KANZHUN LIMITED (BZ) 2026 Q2 法說會逐字稿

Ladies and gentlemen, thank you for standing by, and welcome to Kanzhun Limited second quarter 2026 financial results conference call. (Operator Instructions) Today's conference is being recorded.

各位女士、先生，感謝您耐心等候，歡迎參加看準有限公司2026年第二季財務業績電話會議。(接線員指示) 本次會議將進行錄音。

At this time, I'd like to turn the conference over to Ms. Laura Zhan, Senior Manager of Investor Relations. Please go ahead, ma'am.

此刻，我想將會議交給投資者關係資深經理Laura Zhan女士。女士，請開始。

Thank you, operator. Good evening, and good morning, everyone. Welcome to our second quarter 2026 earnings conference call. Joining me today are our Founder, Chairman and CEO, Mr. Jonathan Peng Zhao; and our Deputy CFO, Ms. Wenbei Wang.

謝謝接線員。各位晚上好、早上好。歡迎參加我們2026年第二季業績電話會議。今天與我一同出席的有：我們的創辦人、董事長兼執行長趙鵬（Jonathan Peng Zhao）先生，以及我們的副財務長王雯蓓女士。

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 risks, 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 cautions you not to place undue reliance on forward-looking statements and do not undertake any obligation to update these forward-looking information, except as required by law.

在開始之前，我們提醒各位，今天的討論可能包含前瞻性陳述。此等陳述基於管理層目前的預期與觀察，並涉及公司無法控制的已知與未知風險、不確定性及其他因素，可能導致公司的實際結果、表現或成就出現重大差異。公司提醒您不要過度依賴前瞻性陳述，且除法律要求外，公司不承擔更新此等前瞻性資訊的任何義務。

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 results, please see the earnings release issued earlier today.

在今天的電話會議中，管理層亦將討論若干非GAAP財務衡量指標，僅供比較之用。有關非GAAP財務衡量指標的定義，以及GAAP與非GAAP財務結果的調節，請參閱我們今日稍早發布的業績新聞稿。

In addition, a webcast replay of this conference call will be available on our website at ir.zhipin.com. With that, I will now turn the call to Jonathan, our Founder, Chairman and CEO.

此外，本次電話會議的網路重播將於我們網站 ir.zhipin.com 提供。接下來，我將把電話交給我們的創辦人、董事長兼執行長Jonathan。

(interpreted) Hello, everyone, and welcome to the Company's second quarter 2026 earnings call. On behalf of all our employees, management and Board of Directors, I would like to express our sincere gratitude to our users and our investors. Today, I will mainly focus on three areas: second quarter results, changes in Company's growth strategy, and shareholder returns.

(口譯) 大家好，歡迎參加公司2026年第二季業績電話會議。我謹代表全體員工、管理層及董事會，向我們的用戶與投資人致以誠摯的感謝。今天我將主要聚焦三個方面：第二季業績、公司成長策略的變化，以及股東回報。

In second quarter, the Company generated revenue of RMB2.4 billion, up 14% year on year. In terms of profitability, adjusted income from operations, excluding share-based compensation expenses, was RMB1.05 billion, up 19% year on year. Our adjusted operating margin was 43.8%, [up] (added by company after the call) 1.9 percentage points year on year.

第二季，公司實現營收人民幣24億元，年增14%。在獲利能力方面，剔除股份支付費用後的調整後營業利潤為人民幣10.5億元，年增19%。我們的調整後營業利潤率為43.8%，[上升]（公司於會後補充）較去年同期提升1.9個百分點。

As of June 30, the total paying enterprise customers trailing back 12 months reached 7.2 million, up 11% year on year. Several key operating metrics reached record highs in this quarter. Average monthly active users, or MAU on our app exceeded 70 million in the second quarter. The average number of matches received per job seeker increased both year on year and quarter on quarter. Once again, even as our user base grew, user outcomes also improved.

截至6月30日，過去12個月的累計付費企業客戶總數達到720萬家，年增11%。本季多項關鍵營運指標創下歷史新高。第二季我們App的平均月活躍用戶（MAU）超過7,000萬。每位求職者平均收到的配對數量，無論同比或環比均有所提升。再次證明，即使我們的用戶規模擴大，用戶成果也同步改善。

Next, I would like to explain how the Company's growth strategy differs from higher-tier and lower-tier cities. The second quarter of this year marks the fifth anniversary of the Company's IPO. Investors who are familiar with us will remember that, throughout the past five years, we have constantly maintained that the core driver of the Company's growth is user growth.

接下來，我想說明公司在高線城市與低線城市的成長策略有何不同。今年第二季是公司IPO五週年。熟悉我們的投資人會記得，在過去五年中，我們始終堅持公司的成長核心驅動力是用戶成長。

This is determined by the size of the market. China has nearly 500 million people in its urban workforce and more than 40 million active businesses. Since its founding, BOSS Zhipin has cumulatively served approximately 300 million users and approximately 22 million enterprises. Even from where we stand today, there is still considerable room to grow.

這由市場規模所決定。中國城市勞動力接近5億人，活躍企業超過4,000萬家。自成立以來，BOSS直聘累計服務約3億用戶及約2,200萬家企業。即便以我們今天所處的位置來看，仍然有相當大的成長空間。

Second, this is determined by our model. BOSS Zhipin pioneered the mobile recommendation and direct chat model. At its core, this model substantially lowered the cost of communication between recruiters and job seekers.

第二，這由我們的模式所決定。BOSS直聘開創了行動端推薦與直接聊天模式。其核心在於大幅降低了招聘方與求職者之間的溝通成本。

This low-cost model enables tens of millions of companies to shift from traditional recruitment to mobile Internet recruitment, thereby digitalizing and mobilizing recruitment on a large scale. For the vast majority of our enterprise users, the first time they used our services was also the first time they used online recruitment.

這種低成本模式使數千萬家公司得以從傳統招聘轉向行動網際網路招聘，從而在大規模上推動招聘的數位化與行動化。對於我們絕大多數企業用戶而言，他們第一次使用我們的服務，也同時是他們第一次使用線上招聘。

Third, this is determined by our strength and user needs. Double-sided network effects give the Company strong vitality. The larger the user base on both sides, the greater the variety of users, the more users express themselves, and the more users interact, the better we can serve them. The process of driving user growth is also the process of continuously producing digital oil for the recommendation engine. Over the past several years, we have consistently seen that, as monthly active users on both sides have increased, user outcomes have also improved. With the engine supported by AI, we saw not only that AI improves the engine's efficiency, but also that the engine helps AI quickly establish its data flywheel. Over the next five years, we will adopt different growth strategies for Tier 3, Tier 4, and Tier 5 cities and for Tier 1 and Tier 2 cities.

第三，這由我們的實力與用戶需求所決定。雙邊網路效應賦予公司強勁的生命力。雙方用戶基數越大、用戶類型越多元、用戶表達越充分、互動越頻繁，我們就越能更好地服務他們。推動用戶成長的過程，也是持續為推薦引擎產出「數位石油」的過程。過去幾年我們一貫看到，隨著雙方月活躍用戶增加，用戶成果也隨之改善。在AI支撐的引擎之下，我們不僅看到AI提升了引擎效率，也看到引擎幫助AI快速建立其數據飛輪。未來五年，我們將對三線、四線、五線城市，以及一線、二線城市採取不同的成長策略。

In Tier 3, Tier 4, and Tier 5 cities, the core driver of growth will continue to be user growth, and our most important objective will remain user penetration. In Tier 1 and Tier 2 cities, while continuing to grow our user base, we will add reasonable price increases as a growth factor. With regard to the pricing of our services, let me first take a look at the actual situation in the second quarter.

在三線、四線、五線城市，成長的核心驅動力仍將是用戶成長，我們最重要的目標依然是提升用戶滲透率。在一線、二線城市，在持續擴大用戶規模的同時，我們將把合理的提價作為成長因素之一。關於我們服務的定價，先讓我回顧一下第二季的實際情況。

Revenue in the second quarter was RMB2.4 billion. That is a 10-digit number. It looks good, but here in Beijing, for many jobs, the price of a one-month job post is just the price of two cups of coffee. The value of many, many mutual matches that happen in every month combined is only enough to buy one bottle of mineral water.

第二季營收為人民幣24億元。這是一個10位數的數字。看起來不錯，但在北京，對許多職位而言，一個月的職位發布價格只相當於兩杯咖啡的價格。每個月發生的許許多多雙向配對加總起來的價值，也只夠買一瓶礦泉水。

What do we mean by mutual match? For those who are less familiar with us, let me explain again. A mutual match on our platform is equivalent to a job seeker submitting an application to a specific recruiter on another recruitment platform, and that recruiter also confirming the acceptance of the application. That is what we call a mutual match.

我們所說的雙向配對是什麼意思？對於不太熟悉我們的朋友，我再解釋一次。在我們平台上的一次雙向配對，相當於求職者在另一個招聘平台向某位特定招聘者投遞了申請，而該招聘者也確認接受了該申請。這就是我們所說的雙向配對。

It is a two-sided confirmation. In Beijing, in Shanghai, in Shenzhen, in Guangzhou, in Hangzhou, in Chengdu, in many cases, one such match is worth only one bottle of mineral water at 7-Eleven stores.

這是一種雙向確認。在北京、在上海、在深圳、在廣州、在杭州、在成都，在很多情況下，這樣一次匹配的價值只相當於在 7-Eleven 便利店買一瓶礦泉水。

To make this easier to understand, let me cite data from the leading recruitment platform in a mature market. According to publicly available information, one click on that platform costs approximately $0.25 to $1, while generating one application for a basic role costs approximately $5 to $10.

為了讓大家更容易理解，我引用一個成熟市場頭部招聘平台的數據。根據公開資訊，該平台一次點擊的成本約為 0.25 美元到 1 美元，而為一個基礎職位帶來一份投遞申請的成本約為 5 美元到 10 美元。

I do not have data on how many applications for such a basic role result in one mutual match. If I assume, based on a high efficiency case, that a recruiter is willing to accept one out of every five applications, that would translate into $25 to $50 per mutual match. This, my friends, gives you an intuitive sense of two things. First, compared with developed countries, as the importance placed on talent increases, and the human resources services industry grows, there is considerable room for Chinese companies to increase what they pay for such services. Of course, this will take time. Time is a powerful force. One example is that today, the salary of a very good software engineer in China is roughly at the same salary as in Silicon Valley.

我沒有關於這類基礎職位的投遞申請中，有多少最終形成一次雙向匹配的數據。如果我基於一個高效率案例假設：招聘方願意接受每五份申請中的一份，那麼換算下來，每一次雙向匹配的成本約為 25 美元到 50 美元。各位朋友，這能讓你直觀地感受到兩件事。第一，與發達國家相比，隨著對人才重視程度提升、人力資源服務產業發展，中國企業在此類服務上的付費提升仍有相當大的空間。當然，這需要時間。時間是一股強大的力量。例如，今天中國非常優秀的軟體工程師薪資，大致已與矽谷相當。

Second, compared with one aspect of the enterprise expense in Beijing, I have seen that many enterprises have achieved mutual match at a unit price, which is about 1/10 of a bucket of mineral water. The total monthly employment cost of junior human resources personnel could buy 1,000 mutual matches. Therefore, we can see that compared with Beijing, the service price in our field also has some potential to be improved. Put differently, if we do not reform this, the human resources service industry is destined not to be valued by companies. It is destined not to receive high quality resources, and it might shrink.

第二，與北京企業支出中的某一項相比，我看到很多企業以大約一桶礦泉水 1/10 的單價就實現了雙向匹配。初級人力資源人員的月度用工總成本，可以買到 1,000 次雙向匹配。因此我們可以看到，與北京相比，我們這個領域的服務價格也有一定的提升潛力。換句話說，如果我們不進行改革，人力資源服務行業注定不會被企業重視。注定拿不到高品質資源，甚至可能萎縮。

Therefore, at the beginning of the second five years, the Company's growth strategy has changed, which is based on the first-tier market and some second-tier cities to improve the user experience while gradually increase the amount of customer payment in mature markets, including a reasonable increase in payment rates.

因此，在第二個五年的開端，公司成長策略發生了變化：以一線市場及部分二線城市為基礎，在提升用戶體驗的同時，逐步提高成熟市場客戶的付費金額，包括合理提升付費率。

This process has been sustained for a while. The result growth that we have seen, part of the reason is because of that. In fact, in the last quarter, we predicted that growth and profit growth in second quarter would be better. That is part of the reason.

這一過程已持續了一段時間。我們看到的業績成長，其中一部分原因正是如此。事實上，在上一個季度，我們預測第二季度的成長與利潤成長會更好。這也是原因之一。

This is the right time and right place to change the growth model. Everyone’s expectation has played a critical role, which is mainly reflected in three elements. First, the large-scale application of AI increased the platform efficiency.

現在正是改變成長模式的天時地利。大家的預期發揮了關鍵作用，主要體現在三個要素上。第一，AI 的大規模應用提升了平台效率。

Secondly, some big customers in the white collar or blue-collar factories agree very much - they believe that the AI powered interview, AI assisted resume filtering, and other competitive solutions will also help them.

第二，一些白領或藍領工廠的大客戶非常認同——他們認為 AI 驅動的面試、AI 輔助的履歷篩選以及其他具競爭力的解決方案也將幫助到他們。

Third, the combination of AI with our platform business is actually consistent with the pursuit of job seekers on the platform and with the recruiter. That is to achieve not the goal to do the recruitment, but to do a successful hire.

第三，AI 與我們平台業務的結合，實際上與平台求職者以及招聘方的追求是一致的。也就是要實現的不是「完成招聘」這個目標，而是「成功錄用」的目標。

This brings us to our closed-loop business. The closer our services get to the actual hiring stage, and the closer we get to charging based on the successful hire, the more this model approaches a closed loop. The Company will continue to invest in exploring this area. One point worth mentioning is that the revenue we received from our AI-enabled closed-loop business grew rapidly quarter over quarter in Q2.

這就引出了我們的閉環業務。我們的服務越接近實際錄用階段、越接近按成功錄用來收費，這個模式就越接近閉環。公司將持續投入探索這一領域。值得一提的是，我們 AI 賦能的閉環業務收入在 2026 年第二季度環比快速成長。

Let me discuss shareholder returns. The Board today passed a resolution approving the distribution of annual dividends of $230 million. Since the beginning of this year, the company has repurchased approximately $300 million worth of shares, representing more than 4.7% of its total share capital.

我來談談股東回報。董事會今天通過決議，批准派發 2.30 億美元年度股息。自今年年初以來，公司已回購約 3.00 億美元的股份，約占其總股本的 4.7% 以上。

In 2026, the Company's total shareholder returns through share repurchases and dividends amounted to $530 million, exceeding 100% of last year's adjusted net income, and also exceeding the 50% we previously committed to. We share the benefits of the Company's growth with shareholders.

在 2026 年，公司透過股份回購與股息實現的股東總回報達到 5.30 億美元，超過去年調整後淨利的 100%，也超過我們此前承諾的 50%。我們與股東分享公司成長帶來的成果。

That concludes my remarks. Next, our Deputy CFO Wenbei Wang will walk you through the financials in detail.

我的發言到此結束。接下來，我們的副首席財務官王文蓓將為大家詳細解讀財務情況。

Thanks, Jonathan. Hello, everyone. Now let me walk through the details of financial results of the second quarter of 2026. We continue to deliver our high-quality set of financial results this quarter marked by solid revenue growth and further improved profitability. Our revenue achieved accelerated trend reaching RMB2.4 billion, representing 14% year-on-year growth.

謝謝你，Jonathan。大家好。下面我將帶大家回顧 2026 年第二季度的財務業績細節。本季度我們繼續交付高品質的財務表現，體現在穩健的收入成長以及盈利能力的進一步提升。我們的收入延續加速趨勢，達到人民幣 24 億元，同比成長 14%。

Recruitment demand in the second quarter remained broadly stable. We drove revenue and profit growth through user base expansion and improved monetization from higher-value services. The number of paid enterprise customers increased by 11% year on year to 7.2 million over the trailing 12 months ended June 30, 2026.

第二季度的招聘需求整體保持大致穩定。我們透過用戶基礎擴張，以及更高價值服務帶來的變現能力提升，推動了收入與利潤成長。截至 2026 年 6 月 30 日的過去 12 個月，付費企業客戶數同比成長 11%，達到 720 萬。

Importantly, the paying ratio among active enterprise users improved for the fourth consecutive quarters, reflecting our sustained progress in monetization. ARPPU for the quarter increased 7% year on year, driven by more efficient and valuable services, including an expanding suite of AI-powered features which encourages higher customer spending. Revenue growth was broadly balanced across different account sizes this quarter with both key accounts and small-size accounts showing healthy momentum.

重要的是，活躍企業用戶中的付費率已連續第四個季度提升，反映出我們在變現方面的持續進展。本季度 ARPPU（每付費用戶平均收入）同比成長 7%，主要由更高效率、更具價值的服務所帶動，包括不斷擴展的 AI 功能套件，促使客戶提高支出。本季度收入成長在不同帳戶規模之間整體較為均衡，重點客戶與小型客戶均展現出健康動能。

Moving to the cost side. Our total operating cost and expenses increased by 6% year on year to RMB1.5 billion this quarter. Total share-based compensation expenses dropped by 19% year on year to RMB186 million.

接下來看成本端。本季度我們的總營運成本及費用同比成長 6%，至人民幣 15 億元。股份支付薪酬費用同比下降 19%，至人民幣 1.86 億元。

As a percentage of revenue, share-based compensation expenses continue its downward trend to 7.8% this quarter, down 3.1 percentage points year on year. We expect share-based compensation expenses as a percentage of revenue to remain at a high single-digit level for the full year of 2026.

以收入百分比計，股份支付薪酬費用延續下降趨勢，本季度降至 7.8%，同比下降 3.1 個百分點。我們預計 2026 年全年股份支付薪酬費用占收入比例將維持在較高的個位數水準。

In the second quarter, we sponsored the FIFA World Cup and increased our investment in AI-related cloud services. Meanwhile, our head count grew sequentially, driven by stable growth in recruitment demand. Despite these investments, our profitability continued to improve.

在第二季，我們贊助了 FIFA 世界盃，並加大了對 AI 相關雲端服務的投資。同時，在招募需求穩健成長的帶動下，我們的員工人數較上一季增加。儘管進行了這些投資，我們的獲利能力仍持續改善。

Excluding share-based compensation expenses, our adjusted operating margin expanded by 1.9 percentage points year on year to a record high of 43.8%. This was primarily driven by our strong operating leverages, disciplined execution, and ongoing efforts to enhance operating efficiencies through AI applications.

剔除股份基礎給付費用後，我們的調整後營業利益率年增 1.9 個百分點，達到創新高的 43.8%。這主要受惠於我們強勁的營運槓桿、嚴謹的執行力，以及透過 AI 應用持續提升營運效率的努力。

Looking into each segment, cost of revenues increased by 2% year on year to RMB312 million this quarter. This increase was mainly due to higher server and bandwidth cost partially offset by lower app store commission fees and improved operating efficiency as we widely leverage AI in our daily operations, verification, and customer services. As a result, our gross margin went up by 1.6 percentage points year on year to 87%.

分項來看，本季營收成本年增 2% 至人民幣 3.12 億元。此增幅主要來自較高的伺服器與頻寬成本，部分被較低的應用商店抽成費用以及我們在日常營運、驗證與客戶服務中廣泛運用 AI 所帶來的營運效率提升所抵銷。因此，我們的毛利率年增 1.6 個百分點至 87%。

Sales and marketing expenses increased by 38% year on year to RMB581 million this quarter, mainly due to the marketing campaign of 2026 FIFA World Cup as well as an increase in sales employee-related expenses related to higher cash revenues.

本季銷售與行銷費用年增 38% 至人民幣 5.81 億元，主要由於 2026 年 FIFA 世界盃的行銷活動，以及因現金收入較高而增加的銷售人員相關費用。

Our R&D expenses were RMB431 million this quarter, up 3% year on year. Excluding share-based compensation expenses, our adjusted R&D expenses increased by 7% year on year to RMB361 million, mainly due to higher cloud service fees and server depreciation expenses related to AI infrastructure investment. Our G&A expenses decreased by 30% year on year to RMB219 million this quarter, mainly due to lower employee-related expenses.

本季研發費用為人民幣 4.31 億元，年增 3%。剔除股份基礎給付費用後，我們的調整後研發費用年增 7% 至人民幣 3.61 億元，主要由於與 AI 基礎設施投資相關的雲端服務費用與伺服器折舊費用增加。本季一般及行政費用年減 30% 至人民幣 2.19 億元，主要由於員工相關費用降低。

Interest and investment income reached RMB1.6 billion this quarter compared to RMB157 million for the same quarter last year. This increase was mainly driven by investment income of around RMB1.5 billion arising from the fair value changes of one of our invested companies, which went public in January 2026.

本季利息及投資收益達人民幣 16 億元，去年同期為人民幣 1.57 億元。此增幅主要來自約人民幣 15 億元的投資收益，係因我們投資的其中一家公司的公允價值變動所致；該公司於 2026 年 1 月上市。

Income tax expenses were RMB550 million this quarter, compared to RMB97 million in the same quarter last year. This increase was also mainly due to the RMB367 million tax impact from the aforementioned investment income. Withholding tax of RMB20 million, as well as RMB10 million provision, for the top-up tax under the OECD Pillar Two rules and higher income from operations. Our net income was RMB1.9 billion this quarter, up 173% year on year. Excluding share-based compensation and net gains from the aforementioned investments, our adjusted net income increased by 9% to RMB1.03 billion. Net cash provided by operating activities was RMB945 million this quarter, down 10% year on year. This decrease was mainly due to higher advertising and marketing expenditures and tax payments, as well as lower interest and investment income received partially offset by increased cash collection from customers.

本季所得稅費用為人民幣 5.5 億元，去年同期為人民幣 9,700 萬元。此增幅亦主要由於前述投資收益帶來人民幣 3.67 億元的稅務影響。另包括人民幣 2,000 萬元的預提所得稅，以及依 OECD 第二支柱規則（Pillar Two）補足稅的人民幣 1,000 萬元提列，並受營運收入較高影響。本季淨利為人民幣 19 億元，年增 173%。剔除股份基礎給付及前述投資的淨收益後，我們的調整後淨利年增 9% 至人民幣 10.3 億元。本季營業活動提供之淨現金為人民幣 9.45 億元，年減 10%。此下降主要由於廣告與行銷支出及稅款支付增加，以及收到的利息與投資收益較低；部分被向客戶收款增加所抵銷。

As of June 30, 2026, our cash position, including cash, cash equivalents, short-term time deposits and short-term investments, but excluding investments in securities, stood at RMB18.8 billion. Our strong cash position and cash generating capability enable us to sustainably deliver our shareholder return commitments.

截至 2026 年 6 月 30 日，我們的現金部位（包含現金、約當現金、短期定期存款及短期投資，但不含證券投資）為人民幣 188 億元。我們強勁的現金部位與現金創造能力，使我們能夠持續履行對股東回饋的承諾。

As Jonathan just mentioned, the Board declared an annual cash dividend of approximately USD230 million. Combined with over USD300 million in share repurchase we have completed year-to-date, which represents roughly [4.7%] (corrected by company after the call) of our total outstanding issuance, our total shareholder returns so far this year exceed USD530 million, representing over 100% shareholder return ratio compared to the adjusted net income last year. Cumulatively, we have now bought back over 10% of our total shares outstanding.

如 Jonathan 剛才提到的，董事會宣告年度現金股利約 2.3 億美元。再加上我們今年迄今已完成超過 3 億美元的庫藏股回購，約占我們已發行在外總股本的 [4.7%]（公司於電話會議後更正），我們今年迄今的股東總回饋已超過 5.3 億美元；相較於去年調整後淨利，股東回饋比率超過 100%。累計而言，我們目前已回購超過已發行在外總股數的 10%。

And now for our business outlook. For the third quarter of 2026, we expect our total revenue to be between RMB2.41 billion to RMB2.5 billion, a year-on-year increase of 11.4% to 15.6%.

接下來是我們的業務展望。對於 2026 年第三季，我們預期總營收介於人民幣 24.1 億元至 25 億元之間，年增 11.4% 至 15.6%。

That concludes our prepared remarks. Now we would like to take questions. Operator, please go ahead.

以上為我們事先準備的發言。現在我們想開始回答提問。接線員，請開始。

(Operator Instructions) Timothy Zhao, Goldman Sachs.

(接線員指示) Timothy Zhao，高盛。

My first question is regarding your AI monetization. Could you share more color on the latest progress of your AI products? For the closed-loop services that you just mentioned, could you share any color on the overall revenue scale? How do you think about the overall AI impact on the matching efficiency. And if there is any quantitative metric that you can share that will be great.

我的第一個問題是關於你們的 AI 變現。能否分享一下你們 AI 產品最新進展的更多細節？對於你們剛提到的閉環服務，能否分享其整體營收規模的大致情況？你們如何看待 AI 對撮合效率的整體影響？如果有任何可以分享的量化指標，那就更好了。

Secondly is on your Nanbeige large language model. I noticed that you recently launched Nanbeige4.2-3B model. Just wondering what is the improvement versus the last generation? And how do you compare the latest model versus the top large language model in the market and what is your different competitive position?

第二個問題是關於你們的 Nanbeige 大型語言模型。我注意到你們最近推出了 Nanbeige4.2-3B 模型。想請問相較上一代有哪些提升？以及你們如何將最新模型與市場上頂尖的大型語言模型相比較，你們差異化的競爭定位是什麼？

(interpreted) Okay, thank you for your question. About AI talent sourcing that we are currently using, there is something slightly different. It is well known that when some of our customers started to use our recommendation system, we began to know them. He used some of the Search functions to help himself.

(口譯) 好的，謝謝你的提問。關於我們目前正在使用的 AI 人才尋源，有一些略為不同之處。眾所周知，當我們的一些客戶開始使用我們的推薦系統時，我們就開始了解他們。他使用了一些搜尋功能來幫助自己。

But the Search functions, as we all know, the problem is the query is relatively short. The large language model just gave us a possibility that you can use a very long query. Also, you can use multiple rounds of conversations to make it look like long, but the system is just coming back to understanding what you really want. That's the fundamental capability the large language model has. So a very long text and multiple rounds of communication that can come out to understanding of our customer demand and better serve some clients who have high requirement, who have the requirement for high professionalism, and which traditional recommendation search model cannot serve. That just bring our search capability to the next level.

但搜尋功能，如大家所知，問題在於查詢通常相對較短。大型語言模型為我們提供了一種可能性：你可以使用非常長的查詢。同時，你也可以透過多輪對話讓它看起來像是很長的查詢，但系統其實是在回到對你真正想要什麼的理解。這就是大型語言模型所具備的根本能力。因此，透過很長的文字與多輪溝通，我們可以更理解客戶需求，並更好地服務一些要求較高、需要高度專業性的客戶，而傳統的推薦搜尋模型無法服務這些需求。這讓我們的搜尋能力提升到下一個層級。

Based on that fundamental we just discussed about, it's quite easy to understand the new value AI sourcing has brought to us. We have already served cumulatively more than 300 million users, but our monthly active users last month is just 70 million. As we further penetrate to new users, the 70 million monthly active users versus 300 million total users might turn into like 100 million versus 400 million. This creates a capability that we can expand our service to a lot of new users that's not within the monthly active users scope.

基於我們剛才討論的這個根本點，就很容易理解 AI 尋源為我們帶來的新價值。我們累計已服務超過 3 億名用戶，但上個月的月活躍用戶只有 7,000 萬。隨著我們進一步滲透新用戶，7,000 萬月活躍用戶對比 3 億總用戶，可能會變成例如 1 億對比 4 億。這創造了一種能力，使我們能將服務擴展到大量不在月活躍用戶範圍內的新用戶。

So the interesting part is customers who have the most need to contact are this kind of silent customers. This kind of customers or this kind of job seekers actually is more senior, more professional, and more likely to be liked by the headhunters. The headhunters always try to contact them. In this process, that's part of the reason we are applying this function to both our own- and third-party headhunter work stream.

所以有趣的部分在於，最需要被聯繫的客戶，其實是這類沉默客戶。這類客戶或這類求職者其實更資深、更專業，也更容易受到獵頭喜愛。獵頭總是嘗試聯繫他們。在這個過程中，這也是我們把這項功能同時應用到自有與第三方獵頭工作流的一部分原因。

This is more and more getting close to our pursuit of closed loop service. My understanding of the closed loop service is more like just one stage to another stage, the process interlocking. We are using AI to help our customers sourcing candidates. We are using AI function to help them to screen resumes. Our AI interview functions are now working on more than 10,000 interviews every day. Stage by stage, we are getting more and more close to our onboarding, and AI is just helping us to accelerating this process and achieving our goal.

這也越來越接近我們對閉環服務的追求。我對閉環服務的理解更像是從一個階段到另一個階段，流程彼此咬合。我們用 AI 幫助客戶進行候選人尋源。我們用 AI 功能幫助他們篩選履歷。我們的 AI 面試功能現在每天在超過 10,000 場面試中運作。一個階段接著一個階段，我們越來越接近入職（onboarding），而 AI 正在幫助我們加速這個流程並達成目標。

For Nanbeige4.2-3B, it is quite a coincidence that early today, a very well-known testing institution, Artificial Analysis, they combined with Liquid AI to do a joint testing on small sized model on both iPhone 17 Pro and Samsung Galaxy S26.

關於 Nanbeige4.2-3B，很巧的是今天稍早，一家非常知名的測試機構 Artificial Analysis，他們與 Liquid AI 結合，在 iPhone 17 Pro 與 Samsung Galaxy S26 上對小尺寸模型做了聯合測試。

So Nanbeige4.2-3B has achieved number one in five areas, including following the two transforming orders, scientific illusion, scientific inference, and mathematic theoretical areas. So Nanbeige has achieved number one in all those four areas.

Nanbeige4.2-3B 在五個領域拿到第一名，包括遵循兩項轉換指令、科學幻覺、科學推理，以及數學理論等領域。因此 Nanbeige 在上述四個領域都取得第一名。

Actually, our previous model 4.1-3B also achieved quite nice results. This model has been quite good in inference writing, fundamental advanced truth reasoning. On the contrary, 4.2-3B, this small-size model, is better to handle more complicated agent tools and functions, including coding intelligence entity and maybe office working intelligence entity, et cetera.

其實我們上一代模型 4.1-3B 也取得了相當不錯的結果。這個模型在推理寫作、基礎的高階真值推理方面表現很好。相較之下，4.2-3B 這個小尺寸模型，更擅長處理更複雜的代理（agent）工具與功能，包括程式碼智慧體，或是辦公工作智慧體等等。

We believe on the road to pursue AGI, there is one way: mega size models, super consuming of electronic powers, and investing a lot of money. That is a lot of big companies are doing. There is another way that maybe a smaller size model can help solving some specific problems and also create its own value. For example, the application in smartphones, on mobile vehicles, on intelligent robots, etc. In those areas, those smaller size model, we are in the leading position and have proven our value.

我們相信在追求 AGI 的道路上，有一條路是：超大尺寸模型、極度消耗電力，並投入大量資金。很多大型公司正在走這條路。另一條路則是，或許較小尺寸的模型可以幫助解決一些特定問題，同時也創造自身價值。例如在智慧型手機、移動載具、智慧機器人等應用上。在這些領域，小尺寸模型方面我們處於領先位置，也已證明了我們的價值。

That is our answers to the first two questions. Operator, let us proceed to the next one.

以上是我們對前兩個問題的回答。接線員，讓我們進入下一個問題。

Eddy Wang, Morgan Stanley.

摩根士丹利，Eddy Wang。

My first question is related to the macro impact. What is your view on the macro impact on our Company, especially for the second half of this year? As most internet company that has reported second quarter results have mentioned that the macro overhangs and the weak consumption. To what extent will BOSS Zhipin be affected under such a macro backdrop? How much of this macro-driven pressure can be offset through our operation improvement?

我的第一個問題與宏觀影響有關。您如何看待宏觀環境對我們公司的影響，特別是今年下半年？如同多數已公布第二季業績的網際網路公司所提到的，宏觀壓力仍在、消費疲弱。在這樣的宏觀背景下，BOSS 直聘會在多大程度上受到影響？這些由宏觀驅動的壓力，有多少可以透過我們的營運改善來抵消？

The second question is related to the AI development, AI service, and the product we have launched and probably will launch. Do you expect they will have different cost structure, and will this affect our overall margin? In addition, do we have plan to materially ramp up the CapEx as we have seen with some of the other internet companies?

第二個問題與 AI 發展、AI 服務，以及我們已推出且可能將推出的產品有關。您是否預期它們會有不同的成本結構，並影響我們整體利潤率？另外，是否有計畫像其他一些網際網路公司那樣，實質性地提高資本開支（CapEx）？

(interpreted) Thank you for your question. Regarding the macro situation, I actually respect your professional observation, and I will not talk too much about it. I have been starting our business for more than 12 years and we have experienced a lot, whether you have experience, or not experience, we have all gone through that. We have always maintained to be a very stable and maybe trustworthy business, and we will continue to maintain this very stable operation.

（口譯）謝謝你的提問。關於宏觀情況，我其實尊重你專業的觀察，我就不多談了。我創業做這個業務已經超過 12 年，我們經歷過很多，不管你是否經歷過，我們都走過來了。我們一直保持自己是一個非常穩健、也許值得信賴的業務，我們也會繼續維持這種非常穩健的營運。

We have a big opportunity here with our potential market size. We have served over 300 million customers and more than 22 million enterprises. It is well known that the average life cycle of enterprise, according to the People's Bank of China's, is less than three years. Within all those 22 million companies we have served, a lot of them are not active anymore. They have turned into new elements and beginning new companies.

我們在這裡有一個很大的機會，來自於潛在的市場規模。我們已服務超過 3 億名客戶，以及超過 2,200 萬家企業。眾所周知，根據中國人民銀行的資料，企業的平均生命週期不到三年。在我們服務過的這 2,200 萬家公司中，很多已經不再活躍。它們轉化成新的元素，開始成立新的公司。

For ease of observation, even we consider those 40 million enterprises as a fixed situation, we have more than double of market to grow. On top of that, a lot of new companies are emerging every year. Our actual market size is even bigger.

為了便於觀察，即使我們把 4,000 萬家企業視為一個固定情況，我們仍有超過一倍的市場可以成長。此外，每年都有大量新公司出現。我們實際的市場規模更大。

The second opportunity is in the paying ratio. Our actual number of annual served enterprises is more than 10 million, and over 50% of them are using our service for free. From that perspective, this is our second driver or second growth opportunities.

第二個機會在於付費率。我們實際每年服務的企業數量超過 1,000 萬家，其中超過 50% 使用的是免費服務。從這個角度看，這就是我們的第二個驅動因素或第二個成長機會。

I will use one first-tier city as an example. We just reported that we intend to increase monetization for certain first-tier cities. In this particular city, including both paid and free service, the average cost per mutual match our customers can get for this city, for example, is like X RMB. If we turn those free customers into our lowest level of paying customers, then those costs will grow by at least 15%.

我以一個一線城市為例。我們剛剛提到，我們打算在部分一線城市提高變現。在這個特定城市中，包含付費與免費服務在內，客戶在該城市每獲得一次雙向匹配（mutual match）的平均成本，例如大約是 X 元人民幣。如果我們把這些免費客戶轉化為最低等級的付費客戶，那麼這些成本至少會增加 15%。

Please rest assured, both our investors, clients or public, that actually this is a very minimal changes. I just explained that for a lot of our customers, the average cost to achieve a mutual matching is only the price of one mineral water at 7-Eleven. Either one bottle of mineral water or 1.15 bottle of mineral water is a very minimal cost to every enterprises.

請投資人、客戶與公眾放心，這其實是非常小的變化。我剛才解釋過，對很多客戶而言，達成一次雙向匹配的平均成本，只相當於在 7-Eleven 買一瓶礦泉水的價格。不論是一瓶礦泉水，或 1.15 瓶礦泉水，對每一家企業來說都是非常小的成本。

Eddy, just as we go through all these years, I am confident we are not only surviving, we should and we will be better and better.

Eddy，就像我們這些年一路走來，我有信心我們不僅能生存下去，而且應該、也一定會越來越好。

About the second question, thank you Eddy for asking me that. Actually, all those large companies who have invest a lot of CapEx for like arm race or things like that, I think they have their ambitions, they have their beliefs, but most importantly, they have that financial capabilities.

關於第二個問題，謝謝 Eddy 問到這點。其實那些投入大量資本開支（CapEx）去做所謂軍備競賽之類的大公司，我認為他們有他們的雄心、有他們的信念，但最重要的是，他們具備那樣的財務能力。

For a company like us, we chose the path of following the taillight strategy, and we prioritize AI applications in the smaller size companies. That is our approach, our strategy to facing this AI disruption or AI impact, what we can do.

對於像我們這樣的公司，我們選擇走「跟隨尾燈」策略的路徑，並優先在規模較小的公司中推動 AI 應用。這就是我們的做法、我們面對這次 AI 顛覆或 AI 影響的策略，以及我們能做的事情。

Thank you for your trust, but I believe our investment in the AI will not impact our overall cost structure and impact our operation capability and financial margins. We will maintain current level of investment. As you know, we have good profitability, so we will maintain around like 20% to 25% of R&D expenses, and we will spend incremental money on the AI and to give more support. But I will not sacrifice our safety on our cash flow. I won't do that. Just don't worry.

感謝您的信任，但我相信我們在 AI 上的投資不會影響我們整體的成本結構，也不會影響我們的營運能力與財務利潤率。我們將維持目前的投資水準。如您所知，我們的獲利能力良好，因此我們會將研發費用維持在約 20% 到 25%，並在 AI 上投入增量資金以提供更多支援。但我不會犧牲我們現金流的安全性。我不會那樣做。請不要擔心。

That's our answer to those two questions. Thank you.

以上是我們對這兩個問題的回答。謝謝。

Wei Xiong, UBS.

UBS 的 Wei Xiong。

First, it is encouraging to see our margins have been maintaining at a very healthy level. Could we quantify the benefits from AI in our internal use to drive better efficiency and lower costs? How much room of further improvement do we see? Also, after the investment in FIFA World Cup, how should we think about the investment plans, the expenses, and the margin trends in the second half?

首先，很令人鼓舞的是看到我們的利潤率一直維持在非常健康的水準。我們是否可以量化 AI 在內部使用上帶來的效益，以提升效率並降低成本？我們還看到多大的進一步改善空間？另外，在投資 FIFA 世界盃之後，我們應該如何看待下半年的投資計畫、費用以及利潤率走勢？

Second, could we please get an update on your overseas business, including OfferToday? How should we think about if there is any plan to expand into other markets?

第二，能否請您更新一下海外業務的情況，包括 OfferToday？我們應該如何看待是否有計畫擴展到其他市場？

I will take the first question on margin. So actually we have been leveraging AI in all aspects of our daily operations, including security, notifications, sales and marketing, and operating employees and everywhere. But to quantify it, it may be easier in the cost lines. Since 2023, we have been witnessing that alongside with our user growth, our overall headcount of operating employees maintain stable. As a result, the employee-related cost as a percentage of revenue continue to go down and help to contribute around 2 percentage points of our gross margin. You can see our gross margin now stayed at a very healthy high 80s level. We believe we at least will maintain this very high gross margin level.

我先回答第一個關於利潤率的問題。其實我們已在日常營運的各個面向運用 AI，包括安全、通知、銷售與行銷、員工營運管理等各方面，幾乎無所不在。但若要量化，從成本項目來看可能更容易。自 2023 年以來，我們觀察到在用戶成長的同時，營運員工的整體人數維持穩定。因此，員工相關成本占營收的比例持續下降，並對我們的毛利率貢獻約 2 個百分點。您可以看到我們目前的毛利率維持在非常健康的 80% 高段。我們相信至少會維持這個非常高的毛利率水準。

For our outlook for the second half, yes, we have sponsored the FIFA World Cup, but the cost will be evenly distributed or recognized within second quarter and third quarter. Apart from that, we will maintain our current investment level of the cloud service rental cost for our AI model training. We are expecting maybe in the third quarter, the margin level should be similar to second quarter. And for the full year as we expected at the beginning of this year, our overall adjusted operating margin can still slightly increase.

至於我們對下半年的展望，是的，我們贊助了 FIFA 世界盃，但相關成本會在第二季與第三季之間平均分攤或認列。除此之外，我們將維持目前用於 AI 模型訓練的雲端服務租賃成本之投資水準。我們預期第三季的利潤率水準應與第二季相近。而就全年而言，如同我們在今年年初的預期，我們整體調整後營業利潤率仍可小幅提升。

(interpreted) Thank you for your concern about OfferToday. Our current goal for OfferToday is in maybe five years from today, it can bring the Company with USD100 million to USD150 million of revenue, that is around about the market size of Hong Kong. We call it maybe a middle dish, not too fast, but not too slow.

（口譯）感謝您對 OfferToday 的關注。我們目前對 OfferToday 的目標是，大約從今天起五年內，它能為公司帶來 1 億至 1.5 億美元的營收，這大致相當於香港的市場規模。我們稱之為「中等菜」，不會太快，但也不會太慢。

The lessons we learned from OfferToday is that it took around two, three years for our new business like OfferToday to enter into a market, and next additional five years to grow to achieve USD100 million to USD150 million of revenue. We consider this kind of place or this kind of city worth investing. Of course, those cities are in Asia and Europe. Of course, we need to avoid those high geopolitical risk areas. To sum up, two to three years of adaption and mature, five years of development, there are still a lot of cities of this size and worth investing.

我們從 OfferToday 學到的經驗是，像 OfferToday 這樣的新業務進入一個市場大約需要兩到三年，接著再額外五年成長，才能達到 1 億至 1.5 億美元的營收。我們認為這類地方或這類城市值得投資。當然，這些城市主要在亞洲與歐洲。當然，我們需要避開地緣政治風險較高的地區。總結而言，兩到三年的適應與成熟、五年的發展，仍有許多這種規模且值得投資的城市。

Also we have some markets we call it slow dish, maybe take a longer term around 10 to 15 years, which can also achieve a revenue like USD100 million to USD150 million. The profile of this kind of market is maybe generally younger in the average age of the citizens and is a developing country, but it is developing quite orderly. Its total population is around slightly less than 100 million. Some place like Vietnam, Argentina or Brazil. In 10 to 15 years, we are hoping this kind of city can accept the new models like we have created and bring about nice profit or revenue to the Company by them. This topic is about OfferToday and what lessons it can give us to developing our overseas business.

另外，我們也有一些市場稱為「慢菜」，可能需要更長的期間，大約 10 到 15 年，但同樣也能達到約 1 億至 1.5 億美元的營收。這類市場的特徵通常是國民平均年齡較年輕，且屬於開發中國家，但發展相當有秩序。其總人口略低於 1 億。例如越南、阿根廷或巴西等地。在 10 到 15 年內，我們希望這類城市能接受我們所創造的新模式，並為公司帶來可觀的利潤或營收。以上是關於 OfferToday，以及它能為我們發展海外業務帶來哪些啟示。

That is our answers to all the questions today. Thank you.

以上就是我們今天對所有問題的回答。謝謝。

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.

謝謝。由於時間限制，今天的問答環節到此結束。接下來我會把會議交回給 Laura，請她做任何補充或結語。

Thank you once again for joining us today. If you have any further questions, please contact our IR team directly. Thank you.

再次感謝各位今天參與。如有任何進一步問題，請直接聯繫我們的投資人關係（IR）團隊。謝謝。

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

感謝您參與今天的會議。本次活動到此結束。您現在可以中斷連線。

Portions of this transcript that are marked (interpreted) were spoken by an interpreter present on the live call. The interpreter was provided by the company sponsoring this event

本逐字稿中標示為（口譯）的部分，是由現場電話會議中的口譯員所轉述。該口譯員由贊助本活動的公司提供。