--- title: "BZ (Trans): Full-Year Profit View Nudged Up; AI Spend Kept Disciplined." type: "Topics" locale: "en" url: "https://longbridge.com/en/dolphin/post/43592210.md" description: "Kanzhun 2Q26 earnings call transcript.Shareholder returns YTD topped $530 mn, already exceeding 100% of last year's adj. net income. However, Q3 revenue guidance is RMB 24.1–25 bn. That implies +11.4% to +15.6% YoY, with no acceleration. The company also made clear it will not pursue large AI capex. R&D expense ratio will remain at 20%–25%." datetime: "2026-08-25T16:05:07.000Z" locales: - [en](https://longbridge.com/en/dolphin/post/43592210.md) - [zh-CN](https://longbridge.com/zh-CN/dolphin/post/43592210.md) - [zh-HK](https://longbridge.com/zh-HK/dolphin/post/43592210.md) author: "[Dolphin Research](https://longbridge.com/en/dolphin.md)" generator: "portal-rs" --- # BZ (Trans): Full-Year Profit View Nudged Up; AI Spend Kept Disciplined. **Trans compiled by Dolphin Research: Kanzhun FY26 Q2 earnings call notes:** **I. Key takeaways** 1) Shareholder returns: The BOD declared an annual cash dividend of approx. $230 mn; buybacks completed YTD exceed $300 mn, equal to ~4.6% of total shares.Total shareholder returns YTD are over $530 mn, more than 100% of last year's Adj. net profit, delivering on prior commitments.Cumulative buybacks now exceed 10% of shares outstanding. 2) Guidance: Q3 revenue is guided at RMB 2.41–2.50 bn, +11.4% to +15.6% YoY.For the full year, Adj. OPM is expected to be slightly above the initial outlook.SBC as % of revenue will remain at a high single-digit level for the year. 3) One-off investment gains and taxes: Interest and investment income was RMB 1.6 bn vs. RMB 157 mn a year ago.The increase mainly stems from approx. RMB 1.5 bn investment gains from the FV change of an investee that will list in Jan 2026.Income tax expense was RMB 515 mn vs. RMB 97 mn; the increase was driven by RMB 367 mn taxes tied to that gain, plus RMB 20 mn withholding tax, RMB 10 mn OECD global minimum tax top-up, and higher OP. 4) Q2 key financials a) Topline: Revenue of RMB 2.4 bn, +14% YoY, with accelerating growth.Adj. OP (ex SBC) of RMB 1.05 bn, +19% YoY. b) Margin: Adj. OPM of 43.8%, up 190 bps YoY, a record high.Drivers include operating leverage, cost discipline, and AI-enabled efficiency gains. c) Net income: GAAP net profit of RMB 1.9 bn, +173% YoY.Adj. net profit of RMB 1.03 bn, +9% YoY, excluding SBC and investment-related net gains. d) Costs and expenses: Total costs and expenses of RMB 1.5 bn, +6% YoY.SBC fell 19% YoY to RMB 196 mn, equal to 7.8% of revenue, down 310 bps YoY. e) Cash flow: Operating cash flow net of RMB 945 mn, down 10% YoY.Drivers were higher ad & marketing and taxes, lower cash collection from investment income, partly offset by stronger customer receipts. f) Cash-like assets at quarter-end were RMB 18.8 bn, including cash & equivalents, short-term time deposits and short-term investments.Securities investments are excluded. **II. Detail from the call** **2.1 Management commentary** 1) Users and monetization a) Q2 user scale hit record highs on both sides, with Avg. MAUs above 20 mn.Over the 12 months to Jun 30, 2026, paid enterprise customers rose 11% YoY to 7.2 mn.The pay rate among active enterprise users improved for the fourth straight quarter. b) ARPPU grew 7% YoY, driven by more efficient, higher-value services.Expanding AI features prompted higher client spend.Hiring demand was broadly stable, and revenue growth was balanced across large accounts and SMBs. 2) Growth strategy: 5-year market view a) Q2 marked the fifth anniversary of the IPO.Management reiterated that China has ~500 mn workers and over 40 mn active enterprises, leaving ample room for online recruiting penetration.Mobile recommendations plus direct chat materially lower communication costs. b) This low-cost model has enabled tens of millions of enterprises to move recruiting online at scale.The core logic is network effects: as the two-sided user base expands and diversifies, with more frequent interactions, platform service capability strengthens.Future 5-year growth will come from lower-tier cities as well as Tier-1/2. c) In lower-tier cities, the key driver remains penetration.Commercialization metrics in Tier-1 cities were adjusted starting H2. 3) AI applications and investment a) Large-scale AI deployment has improved platform efficiency.AI interview features in white-collar and blue-collar (factory) scenarios are priority investments, with related revenue growing rapidly QoQ.Internal AI covers safety, moderation, sales & marketing, ops, and R&D. b) In Q2 the company sponsored the FIFA World Cup and ramped AI cloud services.Headcount increased QoQ, yet profitability continued to improve. 4) Cost breakdown and drivers a) COGS rose 2% YoY to RMB 312 mn, mainly on higher research and analyst costs.Lower app store commissions and scaled AI in ops/moderation/customer service partly offset the increase.GPM expanded 160 bps YoY to 87%. b) S&M expenses rose 38% YoY to RMB 581 mn, driven by the World Cup campaign and higher sales-related costs alongside cash revenue growth.R&D was RMB 431 mn, +3% YoY.Adj. R&D grew 7% YoY to RMB 361 mn, reflecting AI infra spend on cloud services and server depreciation. c) G&A fell 30% YoY to RMB 219 mn, mainly on lower employee-related costs. **2.2 Q&A** **Q: What is the latest on AI products? Any quantifiable metrics for closed-loop service revenue and matching efficiency?** A: No quant provided for closed-loop revenue or efficiency; only that AI interviews average over 10,000 per day.AI talent sourcing differs from legacy search: traditional queries are short, while LLMs enable long queries and multi-turn dialogue to truly capture client intent.This serves higher-expertise clients better and elevates search capabilities. The real value is opening the stock pool beyond MAUs.Cumulative served users exceed 300 mn, but last month MAUs were only 17 mn; with further penetration, that could become 100 mn MAUs vs. 400 mn cumulative.Headhunters most need to reach these 'silent users' who are more senior and professional, and the feature is integrated into both in-house and third-party headhunter workflows. Management views closed-loop services as interactions from one step to the next.AI interviews now run over 10,000 sessions per day and are progressing toward the onboarding stage. **Q: How does Nanbeige 4.2-3B improve vs. the prior gen, and how does it compare to top models?** A: In third-party on-device small-model benchmarks, it ranks first across multiple sub-metrics, clearly following a small-model rather than ultra-large model path.Benchmarking by Artificial Analysis and Liquid AI was conducted on iPhone 17 Pro and Samsung Galaxy S26 for small-size models.Nanbeige 4.2-3B ranks first in instruction following, scientific reasoning, and math. The prior gen 4.1-3B already performed well in reasoning and writing.4.2-3B is better at complex agent tasks, including code intelligence and office scenarios.Strategically, there are two paths to AGI: very large models that are power- and capex-heavy, pursued by big firms, and small models solving concrete problems and creating value on phones, IVI, and robots. The company is in a leading position on the latter path. **Q: How much will macro pressure affect H2, and how much can ops improvements offset?** A: No macro impact quantification; instead, management emphasized two stock pools: enterprise client base and pay rate.They respect sell-side macro calls and will continue stable operations, having weathered multiple cycles over 12+ years.The first space is the client base: cumulative users >300 mn and enterprises >22 mn, but per PBoC data the avg. life of Chinese enterprises is under 3 years, so many of the 22 mn are no longer active. Even assuming 40 mn active enterprises as a static market, addressable room is more than double current customers, with new firms emerging each year.The second space is pay rate: annual active enterprises exceed 10 mn, over 50% still use free tiers.In a Tier-1 city example, converting free users to the lowest paid tier raises cost per successful match by at least 15%, but in absolute terms it is tiny—from roughly 1 bottle of mineral water to 1.15 bottles. **Q: Do AI services/products have different cost structures that could affect margins? Any plan to materially lift capex?** A: No material capex ramp; the strategy is to 'follow the taillights' and prioritize AI applications.R&D as % of revenue will stay around 20%–25%.Those investing heavily in capex have their beliefs and financial capacity; the company will maintain current investment levels. AI spend will not alter the overall cost structure, operating capability, or financial performance.Incremental R&D will go to AI while preserving cash flow safety. **Q: Can you quantify internal AI-driven efficiency and cost benefits, and how much runway remains? Post-World Cup, what are H2 expense and margin trends?** A: AI has cumulatively contributed about 2 ppt to GPM.Q3 margins should be similar to Q2, and full-year Adj. OPM can still tick up.AI is applied across safety, moderation, sales & marketing, ops, and R&D, with more visible impact on costs. Since 2023, as users kept growing, total ops headcount stayed flat and employee-related costs as % of revenue kept falling, adding ~200 bps to GPM.Management believes the current high GPM can at least be maintained.World Cup sponsorship costs are recognized evenly across Q2–Q3, and cloud rental for model training will remain at current levels. Therefore, Q3 margins should be close to Q2, and full-year Adj. OPM is expected to show a slight improvement vs. the initial plan. **Q: What is the latest on overseas biz OfferToday? Any plans to expand to other markets?** A: The target is to contribute $100–115 mn in revenue about 5 years from now, which is roughly the market capacity, with a 'moderate' pace.From OfferToday’s playbook: entering a market takes ~2–3 years, followed by 5 years to reach $100–115 mn in revenue.Asia and Europe have multiple markets that fit this size, but the company will avoid regions with high geopolitical risk. Another 'slow lane' cohort needs 10–15 years to reach $100 mn in revenue.These are typically developing countries with younger populations and total populations just under 100 mn, such as Vietnam, Argentina, and Brazil. \ **Risk disclosure and statements:**[**Dolphin Research disclaimer and general disclosures**](https://support.longbridge.global/topics/misc/dolphin-disclaimer) ### Related Stocks - [02076.HK](https://longbridge.com/en/quote/02076.HK.md) - [BZ.US](https://longbridge.com/en/quote/BZ.US.md) --- > **Disclaimer: This article is for reference only and does not constitute any investment advice.**