Tencent Holdings Ltd

Tencent Holdings Ltd

44.16
USD
-1.97
(-4.27%)
496.99BMarket Cap

Q3 FY2026 · Earnings Call TranscriptNovember 18, 2026

Operator

Good day and a good evening. Thank you for standing by.

Welcome to Tencent Holdings Limited 2026 third quarter results announcement webinar. I'm Wendy Huang from Tencent IR team.

At this time, all participants are in a listen-only mode. After management's presentation, there will be a question and answer session.

For participants who dial in by phone, if you wish to ask a question, please press 5 on your telephone to raise your hand. If you are accessing from the Tencent Meeting or VooV Meeting application, please click the raise hand button at the bottom left.

Please be advised that today's webinar is being recorded. Before we start the presentation, we would like to remind you that it includes forward-looking statements, which are underlined by a number of risks and uncertainties, and may not be realized in the future for various reasons.

Information about general market conditions is coming from a variety of sources outside of Tencent. This presentation also contains some unaudited non-IFRS financial measures that should be considered in addition to, but not as a substitute for, measures of the group's financial performance prepared in accordance with IFRS.

For a detailed discussion of risk factors and non-IFRS measures, please refer to our disclosure documents on the IR section of our website. Let me now introduce the management team on the webinar tonight.

Our Chairman and CEO, Pony Ma, will kick off with a short overview. President, Martin Lau, will provide a strategy review.

Chief Strategy Officer, James Mitchell, will provide a business review, and Chief Financial Officer, John Lo, will conclude with financial discussion before we open the floor for questions. I will now pass it to Pony.

Pony Ma

Thank you, Wendy. Good evening.

Thank you, everyone, for joining us. Our recent significant achievement of base models and applications have accelerated Tencent's evolution for a new AI era.

Hunyuan 4 has achieved China's state-of-the-art AI model status, and Hunyuan 4 Flash brings down inferences cost to deliver cost performance for real-world use cases. We believe WorkBuddy has grown to become the largest AI application in China, measured by total daily large language model token consumption by users.

Our cloud services revenue growth rate increased, driven by WorkBuddy and as we allocate more compute to Model-as-a-Service and GPU rental. Our core businesses remained resilient and vibrant as several evergreen games reached lifetime highs in DAU, and as Video Accounts increased engagement rapidly, particularly among young users, by enriching its content offering.

Looking at our financial numbers for the third quarter: total revenue was 207 billion RMB, up 8% year-on-year; gross profit was 116 billion RMB, up 7% year-on-year; excluding new AI products, gross profit was 118 billion RMB, up 8% year-on-year; non-IFRS operating profit was 68 billion RMB, down 6% year-on-year; excluding new AI products, non-IFRS operating profit was 83 billion RMB, up 8% year-on-year; and non-IFRS net profit attributable to equity holders was 67 billion RMB, down 6% year-on-year. Turning to our key services: for communication and social networks, combined MAU of Weixin and WeChat grew year-on-year and quarter-on-quarter to 1.4 billion.

In games, Delta Force exceed 58 million peak DAU in October, demonstrating our ability to expand our new games' popularity for years after game launch. For productivity tools, WorkBuddy has become the most used AI product in China, measured by large language model token usage.

On the foundation model front, we released a production version of Hunyuan 4 earlier this month. The model ranked first among Chinese model based on Agent Arena leaderboard at launch, achieving China state-of-the-art level.

I will now hand over to Martin for the strategy review.

Martin Lau

Thank you, Pony, and good morning and good evening, everybody. In the past three months, we have achieved transformational breakthroughs in our AI-native businesses, and today I want to provide some important and exciting updates on our initiatives across intelligence, applications, and monetization.

Starting with intelligence, i.e., foundation models, we believe the latest generation of Hunyuan, Hunyuan 4, has achieved state-of-the-art capabilities among China foundation models, as well as exceptional utility in real-world productivity. Hunyuan 4 is a category leader among models of its size and possesses capabilities reaching those of even larger models.

At its launch, Hunyuan 4 ranked first among Chinese models on the Agent Arena leaderboard, for instance, and ranked high across a whole host of benchmarks with zero targeted optimization, which is quite normal in many other models. Our teams substantially scaled up pre-training and post-training for Hunyuan 4, and added vision and computer use capabilities.

As a result, Hunyuan 4 delivers significant gains in instruction following, long context understanding, long-horizon tasks, and deep research compared to our earlier models. To further bring down inference costs and drive broader adoption in more real-world use cases, we also released a customized lightweight model, Hunyuan 4 Flash.

Hunyuan 4 Flash is priced at 1/20 of Hunyuan 4. Its usage way exceeded our expectation and achieved the highest daily token consumption volume on OpenRouter of any model launched this year.

We're pleased with Hunyuan 4's capabilities and its velocity of improvement, and believe we're still at the early stage of our intelligence journey. By deploying Hunyuan models in key products including WorkBuddy, we capture high-value user feedback, enabling continuous model-product co-design and a virtuous iteration cycle, especially for real-world use cases.

We've also recently unified multi-modality development under Hunyuan team. We lead the industry in terms of 3D generation, and we're improving image generation and also developing video generation capabilities.

Moving on to application, we have also achieved breakout success in productivity AI with WorkBuddy. WorkBuddy has grown very rapidly to over 10 million daily active users, consuming over 100 trillion tokens each day.

We believe it has become the number one AI application in China in terms of large language model token usage. WorkBuddy is also by far the leader within productivity AI domain, and according to external surveys and internal data, has widened its lead versus peers in recent months.

While we have historically been cautious about how quickly the enterprise software market would develop in China, we now believe that adding AI capabilities will allow productivity software spending to rapidly and substantially increase, with WorkBuddy as a key driver and beneficiary. The rationale is as follows: Firstly, AI services avoid software piracy problem.

Secondly, WorkBuddy covers large number of valuable use cases within enterprises, such as document preparation, data analysis, project management, research, and software development, which makes the whole package worth paying for. A subscription plus token usage-based monetization model means that customers can unlock a paid experience at low flat entry price, but then spend more as they derive more value.

WorkBuddy's annualized gross revenue over the last 7 days has exceeded $500 million, which is more than 2x of Tencent Meeting, which was launched 7 years ago. In some ways, these dynamics mirror those of China's online game industry evolution, where cloud connectivity also curbed piracy, and an item-based monetization model unlocked a much larger revenue pool, when spending can scale up along the demand curve.

We have already seen enterprises significantly scale their spending up when forward deployment engineers demonstrate how usage can add value. We're working with partner companies to provide FDE support to WorkBuddy customers.

Since we released enterprise version in June, 60% of the top 100 companies in China have become WorkBuddy's customer. This strong early adoption is very encouraging, and we believe this is just the tip of the iceberg.

While we are currently investing to grow WorkBuddy adoption, including via providing temporary free services to certain categories of users such as school teachers, we're confident that we will progressively improve the unit economics for WorkBuddy over time, by increasing the paying user ratio, facilitating higher token spend per paying user, and managing down token costs via inference optimization, model customization, and compute efficiency improvement. By integrating WorkBuddy with Tencent and external ecosystems, we believe we can enhance WorkBuddy's user retention, lifetime value, and overall economics.

While WorkBuddy delivers depth in AI usage, Xiaowei, the personal assistant embedded in Weixin, will facilitate AI diffusion breadth, and make AI accessible to every single internet user in China. Since testing started in June, we have consistently expanded the user testing pool for Xiaowei, and now has achieved 30 million DAU.

We've added more Xiaowei touchpoints into Moments, Official Accounts, and Video Accounts, unlocking greater utility across a broader range of scenarios, and driving higher interaction per DAU. We're in the process of onboarding major Mini Programs merchants and integrating with Mini Shops in order to facilitate transactions.

Because Xiaowei is powered by the purpose-built WeLM model, customized for cost efficiency and optimized for the Weixin ecosystem, inference costs to run Xiaowei are very low, and will continually trend down. In fact, even when Xiaowei is available to entire 1.4 billion Weixin users, we expect the quarterly cost of Xiaowei will remain lower than the quarterly cost that Yuanbao was incurring earlier this year.

Xiaowei's capabilities will be significantly enhanced as compute efficiency multiplies every year. Over time, some inference workloads can be processed on user devices, and that would further reduce costs.

By leveraging Weixin's pre-existing network of merchants across payment and Mini Program services, Xiaowei is able to achieve powerful integration with merchants and service providers at minimal cost, and to deliver synergies rather than competition with them, fostering win-win relationships. On the monetization front, we expect Xiaowei to act as a multiplier that we apply to Weixin's already substantial revenue streams, compounding them further and faster.

Advertising and transaction-based revenues are immediate opportunities, and we can add new revenue streams such as agent-to-agent facilitation services revenue from merchants and service providers over time. We're at the early stages of adding compute capacity at an accelerated pace, and making a portion of that capacity available to external paying customers through Tencent Cloud.

As a result, our cloud services business is now starting to deliver financial benefits from our AI initiatives, initially from WorkBuddy, Model-as-a-Service, and GPU rentals. During the third quarter, cloud services revenue growth accelerated 35% year-on-year from 21% year-on-year in the second quarter, and teens level percentage year-on-year growth throughout 2025.

For WorkBuddy, revenue for the third quarter was over 5x the previous quarter, driven by increasing paying users and strong initial enterprise adoption. For Model-as-a-Service, revenue more than tripled quarter-on-quarter, with enhanced inference efficiency and strengthened operational controls and security guardrails for enterprises, so that they have oversight of their AI deployment.

Our GPU rental services revenue grew more than 30% quarter-on-quarter, driven by enterprise demand for inference and improved scheduling efficiency for our compute capacity. With that, I'll hand over to James for business review.

James Mitchell

Thank you, Martin. For the quarter, total revenue was up 8% year-on-year, with social networks contributing 15%, domestic games 23%, international games 9%, marketing services 21%, and FBS 31%.

Our gross profit was up 7% year-on-year, within which the VAS gross profit increased 7%, marketing services 10%, and FBS 8%. Value-added services revenue was 98 billion RMB, up 3% year-on-year.

Social networks revenue within that was down 1% year-on-year to 32 billion RMB, as revenue from long-form video subscription decreased due to fewer releases of top-tier content, while revenue from app-based game item sales and audio subscriptions increased. Domestic games revenue grew 11% year-on-year, primarily driven by Delta Force, VALORANT Mobile, and Honor of Kings.

International games revenue was down 9% year-on-year, or down 4% on constant currency basis, due to a high base for certain Supercell games and the release of Dying Light: The Beast in the third quarter last year. Moving to communications and social networks: within Weixin, our green bubble gateway for mini drama series achieved rapid user growth, providing free-to-view mini dramas that benefit from Weixin's social recommendations and from high-quality content, including China Literature novels converted into mini dramas.

Developers can now use WorkBuddy to automate their development and publishing of Mini Programs, lowering entry barriers and shortening development cycles. For Video Accounts, we boosted engagement among younger users via deeper content collaboration with our internal game, music, and TV services.

We introduced an AI harness for creators to convert Official Accounts articles into videos, automating storyboarding, scripting, and media generation. Video Accounts total time spent grew over 20% year-on-year, and we believe has room for further growth, noting that our time spent per DAU was still less than half that of the industry leaders.

For domestic games, Delta Force grew its DAU over 50% year-on-year during the quarter to a lifetime high. We deepened the game's core extraction experience with a new map, and we added a new 5v5 tactical shooter mode.

VALORANT Mobile DAU and gross receipts achieved lifetime highs, benefiting from popular blossom-themed and Chinese folklore-themed weapons. Honor of Kings average DAU also grew year-on-year, benefiting from a new aspect system which reimagines classic heroes, and from a fast-paced arcade mode.

Its gross receipts grew, too, driven by successful IP collaborations and by dark fantasy outfits. In September, we launched Honor of Kings: World, a hero card battler set in the Honor of Kings universe on mobile and PC.

The game has ranked among the top 10 mobile games industry-wide by average DAU since launch. Among our international games, gross receipts for VALORANT grew year-on-year in the third quarter, driven by increased player engagement and by the Champions 2026 collection.

The Champions World Finals events for VALORANT in Shanghai featured a pop culture festival of games, fashion, and outdoor experiences. We partnered with the developer of Steal a Brainrot to bring Brainrot IP into PUBG MOBILE's World of Wonder sandbox mode.

The resulting map, called Brainrot Royale, drove the largest DAU increase for the sandbox mode since launch, with 30% of PUBG MOBILE DAU playing in the sandbox mode. On October 29th, we launched Monster Hunter Outlanders, an action RPG for mobile featuring impressive monsters, the series signature combat experience, and a new adventure mode, and the game ranked among the top 10 mobile games by downloads globally in its first week.

For marketing services, despite industry-wide headwinds, revenue increased 19% year-on-year to 43 billion RMB, with growth in impressions and pricing. Consumer goods and local services categories notably contributed to the growth.

We introduced AI-powered advertising solutions for content publishers. For example, we enabled mini drama providers to auto-embed advertisements of optimal length at optimal times, minimizing disruption to user viewing while enhancing monetization efficiency.

These AI solutions now power approximately 80% of ad monetization for mini drama publishers. Video Accounts ad impressions grew over 50% year-on-year, driven by video views and ad loads.

We served advertisements based on our content ranking system now, which increases our ad recommendation relevance and monetization. The green bubble mini drama gateway contributed substantially to marketing services revenue growth within Mini Programs.

FinTech and Business Services segment revenue was 64 billion RMB, up 9% year-on-year. FinTech services revenue grew by a low single-digit percentage year-on-year, mainly due to increased revenues from consumer loan and wealth management services.

For commercial payments, volume increased slightly year-on-year as the number of transactions grew, partially offset by lower value per transaction amid softness in overall consumption. We launched payment app Tenpay Go for foreigners visiting China, which integrates overseas card and digital wallet support with transit ride codes, tax refunds, and AI-powered translation services.

Within business services, as Martin has discussed, our cloud services revenue growth accelerated to 35% year-on-year, benefiting from increased demand for AI-related services, as well as for our overall cloud services. We've integrated WorkBuddy into WeCom, enabling agents to use authorized workplace context to complete tasks, share outputs within the organization, and turn organizational knowledge into reusable capabilities.

Now I'll pass to John.

John Lo

Thank you, James. For the third quarter of 2026, total revenue was 207.3 billion RMB, up 8% year-on-year.

Gross profit was 115.9 billion RMB, up 7% year-on-year. Operating profit was 59.1 billion RMB, down 7% year-on-year.

Interest income was 4 billion RMB, down 6% year-on-year, due to lower average interest yield. Finance costs were 3.5 billion RMB, compared with 3.8 billion RMB in the same period last year, reflecting favorable forex movements, partially offset by higher interest expenses due to increased debt.

Our share of profit of associates and JV was 4.5 billion RMB, compared with share of profit of 7.8 billion RMB in the same period last year. On a non-IFRS basis, our share of profits of associates and JV was 8.8 billion RMB, compared with share of profits of 10.3 billion RMB in the same period last year, primarily due to lower profit from certain associate in the China internet industry, factoring in market consensus estimates and true-up adjustments.

Income tax expense decreased by 7% year-on-year to 9.1 billion RMB, mainly due to lower tax expenses associated with a certain overseas subsidiary. Excluding this impact, our income tax expense would have increased year-on-year.

On a non-IFRS financial figures, operating profit was 68.1 billion RMB, down 6% year-on-year. Operating profit excluding new AI products was 83.3 billion RMB, up 8% year-on-year.

Net profit attributable to equity holders was 66.5 billion RMB, down 6% year-on-year. Diluted EPS was 7.227 RMB, down 5% year-on-year.

Moving on to gross margin: for the third quarter, overall gross margin was 56%, broadly stable year-on-year. By segment, VAS gross margin increased by 3 percentage points year-on-year to 64%, driven by a favorable revenue mix shift towards high-margin internally developed games.

Marketing services gross margin was 52%, down 4 percentage points year-on-year, due to AI-related operating costs, as well as content costs associated with building up our mini dramas service. FinTech and business services gross margin was 50%, broadly stable year-on-year.

On Q3 operating expenses: selling and marketing expenses increased by 6% year-on-year to 12.1 billion RMB, mainly due to promotional efforts to drive a greater adoption of our AI-native products, including WorkBuddy. R&D expenses rose by 42% year-on-year to 32.4 billion RMB, primarily reflecting higher R&D spend to improve our Hunyuan model family, advance Weixin AI initiatives, and develop AI capabilities across our products and services.

G&A, excluding R&D expenses, increased by 2% year-on-year to 11.7 billion RMB, as we controlled non-AI-related expenses. Our third quarter non-IFRS operating margin was 33%, down 5 percentage points year-on-year.

Non-IFRS operating margin excluding new AI products was 40%, broadly stable year-on-year. To conclude, I'll highlight some key cash flow and balance sheet metrics: operating CapEx payments were 63.6 billion RMB, up 257% year-on-year, as we accelerated investment in AI infrastructure to support enhancement to our Hunyuan model family, WorkBuddy inference needs, and development of AI capabilities across our products and services, as well as to meet external demand for our cloud services.

Non-operating CapEx payments were 1.5 billion RMB. Free cash flow was negative 69.1 billion RMB, mainly due to large compute procurement.

Excluding prepayments for compute procurement, our free cash flow would have been positive 16.9 billion RMB. Net debt position was 13.7 billion RMB, compared with net cash of 58.2 billion RMB as at 30th of June 2026, primarily reflecting prepayments I just mentioned, partially offset by net cash inflows of 8.4 billion RMB arising from net divestment activity in our investment portfolio.

As at the end of September 2026, the fair value of our shareholdings in listed investees, excluding subsidiaries, was 505.4 billion RMB, and the carrying book value of our shareholdings in unlisted investees, excluding subsidiaries, was 397.2 billion RMB. Thank you.

Operator

Thank you, John. We shall now open the floor for questions.

If you are dialing in by phone, please press 5 to raise a question, then press 6 to unmute yourself. If you are accessing from the Tencent Meeting or VooV Meeting application, please click the raise hand button at the bottom.

We will take one main question and up to one follow-up question each time. The first question comes from Robin Zhu from Bernstein.

Robin Zhu

Hi, thank you, Wendy. Thanks management for taking my questions.

I guess if I could have two questions. One, we've now had two quarters of very high CapEx and prepayment spend.

Be curious, where is Tencent in the process of this impulse of spend on acquiring compute? How far along the process would you say are we to getting to a good level of compute, however you deem that?

Then how do increased US restrictions impact both existing and future deals for compute? Does that impact the acceleration we saw in cloud and elsewhere in the business going forward?

Then second, WorkBuddy has done very well. Be curious if management could dimension the opportunity from here in terms of user traction, is it mainly power users that are the kind of addressable market, and revenue profit potential?

Could this be at risk given the semis restrictions? Thank you.

James Mitchell

Robin, why don't I take the first set of questions with regard to prepayment, CapEx, and I think impact of advanced semiconductor restrictions on cloud revenue growth. Regarding prepayments, the second and third quarter were abnormally high levels, and we have now entered into the long-term contracts that we wanted to enter into, our prepayment outflows will decline.

In fact, there may be some quarters where prepayments reverse due to the cancellation of contracts that we previously prepaid, but we're now exiting as a result of recent changes. In terms of the CapEx separate from the prepayments, then the current run rate, we'd look at it on a cash CapEx basis, which I think was 65 billion RMB rather than the 81 billion RMB accrual CapEx, because if you're looking at prepayments and CapEx together, then you should look at prepayments and cash CapEx; otherwise, you're double counting looking at prepayments and accrual CapEx.

Anyway, looking at the cash CapEx level, we think that that is at peak levels, and cash CapEx may bounce up and down around this peak level through the rest of 2026 and 2027 on a quarterly basis, and then we expect it to reduce beyond that. As far as how the restrictions impact our cloud business, we actually don't see an impact because our cloud business today is primarily running inference, and in particular, running inference for highly efficient models such as HY4 Flash and other Flash models.

We're now at a stage where that inference can be run very effectively on a surprisingly limited number of domestic chips. I think in the last 12 months there's been some transformations in terms of the inference productivity, and consequently we're able to drive a far higher volume of tokens specifically on domestic chips than we would have imagined a year ago, and it is the inference that drives the cloud revenue.

Actually we think that our cloud revenue will not be substantially adversely affected by the changes.

Martin Lau

Right. Now, a few things to add to what what James has talked about right now.

Number one, in terms of the CapEx, I think, again, right now the CapEx that we spend right now is actually in order to in in infrastructure spent in order to get our AI-native business kickstarted, right? Now it's a little bit like sort of building a factory in in order so that we can do training and you can do token production, right?

The nature of it is that, once you have that, then you can actually keep using that facility to train new generation of models, and you can also sort of keep generating tokens. That's why, once we've spent that, right, in 2026 and 2027, then going into 2028, the the trend would get better because of sort of this nature, right?

Because you can you can keep using the the infrastructure that we've spent to build in 2026 and 2027. Secondly, the pricing that we actually see in 2026 and 2027 is actually elevated, right?

As we go into 2028, even if you buy need to buy the same thing, you should expect the price to come down quite significantly. Thirdly, which is actually I think the most important, is that, we now, after working on the, training and and inference of of our models, we we see that compute efficiency is actually multiplying quite quickly year-over-year.

That that means, we're going to be able to get higher training efficiency, as well as much larger number of tokens, as we go by the year, with the same facility. With these factors, we actually sort of, see 2028 to be much better and beyond.

Now, when when we look at the impact on on our our AI business in in relation to the the latest tighter restrictions, right, I I would say that it's unfortunate, but then, it it a few points, right? Number one, it does set everyone back, right?

It sets everyone back in China. Everyone is actually, having less compute, advanced compute to work with, no matter when when you actually started leasing or or building capacities outside of the US China.

Number two, when we look at training, then, it's going to be impacted, right, but, our training efficiency with Hunyuan 4 is actually very high, and when we look at the next version, right, it would take probably longer time for us to come up with the next version in the short run, but over time, the domestic supply chain would actually come on board. When we look at inference, as James said, it's actually very efficient models that we have built, right, and as a result, it should not be a problem that domestic supply chain should be able to cover it.

When we look at WorkBuddy as a product, then WorkBuddy works with very efficient models first of all, but it also works with a whole range of different models, and it gets a very diversified supply of tokens, some of them provided by us, some of them actually sort of provided by the model companies, that the various model companies that WorkBuddy hosts. As a matter of fact, WorkBuddy has become the biggest customer of almost all of the open-weight models in China.

Over the longer run, we believe, all Chinese players would actually have to rely on domestic supply chains and domestic chips, which we actually see the the industry is actually progressing very fast, and with the latest restriction, we expect it to actually speed up even further, because, all the development effort would actually be have to be focused on speeding up the domestic supply chain. Now, in terms of WorkBuddy, we are very, very excited about the the recent growth, and and, when when we look at the when we look at AI business, right, the characteristic of AI business versus the internet age is that, in internet age, if you double your business, right, and, that's actually sort of, considered very, great result, but, in the AI era, it actually sort of, usually see very successful products and models actually 10x growth, and I think WorkBuddy actually falls into that category.

We're very happy to see that kind of growth traction, and, WorkBuddy, it is is at scale, right, in terms of both the DAU and, and the DAU continues to grow very fast, and it's already the largest AI application in China in terms of token delivered, and if you look at AI business from an intelligence delivered basis, right, tokens as a good proxy for that, and it's already by far the largest AI application in China, and, the token growth is actually also growing and accelerating. We actually see it as a great model, a new model in an enterprise market.

We explain that, there's no piracy, you actually can't have a small team of people sort of, building a makeshift WorkBuddy within the organization, and, WorkBuddy basically provides a a whole package of different functionalities, it's not just a one software, it's almost like, a whole package of software functionalities delivered to you. I think probably most importantly, right, it's actually, a a great business model as we, start, working through the monetization that, the the free tier actually allows you to acquire users very nicely, and then the the paid tier actually sort of allows you to get some people to pay and to give them a a quite quite a significant amount of token, and then for the really heavy users, right, they actually keep on adding spending on the tokens, and and we have seen users spending up to like 40,000 RMB per month on buying tokens, right?

When you look at that, that actually sort of, is very different from SaaS businesses in China in the past in which sort of, you you you can only charge a flat fee based on subscriptions. We actually felt very good and excited about WorkBuddy.

Over time, I think, when when it, number one, it has a fantastic market position. Everybody sort of, think about this productivity AI, I think, thinks about WorkBuddy, and as it collect more and more data, it actually makes our model and also the models who work with us better.

At the same time, it, allows the users to have more and more skills and more and more of their work products, in their memory of WorkBuddy, and as it work into the workflow of businesses, we think, it will be very sticky. We are really enjoying the the, phenomenal growth of WorkBuddy.

Operator

Thank you, Robin. We will take the next question from Alex Liu from Bank of America.

Alex Liu

Thanks for this opportunity. I have two questions.

The first question is, at the recent Tencent 28th anniversary, I think the management talked about organizational change. Could you share a little bit more about the strategic thinking on this, and how broad we should expect the changes to be across different business units and resource allocations, and where should we expect the earliest financial impact to show up?

Martin Lau

We're at a very early stage of this, right, but the insights that we have seen is that, number one, AI really can be applied to very businesses to improve the efficiency of how businesses are run, but in order for you to be able to do that, you actually sort of need to have a pretty significant reorg of the organization, and also sort of you need to re-engineer a lot of the work processes. If we can actually do that, right, then existing businesses can actually use fewer people, right?

Then at the same time, what we see is, AI actually allows small teams to leverage AI to explore new ideas and build new businesses and build new growth drivers, right? If we can actually use fewer people for running an existing business, and actually sort of you can redeploy these people to build new growth drivers for the existing business, as well as for building new businesses, right, then I think that would be the ideal, right, and that's the whole idea behind this.

As we build our AI-native business, we get more and more excited about this opportunity, because sort of we see that in our model team, we see that in our WorkBuddy team, we also see that when WorkBuddy is applied to our customers, right, it actually helped to improve efficiency in a great way, right? That's why we basically say, this is the time for us to make bigger changes to the organization.

If you say sort of what are we seeing already, right, I think the fact that our model team was actually completely rebuilt from scratch, and within a year, they were able to push our model to top-tier position in China, that's one example. I think WorkBuddy team was really sort of a nobody, right, 6-7 months ago, but then sort of they built such an exciting business, I think that's another example, and we hope to see more and more of that over time.

Operator

Thank you, Alex. We will take the next question from Alex Yao from J.P.

Morgan.

Alex Yao

Thank you, management, for taking my questions. Two questions from my side.

Number one is on global AI footprint. As you guys take Hunyuan and WorkBuddy to users outside of China and grow cloud business internationally, how do you think about where AI workloads are best hosted between domestic infrastructure and the international regions, and how much flexibility do you have to adjust that mix as conditions in different markets change?

A second question is about the economics of the WorkBuddy. On one hand, we are glad to see a very rapidly growing user number and also token consumption, but on the other hand, we are also seeing a lot of very highly efficient, but also at the same time very cheap Flash models are introduced to WorkBuddy, which arguably will reduce the overall blended pricing for token consumption in WorkBuddy.

Can you help us to understand the economics between the volume growth and the pricing decline, and are you still seeing a great opportunity in terms of ramping up monetization, even when the per workload cost or price goes down? Thank you.

Martin Lau

In terms of sort of where the hosting should happen, right, I think it's a very complicated question, right, because there's restrictions on GPU usage, but I think I would look at, if there's market opportunities, right, we will be able to figure out a way to solve it, right? We see right now Hunyuan 4 and WorkBuddy's fantastic growth opportunity in China, but we also see a pretty good chunk of the usage of Hunyuan 4 is actually outside of China, right?

In order to serve that, we could serve a part of that from the inferencing capacity in China. Over time, we actually sort of want Hunyuan 4 to be promoted around the world, and we may find partners to do that, right, and there are inferencing partners around the world who can actually do that.

I think that's one aspect. In terms of WorkBuddy, over time, I think we see great opportunities for growing that outside of China as well, and it's a software, it can connect to different models, and as a result, right, it doesn't require necessarily a inferencing infrastructure to support it, because it's very diversified in terms of token supply.

I think that's that. In terms of cloud, right, if it's non-GPU-related, right, we can actually build capacities ourselves outside of China.

I think there's still a lot of such opportunities, because our cloud services is actually very cost competitive. When it comes to GPU, right, then unfortunately we can't actually offer high-end GPU anymore, right, but then we may be able to work with some partners in different places.

I think if we find business opportunities, which we do think for each of Hunyuan 4 and Hunyuan family of models and WorkBuddy, right, and cloud, we would be able to find a lot of growth opportunities outside of China because of the efficiency, because of the cost competitiveness, because of the quality, then we will be able to find the infrastructure to support it. In terms of the use of Flash models within WorkBuddy, I think the Flash models are fantastic customer acquisition models, right, because they offer us the ability to offer very cost-competitive tokens to free users, right?

Then over time, when free users actually sort of want to use additional capability, right, then there's ability for them to get into subscription. Then, over time, when people sort of need to do more and more complicated work, right, then they will move on to more capable models, right?

Then they'll spend more on tokens, right? I think the Flash model is actually a great way to acquire customers, whereas the more capable models are great ways to monetize more by delivering more value to the users.

This actually sort of perfectly fit with what we said about WorkBuddy, which is it actually monetizes at all the different points on the demand curve. It's only helpful in that regards.

Operator

Thank you. We will take the next question from Colin Liu from BNP Paribas.

Colin Liu

Hey, good evening. I have two questions.

The first one is related to international games. When could we see the international games revenue back to positive growth?

With GTA really taking a lot of air time and players' mindshare, do you think that would actually delay the potential timing of the inflection point? My second question is actually also related to one that's already been asked.

You said that reorganization would take time, but AI investment right now flowing into OPEX that weighs on the overall profitability. Are there any more imminent cost-saving initiatives and plan that you can share with us at this time?

James Mitchell

Hi, Colin. As far as your first question is concerned about international games, then I think that in fairness, the slowdown for us predated the run-up to GTA 6 and flowed from a number of internal as well as external factors that we're working to address.

Looking forward, we believe we have a good number of evergreen games, some of which are still early in their growth curve. We have some plans for new games that we hope and believe will become evergreen, and we also will revert to a more normal cadence of more sort of upfront sales-driven games next year, after a period this year when we didn't have any of those premium games.

That's on the international games. More broadly, the game environment globally is one where the rate of innovation has slowed somewhat, and now with AI, there's opportunities to reaccelerate innovation, and that currently is more around the production of in-game content and the pre-production for new games, rather than full production, post-production for new games.

Overall, we've been a leader in terms of bringing innovation for game technology, for game business models, for core gameplay, both in China and more recently internationally, and that's what we will seek to continue to do. If we do that successfully, then we think that our international game revenue growth will look after itself.

Martin Lau

In terms of your question about efficiency, right, I think, once again, last time I talk about how we look at our existing businesses, right, and I reiterate, right, which is we look at our business in two different parts, right, two different buckets. One is the existing core business, and the other one is the AI-native business, right?

With respect to the AI-native business, it's natural for a business like that to go through the stage of investment, and then it would actually start generating revenue, and then generating profit, and we have seen many times business like that happening in Tencent. Because of the scale of the investment, right, this time around, it get people a little bit worried.

I think we are actually very excited about what we have seen and what we have experienced in the past 6 months. There's a real breakthrough in terms of our model, in terms of our products, and now we can actually say that this business is actually on its way to become a really valuable business for us over time.

Then when we look at the existing businesses, right, then we feel that because of the advent of AI and the capability and enabling capability of AI and the read across from how we build the AI-native businesses, right, then there's actually a lot of efficiency that we can achieve within the core businesses. It will both be in terms of extracting more efficiency, but also enabling the teams to create more revenue and growth opportunities.

That would be another initiative that will be pushing along. We see, again, right, also great opportunities for this in the coming time.

That is how we look at your question about reorganization, yeah.

Operator

Thank you, Colin. We will take the next question from Kenneth Fong from UBS.

Kenneth Fong

Hi, good evening, management. Thanks for taking my questions.

I have two questions. The first is on Xiaowei.

As we start the pilot testing and also deepen its integration with Mini Programs, any early user feedback that you can share, and what kind of milestone we need to hit before a more broader rollout? Longer term, how should we think about the Xiaowei monetization timeline, the TAM, as well as the path for monetization across advertising or even subscription, especially if we look at the early read across from consumer agent like Meta View?

My second question is on the advertising gross margin that have contracted notably this quarter on AI investment and also higher mini drama kind of mix. Looking ahead, how should we model this ad margin trajectory, specifically how quickly can inference efficiency gain from the AI investment that offset this kind of AI cost drag, and how much of this mini drama content cost impact will be structurally drag in term of the near-term margin?

Thank you.

Martin Lau

In terms of Xiaowei, we have been gradually rolling out to more and more users, and we have also been expanding the number of touchpoints within Weixin to now more than 30 touchpoints, including Moments, Official Accounts, Video Accounts. What we have seen is as we add these touchpoints, more and more people start using them, and there are also more and more interactions per DAU.

Our DAU in terms of interaction with Xiaowei is now 30 million, and the number of interactions per DAU has been increasing as well. People feel in general, right, it's very convenient to use, there's enhanced utility across social and content scenarios, and a lot of smooth usage, right, within some convenient touchpoints.

Now, I think that that's that, right? A lot of people, if you compare it against Muse, would say, "Oh, it's not as powerful as Muse," right, which is the case by design, right, because we actually want Xiaowei to first come out not as a all-capable agent available to a small number of people who are willing to pay for it, right?

We actually want Xiaowei to be available to all Weixin users, and over time, it will become more and more powerful because of the connection with the different touchpoints, because of connection with the ecosystem, including the Mini Programs. We actually need to take time to have more Mini Programs to have integration with our Xiaowei, right?

A lot of the capabilities would actually sort of be built over time. At the same time, right, because of the cost concern, right, because if we want to make it available to all Weixin users, right, then there is a cost question, right?

That's why we use a small model, and we're confident that even if we roll out to all users, right, it would not be such a big cost, right? In fact, the quarterly cost, as we said, right, will be less than the Yuanbao cost in the beginning of the year, even if we make it available to all users.

Then the question is what are we going from here, right? The answer is we see clear tailwind for us because compute efficiency is actually increasing in a very significant way.

In fact, it's actually multiplying by multiples every year, right? You can expect for each of the next version of Xiaowei launched, the capability would actually jump in multiples times, right?

Given a year or 18 months, right, the capability of Xiaowei would be much better than it is today. At some point of time, it will be as powerful as Muse, right, but we can actually cover all the users within Weixin.

We can actually offer this service to everybody at a manageable cost to us. When that happens, right, we believe it would be a huge multiplier to the ecosystem value of Weixin, right, because then people can interact with their agents on a continuous basis, and it will be very automatically tied in with their daily lives, and they can conduct transactions, they can discover content, they can actually sort of manage the schedule, and they can do a lot of things within Weixin just by interacting with Xiaowei, right?

We believe if you look at the current revenue stream of Weixin, which is, if you combine everything, right, in the ten hundreds of billions of RMB a year, right, then adding 10% or 20% to that would be a very significant number, which will make Xiaowei hugely profitable for us. I think that's the overall equation.

We feel that we will see that in the next 6 to 12 months very clearly. Yeah.

Go ahead.

James Mitchell

Sorry. Kenneth, on your second question around the advertising gross margin: so we have been progressively upgrading our ad tech stack over the past many quarters, and sometimes we incur the cost of the upgrade at the beginning of a quarter, and we only get the benefit of an upgrade later in the quarter or in the subsequent quarter.

Some of that was true this quarter, but that's not new news. What was new news this quarter was really the success, the breakout success of the mini dramas in the green bubble gateway, and the good news aspect is that contributed substantially to our advertising revenue.

The bad news is that we're in the early stages of really kickstarting our mini drama ecosystem and encouraging a number of independent mini drama studios, whose content is not available on our primary competitors' application, to make their content available in green bubble. There is some front-loading of cost associated with driving that mini drama traffic growth, revenue growth.

The mini drama is also interesting, taking a longer-term perspective, because alongside mini games, it represents an example of how we're able to sort of light up these new closed-loop activities within Weixin. Games for some time now have been both substantial in-app purchase, but also substantial in-Weixin advertising revenue generators for us.

Mini dramas are now following in terms of advertising contribution, and the biggest opportunity, which we're still working toward, is with the Mini Shops and capturing more e-commerce on a closed-loop basis. The flip side of what I've just said is that our advertising revenue growth would have been slower had it not been for the upside surprise from mini dramas, and I think that's just a function of an overall difficult macroeconomic and specifically consumption and advertising environment.

At any rate, we have continued to grow through that environment. We've continued to outgrow the market, and with the benefit of AI, with the benefit of closed-loop activities including mini dramas, with the benefit of more traffic for Video Accounts and so forth, we intend to continue outgrowing the overall advertising market.

Thank you.

Kenneth Fong

Thank you, James and Martin.

Operator

Thank you. We will take the next question from Alicia Yap from Citigroup.

Alicia Yap

Hi, thank you, Wendy. Good evening, management.

Thanks for taking my questions. Two questions from me.

First is that we notice that the recent advances from some of the frontier models, including OpenAI, Gemini, Meta, and also the Chinese labs, actually suggesting that the sophisticated capabilities in the interactive storytelling and also this world generations and the AI-driven characters towards the gaming environment. Does Tencent actually view these developments as a competitive threat to the traditional gaming and engagement, or as a potential catalyst that to create a new generations of the AI-native games?

Where does management see the biggest opportunity or the risk for Tencent games over the next 3 to 5 years? A follow-up on the advertising comment just now.

I think, James, you mentioned about the softening macro consumption, and also the offsetting factor we also have the mini dramas advertising. How should we be thinking about, given we are actually coming out from the higher base as well, should we expect to face a greater normalization pressure in 2027, or do we think that the AI-driven monetization and the further market share gain will continue to provide some of the offsetting factors?

Thank you.

James Mitchell

I think on the second one, I answered that already, that we're optimistic about our ability to keep outgrowing the market, and I don't feel we're facing any particular base pressure. I just think that the advertising market has been weak for several months already in 2026, I wouldn't characterize 2026 as a spectacularly good year for advertising.

These things are ultimately cyclical. At some point, consumer confidence, consumer spending will recover, and when it does, then advertising spending will bounce, and so if anything, I'm probably see more upside than downside risk to 2027, but time will tell.

As far as your first question is concerned, then AI can be a threat or an opportunity for the game business, for the game industry globally. We have seen that in some categories of games, such as hyper-casual single-player mobile games, then that there have been sort of rashes of AI-created copycat games that have arisen very quickly as a result of people running decompilers on existing games.

Recently the big model laboratories have taken steps to prevent that decompilation or piracy process, which seems to have curtailed the AI copycatting, but we're very early in the evolution. I would say that the work involved, the expertise required to not only create, but also operate and sustain and grow over time a multiplayer evergreen game that is multi-platform, multi-modal is orders of magnitude greater than the work required to create a sort of a single-player hyper-casual mobile game.

Obviously those multiplayer gameplay-based, multi-platform, multi-modal evergreen games are where we make most of our gaming revenue, and so AI can accelerate aspects of that process, as I mentioned in an earlier question, such as the pre-production drafting effectively of what a game might be, but as of today, AI is not able to take over the mainstream production process or the mainstream game operation process. Thank you.

Martin Lau

I also think that the game industry is not short of games, right? The ability to provide more and more games actually sort of do not really change the dynamics in the gaming industry, which is how do you distribute and attract a lot of users on a sustained basis, right?

It actually will generate, I think, a new generation of AI-native game experiences, and these game experiences can actually be standalone, but they can also be integrated into evergreen game platforms, right? I think we have seen a lot of cases in which if you see an innovation, then the evergreen game platforms can integrate these game experiences into the platforms and sort of benefit from that great idea.

At the same time, the AI game generation tools can actually help to make the game platforms evolving into more and more of a user-generated content, right? It will be sort of both the first-party and third-party game platform, right?

You would have the professionally generated content by the game team, but at the same time, you actually sort of have a place for the users to generate their own content, right? I think we have already seen that doing very well for Peacekeeper Elite, right, and we expect with the advent of more and more of these tools, right, it would actually help to expedite the evolution of many of our evergreen games into evergreen game platforms.

That would be hugely positive for us.

Operator

Thank you. We will take the next question from John Choi from Daiwa.

John Choi

Okay. Thank you, and thanks for taking my question.

I have two questions. On AI investment returns.

I think with these new AI products being launched, we've seen the operating profit reduced quite a bit this quarter. With this new AI products already creating a meaningful drag in operating profit, what kind of milestones should we expect or look for to assess the returns on these investments, and when should we expect the earnings drag to start narrowing?

That's my first question. My second question is on your cloud revenue.

I think we've noticed that you've seen a meaningful acceleration to 35% this quarter, and I think management shared more details of the Model-as-a-Service, GPU rental. With all the limitation of compute, if we've seen both domestically and globally, how should we think about the quality and the sustainability of this AI-related cloud demand, and what kind of revenue and margin opportunity should we be expect going forward?

Thank you.

Martin Lau

In terms of the investment in AI, right, I think we, first of all, sort of have to make the investment, right? Then the benchmark is actually, then, are we creating valuable businesses and products from that investment?

Before looking at the financials, let's look at sort of, are we creating value from that investment? Before our Hunyuan 3 and Hunyuan 4 got launched, right, it's uncertain whether our training investments is generating valuable models.

Now with Hunyuan 4 and Hunyuan 4 Flash coming out and we see sort of the explosive endorsement of users by using up a lot of tokens, right, I think we put a check on it, right, and say, "Oh, the investment in training is now generating very valuable products," right? As we look into the future, the future generations of models would get more and more exciting.

Then, when we look at the products, right, and, oh, first of all, we actually sort of invest in Yuanbao, and you can say, "Oh, Yuanbao was not as successful as we expected." Now, Yuanbao is actually sort of back to not focused on growing user base, but focused on growing its capability, and it's actually sort of growing its capability very nicely, and recently it has launched the AI search function, which I think it's also sort of an addition to the overall product offering, but we actually reduce the spending on Yuanbao quite significantly.

Then the next product that we actually launched was WorkBuddy, and WorkBuddy has become a phenomenal success, right? Now we can actually put a check and say all the investment that we put around AI products is generating positive value.

We are in a very good position to keep expanding on these valuable businesses, and look for new opportunities, too, right? Over time, we know these products will generate, first of all, usage, and then revenue, and then over time, profit.

Now, exactly when would that happen, no one knows, right? We are really operating in a generational technology revolution, right, and I think it's very difficult to build a business like, oh, you build a building and try to sort of get rental return from it, right?

When we look at the investment that's made and we're building very valuable results, right, then we feel confident that over time, it will generate revenue and it will generate profit eventually.

Operator

Thank you. In the interests of time, we will take the last question from

Martin Lau

Then 35%, yeah, the cloud, there's another question.

James Mitchell

Oh, yeah, sorry, John. On the cloud revenue growth, I think your question was around whether limitations on advanced semiconductor importation would disrupt the cloud revenue growth, and in our view, that it will not be the case for us, because the drivers of the accelerated cloud revenue growth include WorkBuddy and include Model-as-a-Service, and for WorkBuddy and for the models that we're deploying within Model-as-a-Service, these are highly efficient models that are being deployed, and they can therefore be deployed on domestic chips, which are not covered by the new limitations.

In addition, and Hunyuan 4 Flash being 95% cheaper than Hunyuan 4 is an illustration of this point, there have been transformational strides made in the volume of tokens that can be generated by fixed compute or GPU capacity in recent months, and so that enables us to get substantially more mileage from the compute we already have in place, while at the same time, we're adding additional compute through domestic supply chain, and those two forces together should support our cloud revenue growth going forward.

Operator

Thank you, James. Management, we will take the last question from Ronald Keung from Goldman Sachs.

Ronald Keung

Thank you. Thank you Pony, Martin, James, John, and Wendy.

First question, want to ask about our profit growth of the core business. I see that the operating profit excluding new AI products, the profit growth has moderated.

Want to hear what are the biggest cost swings for our core business? Is it the mini dramas or attributable depreciation, and how should we think of this underlying high-quality profit growth that used to be the core business?

How should we think of the profit growth outlook into the next few quarters? Second, maybe can you talk about Yuanbao Search?

As we talk about the agentic and Xiaowei and all of that, I also see we're doing some initiatives on the Yuanbao Search, what are the thinking process of going back to the search vertical, and how are we innovating on that vertical? Thank you.

James Mitchell

Why don't I attempt to address the first question on the core or not new AI product operating profit growth, and there's two drivers you can consider. One driver is the weak consumption environment.

That shows up directly in slower growth for our FinTech services business. It also would have showed up in slower growth for our marketing services business, but as I mentioned, we somewhat offset that through the accelerant of the mini drama series, but the mini drama initiative had its own heavy costs associated with kickstarting that ecosystem.

We faced that macro challenge that adversely affected margins. Then the second factor is that, as you're probably aware, many component prices exploded late last year, with DRAM, for example, up around 400% year-on-year as of early this year, and so in this quarter, you saw core businesses, not just our new AI products, but our core products, eating some of the flow-through from that jump in input prices for components, especially memory components, earlier this year.

Now, we and I think many other companies have taken steps to address and manage more tightly our consumption of those components, in particular memory, and I think that, partly as a result, memory prices have now been flattening, and our expectation is that our core business will face those two cost pressures I mentioned for the next few quarters, and then move through them as we digest the period of abnormally high component costs, and as the macro environment stabilizes at some point. Those are the two reasons for the slower growth in core operating profit.

Martin Lau

In terms of Yuanbao Search, it's really a recognition that search is actually a very important capability in the AI age, and it's actually completely revolutionized with AI agentic capability. That when we keep on improving the capability of Yuanbao, the search capability within Yuanbao becomes stronger and stronger, up to a point that we actually felt, as an atomic capability, it actually can be offered to other internet and technology companies to generate revenue.

That's the reason why we actually sort of open it up and start offering it as a service and providing it to other companies. It's a reflection of, one, AI has been transforming the search capability, and number two, Yuanbao, in the process of providing its service, has build up a very significant capability in that area, and three, there's actually external demand.

We have launched it, and would, over time, see some revenue from it.

Operator

Thanks, Martin and James. Thank you all for joining our results webinar today.

We are now ending the call. If you wish to check out our press release and other financial information, please visit the IR section of our company website at www.tencent.com.

The replay of the webinar will also be available soon. Thank you, and see you next quarter.