KANZHUN LIMITED

Ladies and gentlemen, thank you for standing by and welcome to CanChun Limited four-quarter and fiscal year 2025 Financial Resource Conference call. At this time, all participants are in a listen-only mode. After the speaker's presentation, there will be a Q&A session. Today's conference is being recorded. At this time, I'd like to turn the conference over to Ms. Laura Chan, Senior Manager of Investor Relations. Please go ahead, ma'am.

Thank you, operator. Good evening, good morning, everyone. Welcome to our fourth quarter and fiscal year 2025 earnings conference call. Joining me today are our founder, chairman, and CEO, Mr. Jonathan Peng Zhao, and our deputy CFO, Ms. Wenbei Wang. Before we start, we would like to remind you that today's discussion may contain forward-looking statements, which are based on management's current expectations and observations that involve known and unknown risk uncertainties, and other factors not under the company's control, which may cause actual results, performance, or achievements of the company to be materially different. The company caution you not to place undue reliance on forward-looking statements and do not undertake any obligation to update this forward-looking information, except as required by law. During today's call, management will also discuss certain non-GAAP financial measures for comparison purpose only. For a definition of non-GAAP financial measures and a reconciliation of GAAP to non-GAAP financial measures, please see the earnest release issued earlier today. In addition, a webcast replay of this conference call will be available on our website at ir.gping.com. With that, I will now turn the call to Jonathan, our founder, chairman, and CEO.

大家好,欢迎参加公司2025年第四季度及全年业绩发布会。 我请代表公司全体员工 Hello, everyone.

Thank you for joining our company's fourth quarter and full year 2025 earnings conference call. On behalf of the company's employees, management team, and board of directors, I would like to extend our sincere gratitude to our users and investors who have continuously believed in and supported us.

Today's presentation will cover four main parts.

First, our financial results for the fourth quarter and the full year of 2025. Second, the robust growth momentum we are seeing in both supply and demand during the spring recruitment season this year. Third, an update on our AI progress. And fourth, further strengthen our commitment to shareholder returns.

首先,简阳汇报公司2025年第四季度和全年的业绩情况。 2025年第四季度,公司实现20.8亿元收入,同比增长14%, First, let me briefly review the company's performance for the fourth quarter and full year of 2025.

In the fourth quarter, we generated a total revenue of RMB 2.08 billion, up 14% year-on-year. Excluding share-based compensation expenses, our adjusted operating profit reached RMB 900 million, up 37% year-on-year.

四季度是中国传统的一个招聘淡季, 但是剔除季节性的因素后, 且端招聘需求延续了年初以来的一个稳步恢复的趋势。 求职和招聘的供需比也都保持在一个健康的区间。

The fourth quarter is traditionally a low season for recruitment in China. However, after adjusting for seasonality, the hiring demand from enterprises continued the steady recovery trend since the beginning of this year. The supply and demand ratio also remained within a healthy range.

从行业看,制造业电子通信半导体表现突出。 服务业汽车和消费品零售增速相对领先。

By sector, manufacturing, electronics, communications, and semiconductors industries stood out. Urban service, automotive, and consumer retail industries saw faster growth. The Internet technology industry achieved a faster year-on-year growth rate in the fourth quarter compared to the third quarter.

In the whole year, the company achieved a revenue of 82.7 billion yuan in 2025, with a net profit of 12.4%, a net profit of 27.0 billion yuan, and a net profit of 33.8 billion yuan, with a net profit of 45.7 billion yuan.

In full year 2025, the company generated a total revenue of RMB 8.27 billion, up 12.4% year-on-year. Net income reached RMB 2.7 billion. Excluding other incomes such as investment gains and share-based compensation expenses, our adjusted operating profit reached RMB 3.38 billion. up 45.7% year-on-year.

Our adjusted operating profit margin reached 40.8%.

The upward growth momentum in users continues. In 2025, the company acquired nearly 46 million newly verified users. As of the end of 2025, the platform had cumulatively served over 250 million job seekers and 36 million enterprise users, with the total number of enterprises exceeding 20 million.

In the year of 2025, the average number of users for the BOSS-based app was 60.7 million, which is a 14.5% increase. The platform provided more than 22.7 billion users with an exchange based on both parties' consent or contact information, which is what we call the achievement in our daily business. This number is a 22.4% increase.

In 2025, the average monthly active users reached over 60.7 million, representing a year-on-year increase of 14.5%. In 2025, the platform facilitated over 2.27 billion instances, of mutually consented exchanges of resume or contact information, what we commonly refer to as mutual consent, representing a year-on-year growth of 22.4%. The average number of mutual consents per job seeker also rose by 7% year-on-year. This demonstrates our strong double-sided network effect on our platform.

In 2025, the company will continue to stay in the blue. low-end cities and small and medium-sized enterprises. The increase in the number of users in Lanling has led to an increase in the ratio of Lanling's income. The demand for Lanling's recruitment is also steadily warming up. The increase in the number of industries in the field of Internet technology, communication, and semiconductors has led to a significant improvement in 2024, showing structural improvement.

In 2025, the company continued its momentum in penetrating the blue-collar sector, lower-tier cities, and small and medium-sized enterprises. The growth rate of blue-collar users led the overall user base, with their contribution to revenue further increasing. Meanwhile, white-collar recruitment demand also recovered steadily, with sectors such as internet technology and communication Send conductors achieved higher growth rates compared to 2024, reflecting a structural improvement.

From the city's performance, the four-level three-line and lower city income ratio is close to 25%. It can be said that the four-year time has been reversed. This fully shows that our user infiltration strategy in the lower city market is working well. From city-tier perspective, revenue contribution from third-tier and low-tier cities approached 25% in the fourth quarter, doubling compared to four years ago.

This clearly demonstrates that our user penetration strategy in lower-tier markets is steadily gaining traction, while our brand awareness has continued to expand. In 2025, revenue contributed by enterprises with fewer than 100 employees exceeded 50% for the first time. The substantial contribution from SMEs becomes a key driver of the platform's continuous growth.

In terms of commercialization, up to the end of 2025, in the past 12 months, that is, the whole year of 2025, the number of corporate users we paid reached 6.83 million. The share price increased by 11.6%, and the share price increased by 1.3%. The pay rate of active users remained low, and was affected by the increase in the proportion of income of small and medium-sized enterprises. The pay amount of single customers, namely APU, remained stable overall.

On the monetization front, in the 12 months ended December 31st, 2025, which is full year 2025, the number of paid enterprise customers reached 6.83 million, up 11.6% year-on-year, and 1.3% quarter-on-quarter. The pay ratio among active users continued to see modest growth. Due to the increased revenue contribution from SMEs, the average revenue per paying user, which is the ARPPU,

remained broadly stable.

接下来说第二个方面,今年我们看到春招呈现出一个供需展望的良好势头。 Second, we are seeing robust momentum in both supply and demand sides during the spring recruitment season this year. 众所周知,研究行业的人都知道,中国的求职招聘的旺季

每年就是在春节之后发生。 As we know, the peak season for recruitment in China is after Spring Festival every year.

所以我们按照农历对齐的方法进行了研究。 今年春节之后,日新增的用户规模包括招聘者和求职者均好于去年同期水平。

Adjusted for the Chinese New Year holiday, the average daily newly verified users, including both job seekers and enterprise users, exceeded the same period last year, and active users upwards after spring festival reached a record high.

The average daily job postings in the 15 days after the Spring Festival grew by a double-digit percentage year on year.

In terms of supply and demand dynamics, the ratio of job seekers to enterprise users, which is the CQB ratio, on our platform, remained within a healthy and reasonable range.

From the industry point of view, manufacturing, electronic communication, semiconductor, car, radio, media, service industry, and other areas, active job increase is ahead of other industries. Considering that this year's Spring Festival is relatively late, compared with last year, But industry sectors,

Active job postings in areas such as manufacturing, electronics, and communication semiconductors, automotive, advertising and media, and urban services grew at a faster pace compared to other industries. Given that the spring festival fell later this year compared to last year, and it has been only a few weeks since the holiday period ended, white-collar and large enterprises typically resume operations slightly later than blue-collar and SME sectors. However, based on our current data, both newly added and active job postings in fields such as internet, AI, and technology are already showing signs of accelerated year-on-year growth compared to the same period last year.

第三项内容公司在AI上的进展

First of all, with AI's rich potential, the company offers a significant increase in the opportunity to return services. The purpose of the recruiter is to have the right person to work for them.

The purpose of the applicant is to be able to work for the right position.

In this sense, since the birth of Internet recruitment, the value provided has been is a small part of the whole thing.

Third, our AI progress. First, with the empowerment of AI, the company now has significantly greater opportunities to deliver closed-loop services. For enterprise users, the goal is to bring the right people on board. For job seekers, it's about securing a suitable position. In this sense, the value that online equipment has provided since its inception

represents only a small part of the big picture so far. In the past, online recruitment advertising platforms primarily offered value through information dissemination between supply and demand side.

Now, platforms like BossGPin focus on increasing the profitability and opportunity of establishing labor contracts. between enterprise users and job seekers. Moving forward, the online recruitment industry has the opportunity to evolve toward a closed-loop model. AI can assist companies in completing the last mile of work.

In 2025, with the support of AI, the company will try to open up a closed business service. The revenue has reached the level of 100 million. Since then, this part of the business has exceeded the other businesses on our platform in terms of speed of development. In fact, AI can help not only our company, but also AI can help the entire economic operation. We can all see the compatibility efficiency of any position with people. From the perspective of our business, AI can better digitize the needs of both sides, better improve the efficiency of the grassroots personnel, and better reduce the administrative costs of large teams. In this way, large-scale closed service can become a reality for the Internet recruitment platform.

By 2025, with the support of AI capabilities, the revenue from closed-loop services has already reached the scale of hundreds of millions RMB. To date, this segment has been growing faster than other businesses on our platform. In fact, many have recognized that AI can improve the matching efficiency for annual roles in the economy. In simple terms, AI enables large-scale closed-loop services by better digitalizing both sides of demand, enhancing the efficiency of frontline workers, and reducing the management costs of large teams.

In terms of AI, I would like to talk about two aspects. One is about B, and the other is about C. In terms of B, it is based on some of the higher recruitment requirements, and at the same time, there are also higher payable users. We've been using a tool called AIS2CHAO to achieve a good result. We're currently expanding the scope of the experiment. The essence of the function can be simply explained as better memory and analysis. Specific customers, recruiters' needs and preferences are automatically searched in our 2.5 billion recruiters' warehouse. Another progress, I will describe it between the enterprise user side and job seeker side.

The AI quick hiring tool developed for high-end users with demanding recruitment requirements and strong willingness to pay has achieved positive results and is now being scaled up for broader trials. The core function of the tool is to better remember and understand the specific needs and preferences of targeted clients, automatically conduct searches within a database of 250 million job seekers and the continuous interaction with users to improve hit rate. The feature demonstrates high readability, and we believe that it holds strong commercial potential.

For C-DAN, one of our experiments is to provide AI-assisted interview functions for the interviewer in the boss-oriented interview room, which improves the user-reliability of our interview room. What I mean is that its usability is clearly improving. From the JobSeeker side, our AI-assisted interview feature experiment has been fully rolled out to users within both GP's interview rooms.

This has significantly increased both the usage and reusability of the company's proprietary interview rooms, while also enabling the platform to collect a greater volume of high-value interview data.

The third aspect is related to our strategy. Recently, we have taken a more aggressive attitude in the process of using AI Agents to help specific recruiters. To put it simply, for example, we have expanded the use of AI-based full-time management and selection functions in similar scenarios. Of course, for recruiters, the efficiency is significantly improved.

At the same time, we are still insisting

The third progress is that the company has adopted a proactive approach in applying AI agents to assist specific enterprise users. By leveraging AI hosting and screening capabilities, protected delivery efficiency has significantly improved. in a situation like bulk hiring scenarios such as for manufacturing workers and large streamers. This progress makes a business model based on successful hires or interviews viable on a large scale. We consistently inform job seekers that they are interacting with AI agents. We believe that the job seekers' response is quite positive and enterprise users' satisfaction rate is high.

In terms of our progress in AI, I'd like to report on our progress in AI science. In terms of science, the scientists of our company's NANBEIGE laboratory have obtained some top-of-the-line explorations in the field of small model intelligence and ability. . . . . .

有一段时间站在文本模型榜的榜一。 这是一个我们在科研上的一个成绩。 The last approach in terms of scientific research, the scientists at our company's Nanbei Ge Lab have made some industry-leading exploration into the intelligence and the capability ceiling of small models. Among these, the recently open-sourced Nanbei Ge 4.1-3B model with only 3 billion parameters, surpassed models from the Qianwen 3 series ranging from 4 billion to 32 billion parameters on multiple evaluation tasks. We are not intend to compare with others on this scale. In late February, it ranked first in HubInFace trending list for the tax generation model.

The last thing I want to report is our plans for the return on stock. Today, the board passed a resolution. The company's plan for the next three years, starting in 2026, is not lower than the previous year in terms of stock repurchase and share price per year. It has adjusted 50% of the net profit. At the same time, we will pass a resolution The return of $2.5 billion will go up to $400 million. The company will increase the return and respond to changes in the market environment. I think this reflects our company's confidence in itself and, in some cases, even from a long-term perspective, we are willing to use greater effort and sincerity Fourth, our plan for shareholder returns.

Today, the board passed a resolution approving a plan to allocate no less than 30% of the prior year's adjusted net income for dividends and share repurchase over the three-year period starting from 2026. Meanwhile, the company will increase the share repurchase program upper limit from $250 million to $400 million, initially approved in August 2025, who will scale up repurchase as appropriate in response to changing market conditions. And we believe that it fully demonstrates our confidence in the company's long-term growth, as well as our long-run commitment to delivering value to our shareholders and friends who supported and believe in us.

That concludes my part of the call.

I will now turn it over to our Deputy CFO Wenbei for the review of our financials. Thank you. Thanks, Jonathan. Hello, everyone. Now let me walk through the details of our financial results of the fourth quarter and the full year of 2025. We are pleased to deliver a quality set of financial results for the fourth quarter of the full year, with revenue growth continued its accelerating momentum and enlarged profitability. In the fourth quarter, the recruitment market sustained its steady recovery trend throughout the year. Under such conditions, our revenues reached RMB 2.1 billion during this quarter, with growth rate accelerating to 14% year-on-year and grew by 12.4% to RMB 8.3 billion for the full year. Our number of paid enterprise customers for 2025 expanded to 6.8 million, marking a 12% year-on-year growth and 1.3% increase compared to the 2012 month ended September 13th. We witnessed a more balanced growth this year, with faster revenue growth from both key accounts and small-sized accounts. In the fourth quarter in particular, revenue from small-sized accounts grew 21% year-on-year, outpacing midsize accounts and key accounts. As a result of this structural impact, overall R pool remains stable, but R pool for each segment also increasing trend. Moving to the cost and expenses side, our total operating cost and expenses decreased by 7% year-on-year to RMB 1.4 billion during the fourth quarter, and by 7% to RMB 5.8 billion for the full year. Total share-based compensation expenses decreased by 23% and 20% year-on-year, respectively, in this quarter and the full year. As percentage of revenue-wise, SBC went down by 5 percentage points year-on-year, both for the quarter and the full year, and we also expect this trend to continue. Excluding share-based compensation expenses, our adjusted income from operations grew by 37% to RMB $900 million. and 46% to RMB 3.5 billion respectively for the quarter and the full year. The adjusted operating margin for the fourth quarter reached a quarterly historical high of 43.3%, while the annual adjusted operating margin improved by 9 percentage points to 40.8% in 2025, continuing to demonstrate our strong operating leverage and unwavering commitment to financial discipline. Cost of revenues decreased by 1% year-on-year to RMB 309 million during the fourth quarter and decreased by 0.4% to RMB 1.2 billion for the full year. This decrease was primarily due to decrease in employee related expenses and rental expenses, partially offset by an increase in payment processing cost. As a result, Our gross margin went up by 2 percentage points to 85.1% in 2025, marking the second year of increasing. Sales and marketing expenses decreased by 9% year-on-year to RMB $389 million during this quarter and decreased by 18% to RMB $1.7 billion for the full year. The decrease for the quarter was primarily due to improved sales efficiency, which led to a decrease in employee-related expenses. No major market campaign in this year also further decreased sales and market expenses for the full year. RMD expenses decreased by 8% year-on-year to RMB $406 million this quarter and decreased by 9% to RMB $1.7 billion for the full year. Including share-based compensation expenses, our adjusted RMB expenses decreased by 4% year-on-year to RMB 1.3 billion for the full year, and stayed relatively flat year-on-year and sequentially for this quarter. GMB expenses decreased by 7% year-on-year to RMB 256 million during this quarter, which was mainly due to decreased employee rate expenses. J&A expenses increased by 10% to RMB $1.2 billion for the full year, primarily due to a one-off impairment of intangible assets booked in the third quarter. Income tax expenses were RMB $165 million this quarter, up 81% year-on-year. This was primarily driven by higher income before income tax expenses. as well as the provision of a top-up tax of RMB 38 million related to the OECD Pillar 2 Global Minimal Tax Rules. Our net income reached RMB 692 million in the quarter and 2.7 billion in 2025. Our adjusted net income increased by 25% year-on-year to RMB 906 million in the fourth quarter. and 33% year-on-year to RMB 3.6 billion for the full year. Adjusted net margin continued to expand in 2005 and reached historical high of 43.6%. Net cash provided by operating activities amounted to RMB 1.3 billion during the fourth quarter and increased by 29% year-on-year to RMB 4.6 billion for the full year. And our cash position stood at RMB 19.9 billion as of December 31st, 2025. We will leverage our strong cash position and the profitability to invest in future growth initiatives as well as shareholder returns. In 2025, we declared a dividend of US dollar 18 million. And as of year to date in 2026, we have already repurchased a total of US$50 million worth of shares. Furthermore, as Jonathan just announced, the Board has approved a new shareholder return policy that for each year in the next three years starting from June 26, we will allocate no less than 50% of the company's adjusted net income of the preceding fiscal year for dividend distribution and share purchases. and upsized our share purchase program to a total of US dollar 400 million through August 28, 2027. And now for our business outlook. For the first quarter of 2026, we expect our total revenues to be between RMB 2.050 billion and RMB 2.085 billion, a year-on-year increase of 6.6% to 8.4%, This Q1 figure reflects a different signality in this year, as the later Chinese New Year meant a shorter window of the peak recruitment season fell within this quarter. This timing difference aside, our underlying momentum remains strong. With the strong demand we are seeing throughout the spring season, we expect a clear acceleration in revenues over the coming quarters. That concludes our prepared remarks, and now I would like to answer questions.

Operator, please go ahead. Operator, we can open for questions. Hello? Hello, operator. Hello, Eddie, are you able to speak now? Okay. Eddie, I think your line is open. You can start the presentation.

Our first question comes from Eddie Wang from Morgan Stanley. Eddie, you may now speak.

Thank you, President Zhao, Wenbei, and Loach. Thank you for accepting my question. My question may be related to AI, because recently, whether it's China or the United States, there has been a lot of discussion about AI. I believe that investors and the market I think the entire company's influence is mainly focused on two aspects. So I'd like to ask Mr. Zhao to share his thoughts. First, the spread of the AI agent application, will it greatly reduce the demand for white-collar recruitment in the future and affect the long-term growth of the recruitment market? Mr. Zhao, I'd like to share your thoughts on this. The second question is about the rapid development of AI technology. will it be a big challenge for our company's competitive advantage and business? I will translate it myself. Thank you, management, for taking my question. Recently, we have seen a very extensive debate around the narratives of AI disruption and substitution. I think the market and investors have two key concerns regarding the impact of AI on the industry and on our company. So I would appreciate your insights on two questions. Firstly, is that will the adoption of AI agents significantly reduce the future demand of the white-collar hiring and the jobs and affecting the long-term growth potential of the recruitment market in China? And the second question is, does the development of AI technology pose a very big challenge to Boss3Ping's competitive advantage and its business mode? in the future.

Thank you.

Thank you for your question.

I also made certain observation and consideration that I noticed in America, both domestically, some discussions about AI capability to play a human labor force, especially for white collar. And you know, the open clock is quite popular recently. Everybody is talking about open clocks. So that might take some opportunities for human work. I noticed that.

I think my own thinking will be more based on the market I am facing. The first point I want to make is that I think the development of AI, the help of AI for human labor efficiency, I think for China,

But my own thinking is that more based on the market we are facing, my first opinion is that the development of AI, the improvement AI can have to the human labor efficiency, I think is a very good supplement to the overall labor force in the coming 20 years.

We found some data that we could verify with each other. We obtained some public data and did a simple calculation. According to 16 to 59 years old, which is a relatively broad labor population age, in 2026, it will be around 8.8 billion. According to the actual situation this year, if we calculate until 2036, this data will drop by about 9%. By 2046, this data will drop by about 20% to 6.9 billion people. This is the result of our data and a calculation. In fact, the drop of this data by about 9% to 20% in 10 to 20 years, I think it's a pretty big, pretty serious challenge. So what's currently on the table is really

so we have made some quite reasonable and practical calculations that if we consider age between 16 and 59 a broader age for labor force that number in 2026 is about 880 million and uh According to the current situation, we refer that in 2036, that number will go down by around 9%. And until 2046, the number will further go down by 20% to 690 million. So this is just a result based on our calculation and some factual data. So in fact, this 9% to 20% in the coming 10 to 20 years, that decrease in the number, I think is quite challenging. quite a severe challenge. And what we can see can help with this situation is actually the application of AI in this daily work.

In addition, we would like to say that there are two very different places in the United States where there is a trend of replacing workers with AI. First of all, the ratio between blue-collar and white-collar is not the same. In the U.S., there are about 156 million people. White-collar has to account for one billion, so it's a 2-to-1 ratio. From the perspective of China, it's the other way around. There are more blue-collar people.

And another thing different about this AI discussion, replacing discussion in the US, that two different points between US and China. First, is the ratio between blue collar and white collar is different. So there are around like 150 or 160 million labor force in the US. and the white collar has around 100 million, so it's like a two to one ratio. But from China's perspective, it's the other way around, and which means we have more blue collar population.

Another different thing is, let's take a look at the situation of enterprises, that is, the situation of investors. In China,

We all agree that businesses that are active and operating are at least at a level of 40 million. Based on the public data we've obtained, 98% of businesses are small and medium-sized enterprises. From the perspective of a small and medium-sized enterprise, Maybe you still remember why many very successful SaaS companies in China with more than 98% of small and medium-sized enterprises are not easy to be recognized. Because compared to nine people, it may not be as difficult.

So I think we can all agree that I have more than 40 million companies, and I employ about 4.2 billion people.

On average, I hire more than 10 people per company. Because of AI's development, I have to hire 10 people instead of 9 people. I think this can be attributed to the past. A lot of corporate service software is not successful in China. I think this is determined by the size of our corporate employees and the stage of corporate development.

So another different point between US and China is the situation of the enterprises, which is employers. So there is a common upgrade that in China, the number of active operational enterprises is around like 40 million. So according to the public data, we can get 98% of which need small or even micro-sized enterprises. So from a SMB's perspective, But maybe you can record a lot of very successful SaaS companies and not that easily to achieve same kind of success in China. Because where in China, 98% of companies are SMBs, when they're hiring people, for example, they're hiring 8% compared to 9%. The difference is not that huge and it's not that difficult. So it's the same situation here. around 40 million enterprises in China hired around 420 million urban employees. So on average, one company hired around 10%. Because the development AI, the company might hire 9% instead of 10%. I think that can explain why a lot of software or SaaS companies are not accessible in China. That is actually determined by the size of the a company and the different stage of the development of the company.

Of course, I was thinking about this discussion just now. In this generation, the development of AI and the development of AI agents will be discussed in the so-called large-scale process under the assumption of hiring and hiring workers. Next, I would like to talk about the actual data I saw. One of the data is that everyone knows that the development of AI has brought a wave of technology companies in China. We actually see an increase in the demand for talent related to AI. After the Spring Festival, the demand for jobs increased by 172% compared to the increase in the demand for artificial intelligence. Of course, the discussion before was based on the assumption that the AI was largely taking jobs from the human beings.

And now I will share another real data I have observed recently that this AI development has brought about a very good about the entrepreneurs of technology companies and also assert in demand from AI-related talents. So this year after the Spring Festival, from the demand of the enterprise customers, The AI-related, artificial intelligence-related newly posted jobs grew by 172%, 72% year-on-year, and the active online jobs increased by 80% as a very huge improvement.

After talking about the direct recruitment of AI talents, let's talk about the word AI-class. This word is not invented by me. You can recall the word Internet plus. Yes. There are a lot of companies, not only in the first-tier cities, but also in many provincial and city-states. They are hiring people with AI skills or backgrounds. They are adding new positions. I will call them AI plus. This position is becoming more and more common in city-states.

Now about the direct AI, beyond direct AI related jobs, we have another concept about AI plus, if you can recall the concept of internet plus. So more and more companies not only in the first tier cities, but also more and more like second tier cities in the center of the provinces. The companies are recruiting talents or background related to AI and increasing, creating new roles. Now we can call that AI class for now. Actually, those jobs are more and more commonly adopted and more and more further going into lower tier cities and more and more jobs

common or lower in the salary range? ZHANG ZHANG ZHANG ZHANG such as information semiconductors and cars. In fact, the increase in speed is still ahead of the increase in speed of the entire platform. In addition, I also saw some researchers take the data of the Yangtze River, take the data of India, or whose data to look at, And now for the overall number of jobs for the white collar, I actually mentioned in my report earlier that this year,

the overall number of white-collar jobs increased, especially like internet, like information communications, semiconductors, and automobiles. They actually are outgrowing the overall platform. And if you compare it to the other side in the U.S., from Indeed, their data also showed that the number of white-collar jobs increased this year.

And I quite appreciate your concern about our own challenge arised by the AI itself.

We have been paying a lot of attention to this since the beginning of Community 3, since ChatGPT 3.5 moment.

First of all, I think one thing that we can't change, that AI can't change, is that BOSS is actually a human game. It's a game against humans. Both sides are living people. Living people have to find a job and raise a family. Living people have to invite a brother to come back to work. First, one thing has not changed that AI also cannot change.

that bot shooting is actually a platform for the game of people, for a common base, the human being, the both side are like people who are looking for a job to provide for his family. So the person need somebody to help him to do the job and both sides need to be real person, from the supply side and demand side. So that's a marketplace for real person of majority of the population.

In addition, as our analysts and investors are familiar with, we actually have a... In the past, it was a prediction. Now, it's a reality. We already have a big double-sided network effect. As a result... As a result, as our users increase, so do our single-users. I don't think we have felt we didn't feel that our own advantages were challenged by AI. I even think that AI technology will help us to overcome these advantages.

Another point that analysts and investors who have known us quite well know that one point, which is a prediction in the past, but it's a reality now, that we have a very strong bilateral network effect, which caused that in addition of our users continuing to grow, the individual achievement a user can get also increases. So, until now, we haven't felt any challenge arrived posted to ourselves. On the contrary, due to our unique data, I think AI can help us to better solidify these advantages.

Another thing I mentioned in the report is that we have already clearly seen With AI, we can enter a billion-dollar trading market. Since 1997, there has been an online recruitment industry. For the first time, the online recruiting platform has been able to earn a lot of money. I think what I see in this is that AI is helping.

And another point is, as I mentioned in my prepared remarks, that we now can clearly see that with the implement of AI, we can move forward very aggressively into this placement market with multi-hundreds of billions of market size. which is actually since the beginning of online recruitment industry in 1997. And now the first time for the online recruitment platform to have opportunity to achieve like a business model based on results. And AI is actually helping us with that.

So, Eddie, so you see, I actually don't have too much to... I don't think I'm particularly suitable to... There are unique perspectives or there are a lot of research to talk about this prediction for the future. Prediction, in the end, this AI can give us globally, give us these white collar, what kind of disaster will it bring? I have no way to predict this thing. My answer is based on the work I really do and the things I see every day. Just as an answer to you, Eddie.

So Eddie, you can see I don't have any unique point to predict, or I'm not well-positioned to predict what kind of influence or might even disastrous AI can point about to the real world. I just provide you some observations based on current data, based on our daily work. So that's my answer to your question.

Thank you for the question. One moment for the next question. Our next question comes from Timothy Zhao of Goldman Sachs. Please go ahead.

Okay, Mr. Zhao, Wenbei, Laura, thank you for accepting my question. I have two questions here. The first one may be about some follow-up questions about AI. Mr. Zhao just shared some actions about AI last year. I would like to ask the first one is that in terms of AI products, What are the tests and plans for this year? When can we see more results in the currency of AI products? In addition, we also pay attention to the model of Nanbei Pavilion. It has made good progress recently. Please tell us the reason why we continue to invest and update Nanbei Pavilion as a small-sized model. What are the goals and applications we want to achieve through this model? This may be the first question. Thank you, Benjamin, for taking my question. The first question is a more follow-up question on the AI and AI progress. Could you imagine sharing What is your plan for the AI portal in terms of testing, in terms of new portal launch? And regarding the AI portal monetization, what kind of expectations that we should have for this year? And we also note that for your non-big or large-language model, just wondering what is the rationale behind your consistent investments? And what is your expectations in terms of the non-big or large-language model application this year? Secondly, it's regarding the recruitment demand or the recruitment trend after Chinese New Year. Just wondering if you can elaborate more details and especially compared to the previous years, what are the special points about the recruitment demand for this year after Chinese New Year?

Thank you. What money can be generated in the production process? I think it's pretty clear that last year, we made hundreds of millions of dollars in closed-end services. This year, we think there will be a few-fold increase.

I'm pretty sure of that.

Also, we can't say that it's an AI product. It's actually an AI ability to help. So if we take the scope of the discussion back to, with the help of AI capabilities, do we have this kind of currency-based products? Yes, we do. I would rather say that it is something that is widely used in every field of thinking.

So far, I still don't have this.

But we do have a lot in production.

This is about AI making money.

About AI commercialization, so the first thing is about the close-loop placement service, which last year we have already made like a millionth level, a hundred millionth level of revenue. And this year, we expect that number to double or triple, even have multiple growths. That's quite sure. Another thing is not like a native AI product, but I would rather say is AI is empowering all aspects of our operation and technology and the product to help us to provide better solutions and better user experience to our customers.

We still believe that we should keep investing in research. The company has large-scale AI applications, including resources, product managers, and programmers. At the same time, there is a small team in AI Science. There aren't many people in this team. We don't have that many chips. I think the rationality of this is that 我觉得我们有必要在一个 AI 发展如此迅速的过程中 保持住我们所谓的尾灯战略 就是一定要有一支团队清醒的 彻底的盯着最先进的公司 他们怎么了 在干什么 这是南北哥实验室 就是我们存在的一个原因 我们不能只是变成一个 下载别人一个开源的东西 然后科学家 我们家的科学家告诉我说 So actually, the company has made a very significant investment in AI applications, including

resources, product managers, and the technology team. In the meanwhile, we also kept a relatively smaller science team to develop our non-baker model. So the thing about keeping our own proprietary pre-training model, the reason is that I feel that we are necessary to keep our so-called tail light strategy during a world that AI is developing very, very fast. So we must have a team who can clearly follow the most advanced leading company to see what they're doing. So that's the reason why the Nambeg Science Lab exists. We cannot only become the company to only download open source platform. And then sometimes some scientists tell us If we can only download other person's stuff, then we might not understand what is happening. And download something, we cannot properly use that. That's why we kept this tail light strategy.

So we don't want to exceed, we just don't want to a lot. And this is definitely valuable for us in our actual work. We obviously make small models so that we can use them in our own industry.

There's no doubt about that. Another thing is... There are very few scientists in our family.

There are very few young people. I want to take this opportunity to talk about what they told me. They said... One day, it's not about who has more power consumption, who consumes more computing power, who has more data, and who has a higher annual salary. Through these four things, I'm afraid this is exactly what the large-scale model of this model has encountered. um Another point that is quite easy to understand why we are doing this smaller scale model

because it has a relatively very low cost for training and reasoning. And it needs more smart or intelligent ideas, which is quite useful to our daily work. And apparently, we are doing this smaller size model to further help our industrial applications that are out. And another thing is our scientists, which numbers relatively smaller but they are very very very young and talented i would like to take this opportunity to tell you something they are sharing with me that they say that one day there is one day it's not that compared to who can consume more electricity who can consume more computing power who have more data who can give a higher salary compared to all those four things actually is the obstacle that current paradigm is facing for the current model. But so actually the science is not to belong to the money, it belongs to the people who have very strong experience. So actually, when our scientists were talking about developing a smaller model, I didn't tell them to do it by tomorrow for my industrial application. I want to say that we are not a very wealthy company. We're not. But in this progress of human technology revolution, we once have taken the number one of the Hugging Face text ranking. means that a smaller size company can express their own voice during this huge technology revolution.

Yes, this is... Yes, AI, we are all in it. And then the number of seasons after the Spring Festival, I think... Let's have our guest, CFO Wenbei, talk about it.

Yes, about the spring festival recruitment season this year, as Jonathan just mentioned, this year we actually saw very strong recruitment demand with both new and active enterprise data showing significant young year growth and the CP ratio remained stable. I'd like to point out there is one difference this year that the later Chinese New Year actually will result in weaker year-on-year data for the fourth quarter, especially on a gap basis, because there are actually, if you count, there are 20 fewer days for revenue recognition after the holiday compared to last year for things in the first quarter. So that might have a bigger impact on the gap numbers for the quarter. But on the full year, if you're looking at the lunar calendar alignment, our cash question growth is positive. And we are expecting to accelerate further throughout the year compared to 2025. And yeah, that's all to the question. And given the time constraint, I think we can take the last question.

Certainly. allow me to open up the last question. The last question comes from Wei Xiong of UBS.

Please go ahead. Thank you for accepting my question. I also have two questions to ask. First, can you ask the management to update our user growth target this year? And considering the additional marketing and AI investment this year, how should we look at the outlook of the profit and loss in 2026, including the pace and space to improve the future profit and loss? Second, I also want to ask about the progress of our overseas business. Because recently we also saw that Offer Today's user growth in Hong Kong is very fast. What is our next step? I will also translate it myself. So thank you, management, for taking my question. Firstly, could management update our user growth target for 2026? And considering the additional investment in sales marketing and AI this year, how should we think about the margin trend as well as the future for margin expansion? And secondly, regarding our overseas business, could management please give an update there as well? We know there's very rapid user growth of offers today in Hong Kong, so what's our plan for the next stage of development? Thank you.

Thank you for your question. I will answer the first one regarding margin and then Jonathan will take the ORC question. So for 2026, we are still looking at a very robust user growth. At least we want to add at least 40 million of newly verified users this year. And in terms of investment, as you said, we might further invest in AI like computing powers, buying servers and maybe renting some computing power. And about the World Cup, we have not decided yet. But even considering those, we expect that given our strong leverage of our operating operations, we expect the overall adjusted operating margin can slightly increase compared to last year, even taking into consideration of the World Cup and AI investments. From the long term, we still have quite a huge upside potential for our main recruitment domestic business, and we will maintain a steady growth year on year going forward. However, as Jonathan just mentioned, if we have opportunities in new initiatives like closed-loop placement, AI policies, we will not hesitate to invest, and we will communicate with the market in that impact But at least till now, we are expecting our margin to improve suddenly in the coming years.

Thank you for your attention to the development of OfferToday. Actually, these few times in Hong Kong will have a little different feeling than before. Especially now. OfferToday is on the moving end. DAO, if we compare it, it is currently the number one in Hong Kong. And if we add up the number of workers, there are more than 3 million workers in Hong Kong. We can say that every 50 workers in Hong Kong have a day of off-today to look for opportunities. Some of these people have already passed off-today. Thank you for noticing offer today.

So this time we're coming to Hong Kong to do the earnings. I have a slightly different feeling, especially now offer today has now ranked the number one. in Hong Kong in terms of daily active users on the mobile end. If you compare different platforms like mobile and PC, so among the 3 million workforce in Hong Kong, I can say every 50 Hong Kong workers, one of them is using offer today daily to looking for opportunities. Some of them have already secured a new work opportunity through offer today, but now When I'm working on the road, I'm looking at all the potential users for me. I felt quite happy.

Offer Today are recognized in the Hong Kong market. This is a guaranteed success. Our next step is to work on localization, including two parts. The first part is to further localize the product. The second part is to further localize the team, especially because we hope that there will be Or for today, it's not a simple replica of Boss shipping in China mainland, but the core is actually the same.

So in some how, the Boss shipping as a native app has received a certain recognition in the Hong Kong market. Our follow-up job is to further push forward the localization, including two parts. First is the localization of our product, and the second is the localization of our team, especially we are expecting more and more native Hong Kong young talents to join us, becoming the leader, manager of this product.

After talking about our offer today, let's talk about our internationalized business. And move beyond the offer today about our other international approaches.

I can say that we will be more active to looking for opportunities to cooperate with local institutions for opportunities like that. Actually, we are making some progress but still not timing yet to make the report. I just want to clarify, Hong Kong is part of China, so beyond the overseas market. So that's all the answers for the question.

Thank you. Due to time constraint, that concludes today's question and answer session. At this time, I will turn the conference back to Laura for any additional or closing remarks.

Thank you once again for joining us today. If you have any further questions, please feel free to contact us. Thank you.

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