down into the geographic areas, so there are different type of demand drivers. Europe still shows very strong on the demand side, especially for all type of energy-related segments. So including the new AI HMS, so home management solutions, we provide as a total solution or include different type of energy efficiency improvement single device. No matter we provide as a PaaS or we provide as a home and robot products to the solution together. That shows very strong demand still. That's the first one.

And on Southeast Asia and Latin America, the driving forces majorly come from our strong channels in the telecom carriers. So we were trying to establish a strategic partnership along with them around 2.5 years ago, and we're starting to scale and commercialize that part. So through their own channels to deliver some comprehensive total solutions for their users in the IoT field. That's a very strong potential and a very promising one because they're running on a B2B cycle. By the end of the time, it's a B2C, but they run really strong B2B cycles rather than the retail side. They are campaigning on that. That's for Southeast Asia and Latin America.

And the Middle East is still kind of in a pause right now because of the military conflict going on in the second quarter. So right now, we still kind of wait and see. The customer is still there and the customer is still doing a lot of preparations, including the product development and the new concept definitions and type of stuff. But right now, that the -- I think that overall, the business is not coming back yet. And we're looking forward to have better scenarios, perhaps maybe end of Q3 or Q4. We're looking forward to have some agreement for those conflicting countries, and then we'll be able to catch the demand. And so that way overall.

And North America is that the sell-through is still there, but some price sensitive, especially low price type of devices that show kind of fluctuations and by the pricing risk coming from the supply chain side. And so we are -- we structured that type of product mix along with the customers and to deliver a better sell-through in the second half of this year. And so I think that will be overall.

And for China, right now, we see some really good promising categories, including part of the home appliances. But we can find that recently that the major brands right now, they are speeding up the transformation from the legacy type of devices into the smart one. And from first-generation IoT type of smart devices into the AI one. So we are catching the transformation trend and helping a lot of China brands to do that. And the second one is that in China, so some AI-native categories starting to boom in like the AI companion. So our first market, we start to break through for AI companion categories is from China. So that's why [indiscernible] in Tmall. So we see that based on a large target consumer scale in China and where we find the right type of applications and coming on with a very active customer base, and we'll try to find more potentials in the new type of innovations in China.

We will now take our next question from Timothy Zhao of Goldman Sachs.

Congrats on the very solid results. My question is on your gross profit margin. I noticed that in the second quarter, the IoT PaaS margin declined on a year-on-year basis, although stabilized sequentially, while your smartphone and robot products margin actually declined sequentially and year-on-year. Just wondering if you can share more color on what were the margin drivers behind. And what is your margin outlook for these 2 segments for the third quarter and the rest of this year?

Okay. Yes. So first of all, that as everyone knows that the upstream cost fluctuations started to increase over 2 quarters on a global basis. And we are the last one to catch the impact because of buying forces. So for the -- in Q2, what we're doing is that the major of the product we just passed through the cost raise. And so which means that we maintain the gross profit and -- but we don't stick to the gross margin. And -- but till now that we really built a very good buffering on the inventory and cost balance between now and future. And in next 2 quarters or 3 and when we have the confidence that we'll be able to work through a more stable cost level of my major type of materials we needed.

So we're looking for either to stabilize the gross margin, and we figure out whatever or all the possibilities that by offering new capabilities, new technologies, we'd like to improve the gross margins overall. So that's pretty much that. So for the customer side, we really show our kindness that we best pass through the cost. But in the future, well, anything happens, so we don't -- we're looking for the most positive way to help the company to run -- continue to run the business. So it's not stick to the cost, but more stick to the value and the competence that we deliver to the customers to help them get through that.

We will now take our next question from Kai Xiao of CICC.

This is Kai. I have 2 questions, 1 is on Tuya CoBuilder you mentioned in the quarter. I wonder what's the current adoption status of Tuya CoBuilder and what's the company's medium...

Okay. So CoBuilder is something we have to do for a couple of quarters. So starting from second half of last year, some departments in Tuya R&D centers will really start to do the AI coding and to improve our own coding efficiency and also to bring more ROIs on the R&D side. So we start to do that are the major users of that. And while we have enough experience, how we'll be able to use that and deliver the right ROI and be able to know how to manage that. And we start to think about we need to duplicate our experience and open that to our customers. So at the beginning of this year, we'll start to build CoBuilder and we're happy to launch it at the second quarter.

And so we believe that will be the new type of default gateway in the future for many developers, even not only device developers, many developers to lower the bar. And including me like right now, including my financial department. So many of them, they don't know coding at all for their entire lifetime, but they're trying to write their own agent to improve their own workflow, to improve their own individual efficiencies. I believe some of you did that, too. So CoBuilder will be kind of sure where -- how -- I mean, how low the bar can reach and how easy those ideal developers will come with some innovative ideas that they can really quickly to testify the innovations and to validate whether those kind of crazy ideas make senses for some of the users and build the demo and get some pilot users and starting to run, including the [indiscernible] and scale it. So CoBuilder, we believe will be kind of the -- in the hardware world, should be kind of the momentum, like, wow, you have the [ cloud ] code maybe 1 year before.

And we believe that will be default gateway. And so we continue to bring that to -- in Q2, well, after we launched it and then we continue to do a lot of webinar trainings for those developers, even while they don't know what coding means and how they can deal with it, and we're starting to train a lot of developers. And also in the same time, we'll use this tool to attract those not developer at this moment, but they're more considered as a product manager, maybe in some and hardware company. And in the past, the strength or capability for those type of talents, user insight, product definition and interfacing design and psychological understanding.

So right now, we offer them a better tool that they can transfer that part with or without allowing any of their engineers, they'll be able to sketch that out themselves. Yes. So that -- we believe that will be the value of the CoBuilder. And so we can use that to enlarge the developer -- entire developer base by building up a better target and also be able to improve my customers' R&D efficiencies in the long run.

We will now take our next question from the line of Matt Ma of Jefferies.

I have a question on the AI application segment. So it seems like in Q2, the revenue growth has been decelerated from 17% in the first quarter. I'm just curious what is the reason behind that? I calculated it, it seems that Q2 growth is only 3%. And what can get this line back to a double-digit growth? And then also on the segment margin, on the Q1 call, you pointed that a seasonal rebound in device usage from Q2 would help you to increase the margin for this segment, but it doesn't seems that have come through. Could you walk us through what could actually happen in this quarter or the coming quarters to help to the margin recovery for this segment?

Okay. Matt, I'm sorry, I lost the second question. So is the margin for which segment? You mean the home and robot?

AI application.

Okay. AI application. Got it. So the first one is thank you for the question. And so for AI applications, right now, the growth slowing majorly come from the mix of my offering. So as you might know that in that segment, they cover 2 offers. So one is B2B and especially some of the project-based customization services we provide for the key customers. And the second part of that is the B2C, so direct services we offer for the consumer, which are the user of the devices. So they activated my value-added services through subscription. So the growth major is that we gradually still slowing, and we don't want to handle those kind of B2B projects for a long time. So the B2B project made the revenue and the growth slower. But actually, the CN grows good. So my CN services recurring revenue growth in Q2 is 22%. We're happy to see that change because we want to have their segment in the B2C will be able to cover more and more portion of this segment because we believe that will be a better value for that. So that's for the first question.

And so the second question about the -- so the margin is on the application segment, right? You asking the margin for...

Yes.

Yes. So for this one is the same is the segment we want to have more is based on the cloud and based on the AI capability. So that will be a higher valued one. So 70% up is the target margin for this segment. So right now, we'll hit it. So in the future, we'd like to hit between 75% to 80%. And the driver for that, the first one I explained so we don't want to have those kind of project and customization-based services take a larger portion because that's kind of more labor-centric and lower margin type of services. We're trying to lower the entire portion of that. So by increasing more and more cloud-based on B2C side. And on the B2C side, not only enlarge the contribution percentage on revenue. But also in the same time, while we scale the services and be able to improve more and more efficient architecture on the technical side. So we'll be able to figure out a better way to manage the cost and LLM functions in the long run. So through that, we'll be able to push the cloud-based services margin from 70% to 75% and 80%.

There are no further questions at this time. I'll now hand back to the management team for closing remarks.

Thank you, operator, and thank you all once again for joining us today. If you have any further questions, please feel free to contact IR team of Tuya. Good bye and see you next quarter.

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

Good morning, and good evening, ladies and gentlemen. Thank you for standing by, and welcome to Tuya Inc.'s First Quarter 2026 Earnings Conference Call. [Operator Instructions] Please be informed that today's conference is being recorded.

I'll now turn the call over to the first speaker today, Ms. Regina Wang, Investor Relations, Associate Director of Tuya. Please go ahead.

Thank you, operator. Hello, everyone, [ welcome ] to our First Quarter 2026 Earnings Conference Call. Joining us today are our Founder and CEO, Mr. Jerry Wang; and our Co-Founder and CFO, Mr. Alex Yang.

Our results and webcast of the conference call are available at ir.tuya.com. A replay of this call will also be available on our IR website in a few hours.

Before we continue, I'd like to refer you to our safe harbor statement in our earnings press release, which applies to this call as we will make forward-looking statements.

With that, I will now turn the call over to our Founder and CEO, Mr. Jerry Wang. Jerry will give his remarks in Chinese and English translation follows. Jerry, please.

[Interpreted] Hello, everyone, and thank you for joining Tuya's earnings conference call for the first quarter of 2026. During the quarter, despite ongoing uncertainties in the external environment and disturbances from certain regional factors, the company delivered solid growth momentum and strong execution capabilities, benefiting from a continued recovery in downstream demand, our business scale has been expanding modestly since the fourth quarter. Total revenue increased by 8.3% year-over-year with growth momentum improving quarter-on-quarter and posting positive growth for multiple consecutive quarters. Gross margin remained at a healthy level, reflecting the continued enhancement of our product value proposition and platform competitiveness.

[Interpreted] In terms of execution of our key strategies, we continue to advance our AI-driven development strategy. As we have mentioned earlier, AI is shifting from simple feature stacking towards deep integration with hardware devices and vertical industry scenarios. It is gradually evolving from a mere conversational tool into an intelligent agent that interacts and operates in the physical world. This trend has been further validated by a richer portfolio of application offerings and customer scenarios in quarter 1 2026.

[Interpreted] At the same time, we are accelerating the transition of AI capabilities from the platform layer to application layer and scenario-based products with successful deployments across multiple real-world use cases. We keep upgrading our developer tools and platform capabilities, empowering global developers to access and apply cutting-edge AI technologies at lower cost and higher efficiency. The sustained growth in AI-related revenue also reflects steady progress across our commercialization efforts.

[Interpreted] During the quarter, we introduced a range of applications aligned with this direction, including the AI-powered Smart Life assistant Hey Tuya and AI security Guardian. The significance of these initiatives lies not only in single product, but in validating the capabilities of AI agents to move beyond handling digital tasks into physical world execution and device coordination. We are seeing AI gradually develop the ability to operate across both digital and physical worlds, marking a critical step forward in the real-world deployment of physical AI.

[Interpreted] Looking ahead, we will further deepen our strategic focus on the following key areas. First, we will continue to advance AI-native application innovation, centering on high engagement categories such as smart toys. We will accelerate the penetration and large-scale adoption of AI capabilities in consumer scenarios, expanding physical AI into a wider range of everyday use cases. Second, we will scale the global rollout of proven AI solutions, particularly in the energy and green technology sectors. By bringing mature solutions to international markets, we aim to enhance our industry recognition globally. Third, we will continue to strengthen our developer ecosystem. Through open platforms and enhanced tool capabilities, we will lower the barriers to AI application development and work closely with industry partners to drive deeper exploration and commercialization of AI technologies.

[Interpreted] Now let me turn the call over to our Co-Founder and CFO, Alex Yang, for a closer look at our financial performance and business progress.

Hello, everyone. This is Alex. I will provide a brief overview of our first quarter results. Please note that, unless otherwise stated, all figures are in U.S. dollars and all comparisons are on a year-over-year basis.

In the first quarter of 2026, we generated total revenue of approximately USD 80.9 million, representing a year-over-year increase of around 8.3%. Despite ongoing uncertainties in the external environment, the company maintained its steady growth trajectory. Our core platform business remained stable, while AI-related business continued to demonstrate a strong growth [indiscernible].

Our profitability, our operating margin continued to improve. GAAP operating margin reached 9.2%, representing a significant year-over-year increase, while non-GAAP operating margin was 10%. Net margins further improved to 19.5%, reflecting continued optimization in operating efficiency and cost structure. Overall, the combination of improvements in revenue mix and disciplined expense management has driven sustained profitability gains.

Before going into segment details, we would like to note that we have adjusted the name of certain business segments this quarter. So the former SaaS and Others segment has been renamed to AI Application & Others, reflecting our continued push forward AI-enabled software services to add more applications, accurately capturing the transition from traditional cloud services to AI application services. Meanwhile, the former Smart Solutions segment has been renamed to Smart Home & Robot Products, highlighting our increased focus on AI-powered home products, household robotics and scenario-driven AI-initial devices on the hardware side.

We would like to emphasize that these changes are purely presentational and do not affect the revenue composition, recognition methods or historical comparabilities of each segment.

Within our segments, the PaaS business generated revenue of USD 59 million in this quarter, representing a year-over-year increase of approximately 9.8%. As customer demand gradually recovered, we continue to drive steady growth in our core business through ongoing optimization of our customer mix and PaaS capabilities. At the end of this first quarter, the number of PaaS premium customers reached 306, reflecting the variability of our core customer base and the structural resilience of the platform business.

The AI Application & Others segment generated revenue of USD 11.6 million in this quarter, representing a year-over-year increase of approximately 16.9%, continuing to outpace overall company growth. This growth was primarily driven by increased revenue from the cloud software services and AI application services, including AI cloud storage, energy management [ saving ], value-added services like SMS and voice services as well as app OEM and SDK offerings. This reflects the continuous progress in commercialization of our AI applications as more software products completed their AI-driven upgrades. This segment has gradually become a more growth-oriented and software services-centric component of our revenue mix.

The Smart Home & Robot Products segment generated revenue of USD 10.2 million, representing a year-over-year decrease of approximately 6.9%. The fluctuation in this segment primarily reflects our proactive efforts to phase out relatively low-value hardware products and optimize the product mix and reallocate resources towards higher value-added, especially AI-initial hardware terminals. As the segment undergoes structural adjustments, we expect the long-term profitability and scalability to gradually improve with a higher mix of higher value products.

From an operational perspective, several verticals this quarter has demonstrated structural opportunities driven by the integration of AI and smart hardware. Like in the security segment, our smart door lock business achieved 73% year-over-year growth, driven by upgrades in the multi-modeling Wi-Fi solutions, video intercoms as well as AI voices and vision capabilities. PaaS revenue from Wi-Fi-enabled smart door locks increased 65% year-over-year growth. At the same time, the AI revenues from the video-enabled locks increased substantially 500% year-over-year. This demonstrates that AI and multi-modeling capabilities are driving the traditional smart lock vertical to evolve from a stand-alone hardware model into a higher-value business model of hardware plus software service plus AI capability combined.

In the energy sector, related PaaS products, including the EV chargers, metering products and professional metering solutions are emerging as new growth drivers. We are also continuing to advance in the higher-value solutions such as AI-enabled display, gateway and voice capabilities, providing a strong foundation for our customers' product upgrades and future growth.

In the AI energy, demand in the European market for home energy management, energy storage and AI-driven energy saving solutions continue to grow. During this quarter, we made solid progress in advancing AI energy-related initiatives with key milestones achieved in the commercialization of energy storage and ecosystem accessories. Our customers received very positive feedback and secured multiple channel partnerships and orders at the exhibition such as Light + Building in Frankfurt and Solar Solutions in the Netherlands.

In Singapore's HDB project, new capabilities, app panels and deliveries are progressing on schedule. AI Energy is gradually evolving towards a comprehensive solutions model, integrating hardware bundles, software and AI orchestration plus channel operations.

From a regional and scenario perspective, Europe remains a key deployment market for energy and green technology solutions with growing demand for AI energy, smart electrical systems, spatial intelligence applications, AI smart home appliances and AI safety and security protections.

In Asia Pacific region, the Singapore HDB project continue to move through implementation and validation, while Southeast Asia and other emerging markets are beginning to generate opportunities in energy management, spatial intelligence and [ SME ] scenarios as well.

In China, AI-enabled smart door locks, AI toy and AI home products, including AI companion, continue to attract strong customer interest with some customers already advancing project upgrades and solution integration.

On margins, our blended gross margin for this quarter was 46.9%, with slightly year-over-year fluctuation primarily due to the change in the product mix and certain upstream cost variations. By segment, gross margin for PaaS was 46.1%. Gross margin for AI Application & Others was 71.7%, remained stable and reflecting the structural advantage of software and AI-driven business. And the gross margin for Smart Home & Robot Products was 23%, maintaining a level of above 20%.

While advancing AI applications and high value, we continue to focus on the cost efficiency and product value. Our expenses, we maintained disciplined cost management during the quarter with total operating expenses, OpEx, of approximately USD 30.4 million, while continually investing in core AI development and platform capability improvements driven by AI and digitization, and digitalizing operations enabled further operating leverage.

In terms of profitability, we recorded profit from operations of about USD 7.5 million for this quarter. Non-GAAP profit from operations was approximately USD 8.1 million. Net profit reached USD 15.8 million. The improvement was primarily driven by positive contribution from gross profit growth as well as lower share-based compensation expenses.

Our cash flow, net operating cash flow remained positive during this quarter. At the end of this quarter, the company's total cash, cash equivalents, time deposits and treasury securities amounted to approximately USD 1 billion plus. The strong cash position provide solid support for our continued investment in long-term AI capability development, our ability to navigate external uncertainties and opportunities and our [ competitiveness ] to enhance shareholders' returns. We will also prudently evaluate and pursue higher quality strategic investment opportunities.

Overall, the company continued to deliver revenue growth and improved profitability in a complex environment while the accelerated development of AI application business is driving the ongoing evaluation of our revenue mix towards higher-value segments.

And next, I will briefly walk you through our progress in the AI development ecosystem. Within our development ecosystem, during the first quarter, we continued to advance to the open source capabilities of TuyaOpen and further development on our AI agents. So to better address the diverse needs of AI-native developers, we also launched our new offerings, including the ultra-lightweight agent kit for the hardware developers and the vibe coding based on the Tuya hardware applications. So the vibe coding will be able to help lower the bar for many new developers as well.

Those tools enable developers to build a wide range of AI-native hardware products in a more flexible and agile manner. We remain committed to lower the bar for AI hardware and application developments while enhancing flexibility and openness, allowing developers, brands, solution providers to accelerate the process from ideation and prototyping to product commercialization.

At the end of the first quarter of 2026, the number of registered AI developers on our platform exceeded 1.96 million, maintaining steady growth. At the same time, engagement within the TuyaOpen community continued to increase. Based on our current ecosystem data, the TuyaOpen documentation platform has been accumulated over 340,000 views with more than 16,000 community members. It has accumulated abundant open-source project resources and launched a standardized demo cases library, covering mainstream application scenarios and development needs. TuyaOpen is gradually evolving from an open-source framework into an open ecosystem infrastructure for the AI hardware innovation.

From our deployment perspective, AI capabilities are increasingly extending from the platform layer into a broader range of end devices format. Whether in AI-enabled door locks, energy management solutions, sensors, AI companion toys or AI robots, they all reflect the same underlying trend. AI is evolving from isolated functions towards deep integration with the devices, scenarios and user needs. This is fully aligned with our previously articulated vision of physical AI, enabling AI to engage in real-world environments and actively participate in [ centering ] decision-making and execution in real life.

In summary, our first quarter performance further validate the commercial viability of our AI strategy. Our core PaaS business continued to provide a solid growth foundation while the deep integration of AI application services with physical hardware is emerging as a new driver for the value creation. At the same time, we have achieved meaningful progress in deploying AI solutions across high-value scenarios such as energy, entertainment and security.

Looking ahead, we will remain focused on 2 key priorities: physical AI scenarios and high value-added AI products. While maintaining financial discipline, we will accelerate the transition of AI technologies from a 2-level capability to products with tangible commercial value, creating sustainable long-term returns to our shareholders.

Thank you. Operator, we can begin the Q&A right now.

[Operator Instructions] First question, we have Yang Liu from Morgan Stanley.

Congratulations on the solid results. I would like to ask about the value chain because a lot of the sectors are suffering from the chipset shortage globally. So could management update us in terms of Tuya's situation in value chain, especially the chipset sourcing, and also update us the pricing strategy, if there's any shortage or constraint from the value chain and how to pass through the inflationary cost to the downstream.

Yes. Thank you, Liu. Yes, we really noticed those kind of fluctuations around 1.5 quarters ago. So that's why we gave a heads-up of that type of trend around the end of last year. And so the things we're doing is, the first one is that, considering of large buys of some of the major chips in the industry. And so the fluctuations we maintain as limited as we could because of the buying power.

In the same time, for those costs that inevitably we have to increase, so we'll pass through those costs to the downstream side. So that will be the basic idea and how we've been doing. And so you can notice that there are several reactions we've already been doing. The first one is that, in Q1, we already do some strategic purchasing before any cost change.

So you can notice that in our balance sheet that our inventory increased slightly. So that majorly is that it's kind of the procurement. We do that before the cost increased. And so that reflected to my inventory level and including my net cash as well. So that's the first one. So we try to use larger inventories to buy more times to working through the fluctuations.

The second one is that, you already noticed that, especially on the PaaS side, the change or those kind of difference of the gross margin of PaaS reflects that we're really trying to pass through the cost. But we didn't add the margin on the cost change because we don't want to bring more burden from our downstream side. So that reduced slightly on -- my gross margin on the PaaS as well.

We'll continue to keep focus on that and to working along with my customers and through those fluctuations. So no matter, using our scalability to manage the cost difference at the least level as we could. In the same time, we're using our inventories to try to bring more balance coming through with the time. So that will be the basic idea there. But we found that the shortage and the intensity of the momentum continue to increase in Q2 -- in the beginning of Q2.

Our next question will come from the line of Goldman Sachs, Mr. Timothy Zhao.

Congrats on the solid results. I think my question is on the revenue front. I noticed that this quarter, you achieved a pretty solid sequential acceleration on the revenue growth. However, given the very dynamic geopolitical and macro environment globally right now, I was just wondering what is your latest thoughts on the demand outlook and revenue growth outlook for the rest of this year? And what measures have you taken to stabilize or further boost the demand?

And my second question is that I noticed, as you mentioned, you changed the reporting line or changed the reporting name of the 2 of the segments that you report. Just could you further elaborate on the rationale behind and specifically on AI applications and robotic products? Just wondering if you could share more color on your plan regarding these 2 specific subsegments.

Yes. Thank you for that. So first one on the market environment, we already noticed that as we shared the colors when we released our Q4 results, we found that while the trading -- the international trading environment became stable after the November of last year, so the momentum is starting to recover and the customers are starting to return to a growing trend on the business side. So it's not that conservative. So starting from December, we already see that starting to recover. It's not overnight. So they're doing that gradually. And even though in March, we know that there will be a new fluctuation coming.

But overall speaking, the downstream side is recovering. But we have to break down into different sectors. So what we see here is that, like the appliances, like the energy, like the innovative devices, including the securities or the locks, we found that the growth momentum are more positive and almost for sure. So -- because the matter is that we found a more solid [ the amount ] of pain points on the user side and all those sectors, those companies are doing better.

But in some other sectors like the lighting, we don't see significant recovery. So it's kind of still doing what we call is into an evaluating stage on the lighting side. And some sectors that those chipset cost variations, not from our side, but from their own side, like the cameras, so all -- like some control panels with the screen, so the memory chip cost variations will bring a more significant cost difference for the finished product for the device side. And the factories and the brands, they can do less to change that direction. So that price increase might be significant for them, like the camera, for business, the entry-level cameras. Usually the FOB price or the retail price will be like -- retail price will be like USD 20 and FOB will be below USD 10.

But during those kind of memory chip and non-chip increase, we noticed that the FOB price might be able to hit above USD 15, so which means that the retail price have to increase around -- to USD 35. So that significant increase on the retail price side might bring -- might influence the consumers' buying decision. So we already noticed some sectors might be more sensitive on the -- will be more -- will be impacted more on the cost increase. Some will be more resilient. So that will be on the product sector side.

And on the region side, so combined with that is that -- still for the energy that Europe and the Southeast Asia sees strong demand for the -- including Australia, is very strong demand for the energy management solution. Especially in this year, what people are starting to notice is that energy become more and more -- how can I say, energy become more and more crucial and on the cost scalability side. So they have to pre-invest -- they are more willing to pre-invest on any energy efficiency. So for that part.

But for some other regions like Latin America, they are more price sensitive. So like I mentioned, some sectors like the cameras for this market consider that they have a lower buying power for the -- based on the macro economy in that sector -- in that region. So for them, that -- so some sectors will meet some challenge out there.

So for us is that still we're trying to use our very comprehensive hardware category mix and combined with multi-region mix to go in against different type of fluctuations. We're always looking for opportunity in some regions to balance the seesaw on the other side. So that's overall for the macro environment.

And the second one is for the AI transition. Yes, no matter it's AI application or the home robot products, so both sides, we're looking forward to give the market the signal that we're doing so hard to reallocate our resources since 2023 to transit our previous, we'll call, the first version of smart devices offering into the AI-initial offering. So starting from the end of the 2023, we really upgraded our entire platform architectures into large language model hosted, which means that since end of 2023, all those decision-making on the platform side for the device and the software applications can be based on the different large language model or the mainstream one.

And in 2024 -- May of 2024, we really launched our hardware agent platform that enable our customers to design an agent on top of the devices to make the devices be more smart and doing some things autonomously. But even no customers understand what it is, but it's the agent. And in last year, we launched our new AI platform as a new AI foundation, including the multi-modeling offerings, including the open-source projects to open some new doors for the new innovation -- innovative ideas for those customers and then give them a bridge, giving a path that how they can combine technology into innovative ideas and make it come to.

And in April, in our new developer summit, we launched our new offerings, including the agent kit that allow the hardware designers to do things more freely and including our vibe coding tools that right now, they can design any software, including apps, including the firmware on the hardware side, including the cloud services, they can do that all through vibe coding.

So all the things we're doing is that we -- to show that we are kind of an initial AI user and AI enabler. And so for that, we're trying to upgrade our offering in those 2 segments. So take the AI application, for example. We're really starting to provide that for all the cloud storage on the camera side that right now comes in with AI capability. So customers will be able to customize the event. So it's not just [indiscernible] detect any movement on the picture and give you the alarm. And you can find that you [ assume ] false alarm and then you have to turn down the notification, right, because the camera cannot tell whether it's something you should pay attention to or not. Any delivery boys come by, anyone comes from the door, you get alarm.

And starting from there, you can build an event that -- so if it's a package, so don't give me notification. And if someone stay at the front door like over 10 minutes a day, a notice. If someone showed up every day and seems like very suspicious, give me a notice. So people start to be able to create their own events and then have the camera to watch out for it. So that means it will provide significantly more values and creating more [indiscernible] pain point for the end user side. So that kind of things upgrade on those kind of offerings is -- it's a natural upgrade from our previous SaaS offering.

So we think that right now, we're starting to provide more and more AI capabilities seamlessly to the previous SaaS. And then we show that more and more users start to subscribe that services because of the AI offering, and then we're doing the upgrade on other AI applications. And that's the [indiscernible] scale, it's agent or it's purely services on the recurring model.

And for the home products and robotics, so some scenarios is including like the companion that's offering of Tuya toy or AI toy for some customers. Some customers, they have their own brands. They have their own very good toy design capabilities and channel distributions, but they don't have the capability to design things from scratch, especially they don't know anything about coding, they don't know anything about the circuit boards, about the microphone array design.

So for some of the part, we're starting to offer the entire solution. And to do that, we'll put more focus not on some, what we call is the, first generation of smart devices; we're starting to focus more on the AI, what we call AI-initial devices. So like the toy, they need the multi-modeling capability. They need the reduced noise canceling and microphone array design and they need the string projection and technologies to reflect different type of reactions from the toy side. So for that part, that's how we allocated the resources since last year.

And so right now, for this segment, the direction is that we guide the entire department to put folks on all those kind of AI-enabled and AI-initial devices. And usually, those devices will come naturally with not only the AI feature, but combined with a larger opportunity for the AI application business. So that's how we're driving folks to do that. So not a kind of connected devices segment anymore. It's become a more AI-initial offering for those customers by helping do that. Yes. So that will be the typical use cases.

Our next question will come from the line of Kai Xiao of CICC.

This is Kai, and I have 2 questions. First one is on competition. So following the emergence of agents, on-device agent deployment has become an industry trend. So could you share how has the competitive landscape evolved in Q1? And how do you view Tuya's advantage in this field? And my second question is on R&D. So could you share how is the company applying AI tools like agent coding tools in internal R&D? And what's the potential impact on margins and profitability?

Yes. So the first one, I already covered some of the parts in the market environment side. So as we see here is that 2 things. The first one, since the customer is trying to kind of escape from over-conservative momentum in our shares, this time to get back into the growth path, so what we're doing is that we just identify the right road map along with them and to fulfill that and help them to providing better products, better offering on their shelf, on their own channels and to catch the customers, catch their own end users, what I mean.

And in the same time, we already see that the end user stickiness on AI are growing very, very healthily. So which means that more and more users are trying whatever AI features and AI offerings. And I believe that it's not that significant right now. But in the near future, the consumers when they're sourcing the smart devices, AI features or what type of AI features will be kind of the key differentiations or key factors for them to make the decision. And so we are very happy to see that since second half of last year that our penetration among my ecosystem to integrate the AI capabilities we offer to the new product design and become significant and improved. So that will help us capture the trend. So that's for this part.

So what we see is that we'll always be kind of the early adapter and to notice the trend for the industry, maybe 2 or 3 quarters ahead because I can see that how -- what type of technology my customer is trying to pre-study, starting to try and when they start to implement that into the new product road map and produce that. So what we see is in us to consider as early education for the entire industry or in most of the sectors we cover to give them type of the right educational approach that AI will be considered as the next generation of key differentiations for any new things they build and to the market. And so they need to try that or need to try to understand to learn that.

Starting from the second half of last year, the customers, majority of the new products or the new projects that they kicked off, they tried that. And so then the new products they start to offer maybe at the end of the last year or at the second of this year, bring that into market going through a long procedure into the development, manufacturing, logistics and to the end design. So that's what we see here. And so it will be a very positive trend.

And on the second part is for the AI usage, I'd like to share some things. First one is that at the end of last year, the front end, which means that those ones that design the UI, user interface, and the UX, user experience, are using AI in most for our R&D side overall. And so at the end of last year, around 40% of the codes we designed for UI side are doing through AI. So that's the first one.

And we're improving that as well, considering that in this year, in this Q1, the AI coding capability improved a lot. So we found that we can use more AI to do more terminals, including the agent kit I mentioned, for the hardware designers. And the agent kit, a significant part of that is doing by AI. And while we're offering that kit, we also combine with the vibe coding tool for that kit as well. And which means that not us design the kit for AI, but customer will be doing that through the vibe coding more freely as well and very quickly to turn that into a hardware prototype.

Also at the same time, the AI usage is not only used for the R&D. So all our departments, including the financial, including the human resources, including the legal department, will be using heavily through AI. So no matter, it's improving our efficiencies on some office processing -- office work processing but also including the data analytics, BI and decision-making, et cetera.

So we consider AI to improve the efficiency in 2 parts. The first one is that to release some of my labor to focus more on higher-value work. That's the first one. And the second one is that for the -- even on the coding side, on the development side, that is to enlarge our capacity to meet the future demand growth because we already noticed that while more and more AI-initial developers coming in, that trend is a very good one, is that in this year, we noticed more and more new developers did not come from the hardware industry, which means that people starting to identify that the AI capability might be a new opportunity for new team to engage in the new smart devices business sectors that only come in the new idea and something that didn't happen in the hardware world before.

So special one is like the toy companion ones that many of my very fast-growing customers in the toy sectors, they are not toy players out there. And when we'll see, including some of the, what we call, the youth market like they do the batch is -- animation batch is focused on the cartoon. And those batch players, they don't have that business before. So that type of industry breakthrough of crossover players, they require more on the -- they rely more on the AI capability usage themselves and also they are more come with AI-initial ideas or native ideas.

So not only to reduce the cost, but also use the same level of cost to improve the capacity to capture those demands. And that's where we'll have more priority to check out too. So like I mentioned that -- so the net cash flow is considered as a strategic strength for the company, not only for the future competition, but also for the future opportunity. I think that's even more important is that while the industry are growing faster and some breakthrough happen, especially like the crossover happens, that we will not hesitate to increase the investment to capture those demand.

So I think that will be the overall momentum and so how we use the AI and we empower customers with AI. What we see is that we need to be a very powerful AI user. And until then, we will be able to empower customers.

Our last question will now come from the line of Matt Ma of Jefferies.

I have 2 questions. So the number one is on the Smart Home & Robot Products segment. I would like to know how do we think about the growth trajectory of this segment in 2026? Should we expect a growth recovery in the coming quarters? And my second question is on the AI application segment. We are seeing that the gross margin of this segment has declined by 2.7 percentage points year-on-year in the first quarter. Are there any specific reasons behind that? That's all.

Yes. So the first one is for the AI home and robot products that we're looking for to have the recovery in the coming quarter or in the coming 2 quarters. And because these are structural changes, we have to make the hard decisions. You can see that even to maintain the revenue and the gross profit growth, but in the same time, we cut off some of the products. Even we got the orders, we decided we're not to do that anymore because we don't like the model out there for the long term.

And so there is structural hard decisions, even we made some not that good numbers, but we're looking to speed up to catch it up. So we have the new offerings starting to take place in Q2 and looking for to capture orders and deliver that to make it up. So either it's end of the Q2 or it's Q3, we're looking to get the recovery. So that's the first one for the home and robot products.

And for the AI applications, yes, we found the seasonal difference. It's very interesting that we find that the key part is that the AI applications is relying on the usage of the end users based on the devices that are running. And the typical thing that we found maybe is that in the Q1, the usage is always kind of the lower season for the entire year. So that's why the usage is kind of low, so the service basis revenue has become lower for us.

Maybe one of the reason is that, the Q1, many of the users are kind of the new users and will have the new devices for the Christmas, for the holiday season promotions. And while they're starting to try the products, usually combined with some of the vacations, the usage starting to drop. So we're looking for to see the natural recovery and on the usage side, we're starting to take place on [indiscernible]. So that will be the stuff. So it's kind of very interesting one.

I will now hand the call back to management team for closing remarks. There are no more questions from the line. Allow me to turn the call back.

All right. Thank you, operator, and thank you all once again for joining us today. If you have any further questions, please feel free to contact Tuya's IR team. Goodbye, and see you next quarter.

That does conclude today's conference call. Thank you for your participation. You may now disconnect your lines.

[Portions of this transcript that are marked [Interpreted] were spoken by an interpreter present on the live call.]

Good morning, and good evening, ladies and gentlemen. Thank you for standing by, and welcome to Tuya Inc.'s Fourth Quarter and Fiscal Year 2025 Earnings Conference Call. [Operator Instructions] Please be informed that today's conference is being recorded.

I'll now turn the call over to your first speaker today, Ms. Regina Wang, Investor Relations Associate Director of Tuya. Please go ahead.

Thank you, operator. Hello, everyone. Welcome to our fourth quarter and fiscal year 2025 year earnings call. Joining us today are our Founder and CEO, Mr. Jerry Wang; and our Co-Founder and CFO, Mr. Alex Yang, the fourth quarter and fiscal year 2021 financial results and webcast of the conference call are available at ir.tuar.com. A replay of this call will be also be available on our IR website in a few hours.

Before we continue, I refer you to our safe harbor statement in our earnings press release, which applies to this call as we will make forward-looking statements. With that, I will now turn the call over to our Founder and CEO, Mr. Jerry Wang. Jerry will deliver his remarks in Chinese, which will be followed by a corresponding English translation. Jerry, please?

[Interpreted] Hello, everyone. Thank you for joining Tuya's earnings call for the fourth quarter 2025. In 2025, against the complex and evolving external environment, we maintain stability across our platform business, delivered study full year revenue growth and achieved a notable improvement in GAAP profitability. At the same time, we made solid progress in building a more systematic AI capability framework.

For full year 2025, we generated total revenue of USD 320 million representing a year-over-year increase of approximately 7.8%. Profitability and cash flow quality continues to improve. This result reflects the resilience and stability of our core platform business as well as our ongoing process in prioritizing resource allocation and execution discipline.

On the strategic fronts, we continue to incubate new AI plus IoT application scenarios and accelerated the systematic integration of AI capabilities across our platform and the device ecosystem. AI is moving from a mere overlay of these great features into fully deployable appraisal applications.

As part of our AI strategy, we introduced the AI-powered smart case assistant Hey Tuya CES through a more curative and tangible entry point integrating AI agents with hardware devices, we aim to help users entry a more comfortable and able home experiences, accelerating the real-world adoption of AI capabilities across a broader range of are scenarios.

Our understanding of the integration pathway between AI and smart products is becoming increasingly clear. AI is progressing beyond the stage of capability overlay and enduring a face of deep integration with device form factors and industrial specific scenarios. Its value is increasingly reflected in application maturities, improved revenue structures and enhanced operational efficiency.

We believe that as AI evolved from a conversational tour into intelligent agents capable of engaging in rural operations, industry expectations for underlying system stability real-time responsiveness and scalability are increasing significantly.

The impact of AI extends beyond enhancing product experiences, it is also reshaping application architecture and transforming modes of ecosystem collaboration as AI applications continue to mature, their value will increasingly be reflected in their replay capability and capacity to scale effectively across real world deployments.

Looking ahead, we will continue to advance our strategy across 3 key priorities First, we will further strengthen our AI native platform capabilities enable them to more effectively support millions of developers in creating a diverse range of next-generation AI devices and applications.

Second, we will accelerate the deployment and scalable expansion of AI application service across key scenarios.

Third, we will deepen our investment in developer ecosystem growth and enhance our support for developers fostering rebrand unit granted in innovation and commercial success.

Now let me turn the call over to our Co-Founder and CFO, Alex Yang, who will share more details about our financial performance and the business progress.

Hello, everyone. This is Alex. I will now provide more details on our fourth quarter and full year results. Please note that all the figures are in U.S. dollars and all the comparisons are year-over-year as stated otherwise.

So in the fourth quarter of 2025, we generated total revenue of approximately USD 48.5 million, representing a year-over-year increase of 3%, against the backdrop of the continuous conscious industry demand and a more conservative customer procurement cycles.

We achieved our tenth consecutive quarter of year-over-year growth. In the fourth quarter, our blended gross margin was 47.6%, while non-GAAP operating margin improved to 11.1% compared with 10.3% in the same period last year.

Non-GAAP net margin reached 24.4%. Net operating cash flow totaled USD 23.5 million, making the 11th consecutive quarters of positive operating cash flow. Gross margin remained stable underscoring the company's pricing power driven by the product value and technology capabilities as well as the strong competitive positioning of our platform-based business model in a dynamic market environment.

From a full year perspective, our stable growth in 2025 became even more pronounced. Our full year revenue reached over USD 322 million, representing a year-over-year increase of 7.8%. Blended gross margin of the full year improved to 48.2%, and up 0.8 percentage points from 2024. Non-GAAP operating margin reached to 10.5%, an increase of 2.9 percentage points year-over-year, while non-GAAP net margin rose to 24.9%.

Full year non-GAAP net income reached to a record high of USD 80.1 million, up approximately USD 4.7 million compared with 2024. Among our segments, so the past business delivered stable performance, generating revenue of over USD 230 million, representing a year-over-year increase of 6.5%, against the backdrop of extended customer margin cycles. We maintained stable growth in our core business by optimizing our customer mix and enhancing our product capabilities, but empower my customers to provide a more competitive applications.

At the end of 2025, the number of past premium customers reached to 291, continuing to contribute a structurally stable revenue to the past business. such a diversified structures without reliance on any single customer group has further strengthened our resilience in a vital operating environment.

The certain others business generated a full year revenue of USD 44.8 million, representing a year-over-year increase of 13.4%. And -- of this total, recurring services revenues rose by 37% year-over-year, emerging as a key growth driver of the SaaS. So we're looking forward to enlarge this segment. faster by this recurring model.

On a full year basis, the revenue growth from the SaaS and other business outpaced the company's overall revenue growth. This strong performance highlights the continued expansion of cloud software revenues, especially those AI-enabled software and reflects the gradual realization of the life cycle value from the platform software capabilities as the installation base of the device expense.

Our Smart Solutions business generates full year revenue of USD 45.7 million, making an 8.9% year-over-year increase. In this segment, we observed that [indiscernible] are stimulating demand in certain new product categories while also enhancing the overall pricing power of our product offerings.

At the end of 2025, our total cash and cash equivalent amounted to 1 -- over USD 1 billion, process will be USD 1,017 million, together with the term deposit and the treasury securities record as a short-term and onto investment. This net cash providing ample flexibility to support capability development, ecosystem expansion and potential capital allocation initiatives. So full year profitability was primarily driven by 3 factors. First, the continued stability of our core platform business. Second, the initial revenue contribution from AI-related products and applications; third, disciplined experience management and the realization of operating leverage.

So on the AI ecosystem side for the developers. So within our developer ecosystem, we continue to advance the open source capabilities of Tuya open and further development of our AI agent platform.

So by end of 2025, the number of registered AI plus IoT developers exceeded to $1.8 million, representing a 37% year-over-year increase. The cumulative number of AI agents on the 2-year platform reached about 16,000, spanning a wide range of smart product categories.

So as the application deployment level, AI capabilities are being integrated across a variety of end-user products gradually establishing standardized pathway for AI applications. Recently, we host our overseas development events centered on hands-on AI hardware applications. So including the first actions held in [indiscernible], this event attracts over 300 developers and which about 90% of them are from overseas.

All participating projects were built and demonstrated on the real hard growth using Tuya T5 AI development board. Completing the journeys from concept to a functional prototype within only 48 hours. This enabled Silit to be able to operate directly on physical devices. So those products span multiple scenarios, including AI, companion, wearables and desktop AI terminals as well as applications in education and security.

So some of those products have already enter subsequent incubation stage and attractive commercial interest. Beyond customer-facing products and ecosystem development, we have rapidly applied internally to enhance the development efficiency. So for instance, like in short-term front-end development process, nearly 40% of the code is generated with AI systems.

This has significantly shortened our R&D integration cycles and reduce the cost of the repetitive development. So those efficiency gains enable us to maintain the pace of the product and solution integrations while controlling the headcount growth.

So building on this foundation, we plan to launch the AI development tools for the developers within this year and through the AI holding services, web coating we aim to further lower the barriers for AI hardware development and a boost to our developer efficiencies but enabling more low-code and no-code developers to participate in the AI hardware. And industry and application ecosystem. So this initiative will help expand the developer base while accelerating the commercialization of AI applications.

Finally, so with the maturation of the physical AI technology. So the opportunity for deep integration between AI and physical world has arrived. Our launch of Hey Tuya is to build on this site without waiting for the large scale of deployment of likable, embodied robots, Hey Tuya leveraged hundreds of millions of the existing Parbati smart devices worldwide to enable AI to perceive and proactively interact with the real world today.

So it draws on understanding and reasoning large models while seamlessly interacting with the smart devices that helps manage daily tasks. So this represents a new form of integrated situational AI that's making the benefits of AI tangible and immediately accessible rather than distant other products.

In summary, the 2025 showcases the company's continuous progress across its business structures, profitability models and competitively frameworks on the technical side. So throughout 2025, Tuya's fiscal AI technology was validated for visibility in smart devices, giving rise to a word range of hardware forms, leveraging our accumulated strength across our developer communities, hardware acquisition and global delivery capabilities.

So we are well positioned to a continuous advance in AI deployment and transforming it into a sustainable, long-term competitive advantage. Looking ahead, we'll continue to focus our efforts in this direction. First, we will further categorize the platform level capabilities to enable more efficient applications of AI across diverse device and industry sees, but lower the technology barriers, we aim to help new players breach the technology gap and accelerate this adoption of AI innovations in the hardware industry.

Meanwhile, so our Hey Tuya, our next-generation AI assistant, we will establish a new standard for interactive experience in smart devices through accelerating a mass market penetration of smart products.

Finally, we'll maintain cost displace, consistently improving our profitability quantity and long-term competitiveness.

Thank you, all, operator. Right now, we can begin the Q&A section.

[Operator Instructions]. We will now take our first question from the line of Yang Liu from Morgan Stanley.

Congratulations on the solid results. I have 2 questions. The first one is regarding the recent tax rate change at the U.S. side, whether that will have any impact to our business outlook going forward. And my second question is regarding the recent upstream memory and other chipset supply constrained and whether it will impact Tuya business? this year.

Yes. Thank you, Mr. Liu. So the first question is, yes, that consider are the positive indicators that about the tariff reductions. -- and recently. But the demand didn't didn't react immediately yet. But we really see that the customers' confidence levels about the a better environment to do the business, especially global manufacturing trading, so should improve.

So people have a more positive and more confidence that macro economy will become more stable and better this year. But the demand and order didn't show up immediately. Two reasons. The first one is that still, people will consider the the global situation will be more dynamic. So -- and those type of reason to reductions maybe will not be a sustainable level. in the near future, maybe in March that maybe new executive order will come up. So we'll just like reset the tumor of the tax level, maybe into the 15% or a little bit higher. So that's the first one dynamic. So people rather not overrated.

And the second one is that this kind of new is happening during the Chinese New Year. So until now, most of the manufacturers, they started back to work today. I mean, today that we did today. So many of the manufacturers didn't starting to offer new price and try to make it new orders. So we'll see. But anyhow, we'll be very positive and directions looking forward to. And while overall costs eventually will bring down it somehow.

And so the customers will be able to have more confidence to march the demand. That's the first one.

And the second one is, yes, since last Q4, we're really starting to notice that the shortage of the production capacity of the semiconductor side. And the first one is that the strategy will not impact us because considered as significant buyers in these sectors. So many of our -- I mean all our suppliers will ensure that we will get we will get a fulfillment of our orders, no matter what. That's the first one.

At the same time, since last Q4, we really starting to prepare how can I say, quite good inventory levels to going against those kind of dynamics in the supplying cycles. So that's the first one. So shortage is not a problem for us. And about the cost rate, we continue to keep closing on that. Right now, we -- we didn't meet that immediately increase, like I mentioned that because of the buying process.

But if this kind of intensity is starting to increase without limit. We're not sure. So we'll keep closing on that. But anyhow, because of the special value position that the company will be doing so far, so that kind of increase on the supply side will not impact on our demand or significantly on our gross margin side. But we'll keep closing on that. seems that it will be less for another 1 or 2 quarters. Thank you.

We will now take our next question from the line of Timothy Zhao from Goldman Sachs.

Great. congrats on the very solid results. I also have 2 questions here. One is, I think, a more broader question about the company's position in the Agentic AI world. Given we have seen continued progress in the Agentic AI, how should we think about as value proposition to the customers in your path and SaaS business? And will the technology advances actually enhance the self-development abilities of your customers. And how should we think about the long-term rentership between Tuya and your customers? And it's actually, I think you mentioned that -- in the SaaS business, the recurring revenue actually increased quite dramatically last year. Just wondering if you can further elaborate on that.

And second question is that also in your remarks, you talked about going forward, you want to accelerate the AI deployment of the application scenarios. Just wondering if you can also further elaborate on this as well. For example, what scenarios that you see more promising. And just wondering if you can share more details.

Thank you, Timothy. So yes, it's a question. So first one is that about the macro side, we are happy to see that more customers starting to thinking on how they can create their own differentiation, how they can build their own capabilities in their own R&D side because we're happy to see that. Otherwise, we have to offer that.

So I think that AI makes no difference for past 10 years' experience is that we're starting to enable the manufacturing players to embrace the and the smart technology is starting with IoT. It's the same stuff. If they cannot do that, they want it, we have to offer it.

So for all the time, of the company's history, we continue to offer 2 things. The first thing is that if they don't have the capacity right now, we'll offer them the off-the-shelf solution turnkey. And if they will have some capability, we continue to educate them to do that and then we offer them infrastructure to allow them to do that some extra values, they want to create more freely.

So I think that's the what we call ecosystem were to create, so it's not like just keep selling stuff. They don't have to do that now. So we're happy to see that we already have a significant amount of the customers who already have their own kind of in-house capability to create their own depreciation and make their own innovations. We're happy to see that.

So the same as that we continue to enable our customers to build their own like the device level, innovations and application network. So I think AI makes no difference. We also continue to do the same thing including 2025, the showcase is that for some new players, they don't know nothing about that, but they only have some ideas how they want to bring AI into their business. we increase some term solution for them, they can grab and go.

At the same time, we'll continue to have the very deep and active conversations with their engineering team. Okay, what they can take for now. and what they can be in the future and what -- how to able to enable them to do that more efficiently than faster without the overwhelming boarding. So we'll continue to do that the same way.

So -- but what we think that make us very excited about is that several years ago, you still need to convince tell people how the smart devices are promising business. You still need to tell them that this will be in the future. But right now, you don't have to tell people that AI future. Every people are buying that.

So the key part is that they really have the concept in their mind and how you'll be able to help them to make that faster and more efficiency. And more competitively, I mean, on the user experience side. I think that's the first one.

And so the second one is -- on the SaaS recurring staff.I think the key driver for that is that remember our past, we continue to deploy a significant amount and scale of the devices overall with or without any type of recurring services out there. So which means that we will have a large base.

And at the same time, coming along with AI. So some what we call existing categories only come with IoT before, and we really see that combined with AI capability, we will be able to offer some extra experience and values on the same type of the depiction and which already deployed on the household.

So in 2025, we continue to offer some new services on the same type of the hardware. And then we see that it should work out. And even on the existing recurring services, like some storage services by offering extra AI capability, we make the services more valuable or more feasible for the end user side.

So we either continue to enlarge our recurring consumer base at the same time, we're trying to offer more recurring services out there. And we believe that will be a long term, especially for some initial products, which will mean that the new type of applications since they won those kind of new recurring models, we started to took in place from the beginning. So I think that's for the sash recurring. We continue to grow that.

I think that will be one of the fastest growing segment in our middle term. for the recurring. And the third 1 is for the AI applications, I think that we already shared some of overviews and in late last year. So for the -- those segments that I will be able to provide more significant values, we believe will be -- right now -- the first one is that all the multi modeling applications, including the video and audio interactions and interactions and analysis. So including the companion toy securities. So those type of products will really have a significant base, and we have new players coming in.

But come along with AI, so either you make those device interactions more smoothly and also combined with the perceptions of the video and audio, the devices will be able to provide more sense the security side that you will be able to protect the people's home more precisely without to bring any fortune.

And like for the companion side, so you -- or choice you really could be able to provide some educational level of the interactions by providing the language, providing the right understanding drive emotion, providing the right feedback and provide the right type of knowledge to the targets to the target customers. So that will be the first one. Martin modeling applications, especially on audience video interactions.

And the second one is data analytics and decision-making. So our typical use cases is for energy management. So coming on with a full cycle device deployment for the energy web cycle, including the generation of energy storage, consumptions and metering, you'll be able to understand how the electricity will be moving, I mean, translate from the grid into each of the devices, how people want to manage the flow.

And through all the data you'll be able to know and then the AI will be able to jump 1 step ahead is not only providing you the data analytics and suggestions, but they will be able to make the decisions that how you'll be able to control your dishwasher or a different way, how you'll be able to manage your battery bank different way, how we'd be able to manage the AC and heating system in a different way, combined with variable pricing in different timing combined with the generation of your solar panel combined with what kind of battery we have in home rent.

So -- and either to reduce the total cost directly, so that will be a typical showcase is that AI is not providing a 2. AI will be able to provide the outcome. So people will really see directly that what will be the TCO, what be the the total values they can get for the life cycle of the usage of these type of devices and they pay for the services as well.

So the data and analytics and decision-making will be another part. -- beyond energy, we're looking for more scenarios in that segment as well. So that's estimate.

We will now take our next question from Mingran Li from CICC.

Congrats on the results. My first question concerns the milestone. Given the recent geopolitical risk, how to meet assess the potential impact on Tuya's international operation. Looking at the current environment in this year, how do you perceive the recovery in demand across the overseas markets?

And my second question, I would like to ask about the shareholder returns. Data very healthy cash position and your profitability continues to improve could management share if there are any more specific plans or cancelations for shareholder returns has been moved to 2026.

Yes. So thank you for the questions. The first one is that really part of that from MS, that's the tariff questions. So the first one is that, yes, the global situation will become more dynamic, correct? We're trying to get used to that come along with our customers as well.

So right now, we see that we get to be able to see more positive indicators in that direction, either reductions of the tariff on the global side anyhow to any type of pathways, but we really see that people require -- the commerce require a better environment to do the business, and people cannot cut each other off. So we really see that. So the end demand continued to increase because the technology really provides value for the end users and they want it and they use that more and more often.

So that's what we see. And this is inevitable. I mean you can ever return on that. So coming out with the end demand increase. And so all the commerce level that people just figure out a way how they'll be able to fulfill the demand. and go through -- navigate through all the dynamic factors, including the tariffs, including the reallocation of the supply chain globally, et cetera.

So for us, we just follow the flow is that we come along with the customers to focus on. First one is to provide our offering technique offering to help them to build whatever application that makes sense for their end users and be able to scale it. That's the first one, to make them be able to provide the right thing. In the same time that we continue closely to manage the cost to come on with a different allocation of our services on the global side. right now, the -- we can deploy the services on whatever countries my customers are, we really did -- and right now, my customers are really starting to build a different type of production, and they really have different type of production centers across 11 countries all over the world.

So we just follow the flow and help them to achieve that more identity. So I think that that overall what we see for the global situation side. And so this year, we really see that people looking forward to have the rebound versus 2025 because 2025 will be kind of the over conscious situations. And people don't know what will happen and things happening like every week. And so people don't people are not willing to do even a long term across a portal like the decisions. So they keep the decision very, I mean, frequently and precisely what macro decisions.

But this year, people will already see that the sustainability on the situation, we're starting to build some better. So they try to rebound from the over conscious companies level. Yes. So that's for the macro side.

And so for second side, for the return of the shareholders, as our -- as what we've been doing for for the past 2 years, we continue as a shareholders' return as one of the prioritized target for the company as well. So we continue to provide a very sustainable and strong foundations on the operations side, including the net cash flow, including the profitability, including the growth of the revenue, including the health of the revenue structures and the margin.

So the return of the shareholders will become a long-term strategy as well. So we just announced, we have a new round of the dividend for the shareholders be -- so coming on is continue as a practice for us is that the 1 or 2x dividend a year. So that will be what we're doing for the shareholders' returns. And also, in the same time, the dividend will be more reflected on our level of net operating cash flow and profitability. So that's where we feel.

Our next question comes from Matt Ma from Jefferies.

Congrats on solid results. My question is regarding on the Smart Solutions segment. We noticed that the company showcased multiple AIoT products at CES last year. And which product categories does the company have higher confidence in sales growth in this year? And when we are thinking about product category expansion with our thought process and could we expect a relatively strong growth in the Smart Solutions segment in 2026.

Yes. So thanks for the question, Matt. So the first 1 is that I think combined with the previous questions and answers. So for the more promising, that's promising. Maybe I mean we will have more confidence level categories that can achieve a higher growth enabled by, so those categories will be those devices that can use more capabilities naturally.

So including those kind of video and audio interactions and safety stuff. And toy, what we call entertainment stuff and appliances. So those energy and those will be those segments will find that AI can use more. they can use more capabilities than ever. And some of the capability will directly deliver as a value that becomes visible for the end users. And so that's the one. And so we have a more confidence level in that segment.

At the same time, it continues to reach other segments and what will be the new innovative ideas that combined the AI deeply integrated with the existing device capabilities. We continue to sort boat as well. And but which we -- what we're looking for, we think that we're going to see in 2026 is that gradually, you fund more and more new type of advice that didn't exist before, we're trying to occur because of the AI.

So that will be 2, 3 new stuff. [indiscernible], nobody thinks that a companion type of toy will become resisting before 2025. So this type of new concept of applications. We're looking for have more because we have more talent coming into the industry. We have more players coming into the industry. The new ideas come across different world will create a very, very interesting chemistries out there.

So new categories, which I don't name that, even we don't know how should we call that, but we'll find more issue. That's the first one. That's the first question.

And second one is [indiscernible] Smart Solutions. So let me describe the better proposition of Smart Solutions is that is those type of hardware type that help our customers to attrition themselves and those differentiators, the customers prefer yet to do that because that either there will be more efficiency or that will be a must be. So a significant -- I mean, typical use cases for that is like the bird feeders, I mentioned a couple of times out there. is that that's just a concept of ideas. That's my work.

So the customers come from the -- can I say the pet products work, they know that some some of those -- their customers are looking for to interact with white life like that. So that's customer and consumer or user insight and concept ideas. So if they want to do that, they have to cover all the technology gap. And will be kind of overwhelming for them and not only because of the lack of capabilities of the engine maintain, but also that investment can be huge. I mean for them, if they do that individually.

And also, in the same time, that type of innovations need a deep intuition on the software and hardware development directly. So instead of waiting for Tuya to offer the past maybe that doesn't show up in our past road map ever. So this is that how they can work closer with Tuya if we can make that happen.

So through that, we think that we buy in this concept and then we make it as we'll offer it as a solution because we can directly make that happen and then they can try out the concept. So that will be the typical situation for the smart solution is actually we're looking for those differentiated type of offering to the market that can help my customer outstand themselves in their own segment.

In different regions, in different categories, in different vertical channels, et cetera. So we only focus on this. So that you can see that for the smart solutions, even on the hardware business, we maintain as plus margin. Reason being is that we only choose those higher value products with the differentiation and with the special technical offering and touched as a very precise targeted consumers that they're willing to pay high -- so that will be how we do.

So consider smart solution will be kind of the higher value segment type of the devices among all my past orders. So this will continue to do. So really, much our solutions will become the flagship model for my past customers, specific customers in the new year. So we do -- we continue to work on with our product road map year-over-year and the flagship types they ask us to offer as a solution.

Thank you. There are no further questions at this time. I will now hand back to the management team for closing remarks.

Thank you, operator, and thank you all once again for joining us today. If you have any further questions, please feel free to contact to IR team. Good bye and see you next quarter.

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

[Portions of this transcript that are marked [Interpreted] were spoken by an interpreter present on the live call.]

Good morning, and good evening, ladies and gentlemen. Thank you for standing by, and welcome to Tuya Inc.'s Third Quarter 2025 Earnings Conference Call. [Operator Instructions] Please be informed that today's conference is being recorded. I'll now turn the call over to your first speaker today, Ms. Regina Wang, Investor Relations Senior Manager of Tuya. Please go ahead.

Thank you, operator. Hello, everyone, and welcome to Tuya's Third Quarter 2025 Earnings Conference Call. Joining us today are Founder and CEO of Tuya, Mr. Jerry Wang; and our Co-Founder and CFO, Mr. Yi Yang. The third quarter 2021 financial results and website of today's conference call will available on our IR website at ir.tuya.com. And the replay will be posted shortly after our conclude.

Before we continue, please note that our safe harbor statement in the earnings press release applied to today's call. As we may make forward-looking statements. With that, let me now turn the call over to our Founder and CEO, Mr. Jerry Wang. Jerry will deliver his remarks in Chinese, which will be followed by a corresponding English translation. Jerry, please?

Hello, everyone. Thank you for joining Tuya's earnings call for the third quarter of 2025.

In the third quarter, the external environment remains relative continuing to in since the beginning of the year, the global consumer electronics industry experience and even recovery with customer demand becoming more cautious in the ongoing macro uncertainties. In addition to the high base from the same period last year created added pressure on year-over-year growth. Against the backdrop, our total revenue for the quarter reached approximately USD 82.5 million, marking our ninth consecutive quarter of year-over-year growth and boring the strength of U.S. business model.

Gross margin remained above 48%. This result further reflects the resilience of our business structure and the steady improvements we have made in product mix and operating efficiency over recent quarters.

In terms of profitability brought by an improved gross margin profile, greater expense efficiency and sustained leverage. Our non-GAAP net margin reached 24.4%, while GAAP net margin were 18.2%. Notably, GAAP net margin expanded by more than 23.6 percentage points year-over-year. Overall, we'll maintain key investments in business development, we have continued to optimize our expense structure, enabling incremental revenue and gross profit to translate more effectively in operating profit.

At the same time, on the strategic execution plans, we continue to fully invest AI and deepen its integration across our ecosystem. As of the end of Q3, smart devices equipped with AI capabilities accounted for 93.99% of total shipments and equipment from the previous quarter demonstrating that AI is swiftly becoming the default ratio from smart devices. .

On the user side, AR adoption is also scaling quickly. AI has clear moves beyond single category features like AI [indiscernible] to a broader backing of product categories. AI agent service now handled 15 million daily interactions for global users sporting diverse scenarios such as [indiscernible] AI translates, AI Health, AI Energy, a Paccar Air, airings pure car and AI robotics AI continues to penetrate the broader range of daily devices and life scenarios, laying the formation for large-scale product innovation and long-term valuation creation.

During the quarter, we also began global beta testing of our new agent app with Tuya ecosystem users live with our smart living mission, we are tensely developing universal AI live assistant global users, which is scheduled for official release at the actual in the United States in just over a month.

Now let me turn the call over to our Co-Founder and CFO, Yi Yang, who will share more details about our financial performance and business progress.

Hello, everyone. This is Alex. I will now provide more details on my third quarter's results. Please note that all the figures are in U.S. dollar based and all the comparisons are year-over-year based. And we delivered total revenue of approximately USD 82.5 million in the third quarter, representing a 1.1% year-over-year increase. Despite a strong person base last year and continued caution in external demand, we achieved our ninth consecutive quarter of year-over-year growth. underscoring our reserves and scalability in our business.

With the total revenue, our past business delivered strong results, generating USD 59.2 million, a 2.4% year-over-year increase driven prevalently by our strategic focus on the customer demand and product optimization. In Q3, the number of past premium customers reached 280 further strengthening our core customer base. In addition, fueled by growth in the cloud software products revenue, the SaaS and others business show consistent expansion, generating USD 11.5 million this quarter, a 15.4% increase year-over-year. Momentum was driven by continued rise in devices and a high proportion of recurring revenue [Audio Gap]

Revenue from Smart Solutions reached USD 11.8 million during this quarter, we strategically scaled by lower efficiency projects and prioritize scalable, high-value solutions such as AI energy management solution and spatial AI solution to further improve overall gross margin and cash recovery efficiency. For a regional perspective, in China market, AI toy continued to show healthy growth in the third quarter, more than 50 customers, including brands, channel partners and solution providers. They launched the products part back to -- and key products' capabilities also continue to advance, such as multimodel interactions, long-term memory and emotion expressions with several connectivity versions coming soon as well.

These improvements further strengthen the foundation of expanding into new product categories and regional markets. In the European market, demand from AI power solutions such as AI cloud storage and AI energy-saving solutions continue to rise. At the same time, we added several new industrial clients in energy and HVAC sectors during this quarter. In Asia Pacific, the point of cube, the privatized platforms for several Southeast Asian telecom operators are scaling rapidly with additional cities entering the delivery phase, the Singapore HTB housing deployment sports of Singapore projects also progressed into implementation with the first branch of the hardware and software solutions delivered and in stores in this quarter.

In North America, AI-enabled products such as smart bird feeders continue to record healthy growth, the strong adoption validates the commercial potential of niche that integrates emotional values, frequent content interactions and nonterm subscription model. and underscores the structural growth opportunities for AI products in mature consumer markets.

In summary, despite pressure in the global consumer environment, Tuya leverage its diversified product portfolios and strong software capabilities to achieve a structural growth. Those trends further strengthen our resilience against external macro volatilities and uncertainties.

Moving to gross margin. Our blended gross margins for Q3 in 2025 was 48.3%. Total gross profit reached approximately USD 39.8 million, representing a 6.1% year-over-year increase. This growth was primarily driven by concurrent improvement in both our revenue mix and cost restructuring. By segment, the past growth rose to 48.8%, continued to upward strength from the second quarter of 2025. SaaS and others maintained a strong gross margin of 70.8%, remaining above 70% level. Smart Solutions posted a gross margin of 23.8%, slightly higher than last year's 23.5%.

Overall, our Q3 performance in line with our expectations and continue to reinforce the profitability foundation at this stage. On the expense side, we continue to maintain prudent and disciplined financial management. Even as both our scale and profitability expanded, total operating expenses declined to $36 million. down 34.1% year-over-year. GAAP operating margins improved significantly to 4.6%, and GAAP net margins increased to 23.6 percentage points year-over-year to 18.2%, while ensuring that R&D investment in key AI initiatives and platform development remain intact and we continue to exercise strategic cost control to balance growth, quality and profitability.

On the cash flow front, operating net cash flow continued to grow steadily this quarter. reaching USD 30 million, a 25.7% increase year-over-year. Our cash collection cycles remain stable and the cash flow quantity materially improved. At the end of the Q3, our net cash balance stayed above USD 1 billion, giving us ample flexibility to balance shareholders' returns manage external uncertainties and support long-term strategic investment.

Next, I'd like to briefly highlight some recent progress in our AI capabilities and developer ecosystem, which serves as a crucial foundation for 2-year long-term growth. At the end of the Q3, Tuya's platform had 1.62 million registered developers representing a 23% year-over-year increase. AI adoption across smart devices also continue to accelerate. Commercial AI developers have collectively create more than 12,000 AI agents on the Tuya platform, covering a broad range of smart products categories, including toys and pet products, electronical, home appliances, IP cameras and wearables.

Meanwhile, we continue to deepen and strengthen our AI developer ecosystem, anchored by Tuya OS, Tuya open and T-Series AI developer Board. On the open source front, Tuya Open has seen steady growth in both documentation and code engagement. Since the beginning of this year, the GitHub repulsor store count has increased by about 80%. To date, over 2.3 million of codes have been contributed to open source projects. Beyond the rise of the Tuya developer participation, the overall quantity of the ecosystem is also improving significantly.

In summary, despite the prevailing external uncertainties, we still demonstrate strong resilience and operational agility. achieving solid financial growth and impressive profitabilities, which steadily advance in the AI plus IoT developer ecosystem across our core business segments. Thank you all, operator. we can begin Q&A session right now.

[Operator Instructions] Our first question comes from the line of Yang Liu from Morgan Stanley.

I have one question regarding the business outlook with more and more state deals that in down in the international trade market, what is the business outlook going into fourth quarter this year, which is a big season. And also what is your early look for customers' demand going into 2026?

Thanks, Liu. I have to share 3 points. So the first 1 is that this year, we still see that will be the kind of the soften demand on the growth side. And because of the uncertainties on the global macro economy situations this year. And so this year, the Q4, we'll see that the regular promotion season will be kind of the softer versus the dose. So we will keep closing on to review that while we already have the stable -- kind of the stable turns across multiple countries, that's whether the demand will begin to return steady on December.

So that will be the short term. And for 2026, what we see here is that because like Jerry shared earlier before that all those kind of AI features and smart home portfolios become more and more inevitable trend for the entire sector so which means that more and more animal consumers are already starting to familiar with this type of products that really become the beginning users of these type of things. And all the major brands and the players, manufacturers in industries already starting to enter these sectors and bring that into their the growth factors. So those type of trend will never stop.

So for 2026, we'll have a very positive outlook about the growth -- keep growing the entire business sectors. And the third 1 I'd like to show is that by reviewing all the technology improvements in the past decade, And -- we'll review AI will be 1 booster that bring the IoT experience into NextNav -- because in the past, the smart home experience, it's majorly focused on the connectivities, some automation and control. But while coming out with the AI capabilities, the user experience will come into a next level to more friendly, more easy to use and more smart. And so that's why we decided to provide a new AI system for Life, which is connecting the all the home scenarios and to ordinary people and have more people be able to enjoy the smart devices experience.

So that will lower another bar for the entry user. So combine that 3 together, the short times, we'll see that 2025, there's still some uncertainty and pressure on that. But it's become more and more inevitable and become a default options for major brands and players there. And we're trying to bring the [indiscernible] for more users who are not become the smart devices user as well. So I think that will be the overall. It's very positive in the long term and constant in short term.

Our next question comes from the line of Timothy Zhao from Goldman Sachs.

Congrats on the solid results. I have 2 questions here. One is regarding the AI home agent that you just mentioned. Just wondering if you can share more color on the detailed specs and the use case of these AI agents that you are going to officially released at CES next month? And how do you think about the impact on the overall business of Tuya with this new product?

Secondly, it's about the AI overall impact on your PAT and [ SatSolution ] business. Just wondering, for example, for the segment growth this quarter, would you please break down in terms of by volume and by pricing? Has AI brought any positive impact on the pricing of your product and services and also on the impact on the gross profit margin.

Okay. Thank you, Timothy. So the first 1 is that so we'll define this as an AR system, so it's bigger than an agent. And because we think that if we review the life scenarios. Even only for home, that you find that you have a multiple since you want someone to help you with. So this is AI Assistant to come with multiple agents that can help you to do almost everything you need in a home. So that's the first one. That's how we designed this new assistant. And the key value for that part will be in 2 things. The first 1 is that while coming on with the adoption of the [ Gimi ] an app, including the GPT, including China, et cetera, that you found that the air can help you to do a lot of things, a lot of tasks on the software side. But there is no assistant focus on home.

That's what you need for your home and how you on taking a of the home. So for us, is that we design the different type of agents and capabilities focus on those scenarios, people want to interact and people want to have a better life quality or easier life experience in home. That's the first one. And the second 1 is that the key differentiation of this assistance among any generic assistant is that this assistant will naturally be able to interact with the physical scenarios through the hundreds of millions of the [indiscernible] devices. So which means that we're trying to bring kind of science fiction to come true, like the drivers in Ironman's house, every people appreciate that. Every people, I mean, admire that. but there is no that type of jobs yet. So we want to create that type of experience for the global people.

So that's how we define the key features and value for the user side. And I think the -- what does it mean for the ordinary users. I think the key part is that right now, we found the smart devices is still kind of complicated is it become way more easier than 10 years before, but still kind of communicated to -- I mean, to [indiscernible] to use, to interact with by many nonuser. I mean, for those beginnings. So those parts still there. But coming out with the assistance, so you don't have to learn to use the app anymore. And you don't even -- you know how to speak, right? Like how you can tell the housekeeper to do something, how -- to servant to do something. It's a similar like the assistant will be to take the orders and due to all those complicated operation for you.

So we believe that will lower the bar significantly to the -- for the new users for home. And when we see that while the penetration of even smart home is still in a low no digit. And by [indiscernible] the entry bar we'll be able to open more doors for those new users while they found that the smart devices will be accessible for them to use. So that's for the AI system part.

And the second question is for the AI. And so first 1 is that this year, we consider at the beginning of the year of the AI device. So we are very happy to see that finally, our execution to the market, to the developers, to the customers or really starting to offer some feedback. So like the numbers we shared before by end of the Q3, over 93% of the products we shipped this year already been turned on some AI capability. So which means that my customers, my developers already very actively to try whatever AI features or capabilities they can provide through their devices, even their existing devices.

So that's the first one. So we really have a lot of innovative developers tying to try the ideas and try to educate end users and test end users' feedback. And will be that will be up very, very typical strong point for any new technology adoption. And so we really have that kind of scale test build to complete it. And the second 1 is that we still provide the AI seamlessly through our 3 business models. So including the past, including the solution, including the SaaS, right now, we have different type of AI offerings in different business model as well. So which means that for my customer side, on the procurement perspective. So they don't have to learn how they will be able to put something from differently. It's a similar like offering, but come with different features.

And maybe it comes with a different pricing, maybe not. So for that part, is that we try to open -- have almost all my customers [indiscernible] be able to try to bring AI into their existing products and solutions. And through that seamless integrations into my existing business models, we believe that, that will help in 2 things. The first 1 is that coming on with the new feature set, any new feature set will win new demand. So that will be able to speed up the penetration and adoption of the entire market. And we long for to have the AA coming as a booster.

And the second 1 is that with some really new feature sets that we reprice that and that improve our GP as well. But we're looking forward to have the GPM impact coming very soon because it's just in the beginning. So we try to promote the market. We try to incubate in the market in the beginning but not running really aggressively on the profitability side on that type of niche sector. So that's the overall outlook, Timothy.

Our next question comes from the line of Mingran Li from CICC.

Congrats on the solid readout, and 2 questions from my side. First is following adjustments of racing global tariff policy, could management share more color on the downstream or the recovery programs in the overseas markets, especially North America. My second 1 is that could management share the latest progress on the AI technology, particularly in terms of commercialization?

Yes. So the first 1 is that a couple of weeks ago that we get a temporary 1-year turns between China and U.S., right? And so which means that all the merchants in portals right now, they have they stable cost levels at a specific timing. And so that will be good things, at least we get some certainty. But the promotion for the issue will be pretty locked in. So those kind of new terms will be able to impact for next year's demand. So we're looking forward to have -- to be a positive impact.

And right now, the customers and importer size, this deal kind of review, okay, what will be the time for next years and we'd like to review what will be the turns out sell-throughs for this promotion season starting from this week, right, we'll have the Black Friday this week. So we're looking forward to have more feedback on December, like I described. And while people already know that what cost they can get for next year over a year, and what will be the demand looks like? And then how they set the teams for the new projects and the new sell-in reordering.

So that's the first one. So it's still under review. And the second 1 is for the AI. I think that I already answered part of that to Timothy earlier for the earlier questions. So the first 1 is that right now, we're offering AI across almost all my categories have some generic capabilities, works on anything. And we also have some differentiated vertical AI capability for a specific type of the product. But all those kind of orphans are seamlessly into my existing 3 services, the past, the solution and the SaaS. That's the first one. And in this year on the new device side, including the past and solutions, we really see -- happy to see some breakthroughs into some new sectors like the toys, and we shared that earlier last quarter as well.

So this will be a totally new vertical categories come with a large total addressable market size and that we don't touch before, and the IoT never get to be able to enter the sectors. But coming on with AI. So right will open the door. And in this year only, we run in 3 quarters only. That's many of the key players in the industry, starting from China, in the toy industry and already started to cooperate with us. And in Q3, we really helped the customer to launch a lot of the use cases to test the demand.

And it turns out that the end users love it. And so I would say the trial calls for many of the customers works out. So we're looking forward to continuous to improve the experience and also, the customer start to be ordering and running a new type of promotion classes or sales channels to scale it. So that's what we see that AI open new doors.

So that's the second one. And the third 1 is that not only upgrades kind of upgrades on existing categories and open new categories. So the third 1 we try to open is the 2C experience. So we use -- we were looking forward to using the new AI assistant to open all the new home user stores, especially for those ones who still don't have any smart devices, they still consider that type of devices will be kind of complicated for them. So we're using assistant to have up -- that's it.

Our next question comes from the line of Matt Ma from Jefferies.

So I just have 1 question regarding Smart Solutions. So the Smart Solutions revenue declined by around 14% in third quarter, just wondering what is the reason behind it? And then could management provide any growth outlook for the segment in 2026? And also, any thoughts on product category expansion going forward?

Yes. So I think the first 1 is that in 2026, we're looking to have a better year versus 2025 because we should have less turbulence for the economy side on the global basis. And so like I described that the customers right now in many vertical sectors, the customers already think that the AI features or ALT features will become more and more default for them. So like some categories that every single new projects they've been doing, they have to come down with the IoT. So we become to take a larger portion in their pie. So that's the first one.

So we will see that the penetration will grow. I mean for the overall industries, we continue to grow steadily, no matter what. It's only a matter of of speed, which year will be the tipping point. And so that's the first one. 2026 and we keep closing on. We think that we can share more colors around the second half of December, while the customers have more feedback on the end demand side. And while they're starting to have the tubes for 2026 because they don't like to run it in a very conservative operation and base for a long time. They really run in for 2025. So that's I think for the first one. And the second 1?

Yes, it's also regarding Smart Solutions. So I just want to understand what is our thoughts on product category expansion for Smart Solutions going forward?

Sorry, I missed 1 part. And so I think that for smart solutions, and we're very carefully to looking for the expansion to new categories because we're really learning that business model for over 2 years. So I think that for Smart Solutions, we're still kind of focused on strategically highly value categories. And for those ones that the AI can bring to difference, like bring some innovative ideas to come true without AI. It never exists and also to some categories that we're really helping customers to do a differentiation to have them out.

So really, the solutions is the 1 we design for the customers for their flagship model. So that's what we put out. So right now, the solutions, the major categories will be the video and related market modeling capabilities, the control panels that's super comprehensive interactions on the touch panel side and including the gateways, focused on specific scenarios and energy.

So I think that's for the middle term that we continue to put focus on scale of those kind of verticals. Unless we see some opportunities with scalabilities in some new vertical categories.

There are no further questions at this time. I will now hand the conference back to the management team for closing remarks.

Thank you, operator, and thank you all for participating on today's call and for your thoughts. If you have any further questions, please feel free to reach out to our IR team. We look forward to speaking with you at our upcoming investor events. Thank you, everyone, and have a great day.

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