--- title: "Tencent: No Longer Resting on Laurels; AI Is the Way Forward" type: "Topics" locale: "en" url: "https://longbridge.com/en/topics/40736371.md" description: "$TENCENT(00700.HK) Q1 results are out, and the read is mixed with plenty to unpack. There are several points that warrant a closer look.On the market's key question of AI spend vs. returns, Dolphin Research offers a quick take.1) AI spend is clearly rising. Q1 capex recognized was RMB 31.9bn (+16% YoY), which looks like a modest growth rate.But cash outlay reached RMB 37.0bn, indicating another round of prepayments for capacity yet to be delivered. Compute remains scarce, especially at the high end.Since Tencent began iterating its foundation model in earnest, securing high-end compute has required heavy upfront orders. Looking back over the past year..." datetime: "2026-05-13T14:50:30.000Z" locales: - [en](https://longbridge.com/en/topics/40736371.md) - [zh-CN](https://longbridge.com/zh-CN/topics/40736371.md) - [zh-HK](https://longbridge.com/zh-HK/topics/40736371.md) author: "[Dolphin Research](https://longbridge.com/en/news/dolphin.md)" --- # Tencent: No Longer Resting on Laurels; AI Is the Way Forward $TENCENT(00700.HK) released Q1 results, a mixed bag overall with several points worth a close look. Focusing on what the market cares most about — the AI investment-to-return — here are Dolphin Research’s quick takes: **1) AI spend is set to rise:** Q1 capex booked was RMB 31.9bn (+16% YoY), which does not look fast on an accounting basis. But cash capex paid reached RMB 37.0bn, indicating another round of prepayments for equipment still in the pipeline. Capacity is tight, especially at the high end, and Tencent’s push to iterate foundational models requires aggressive procurement. Over the past year, capex cash payments have exceeded accounting recognition every quarter. **2) Margin drag is showing, with some offsets:** AI spend started to weigh on profits in Q1. For example, ads — benefiting from AIM+ and stronger marketing tools — booked part of AI depreciation, taking the segment GPM down by 50bps YoY. That said, Tencent’s diversified portfolio — particularly virtual content with paid models — creates room to flex profits and cushion part of the AI drag. This has been our consistent view. Q1 VAS GPM rose further to 63%, helped by a higher mix of self-developed games (especially evergreen titles with low marginal R&D) and lower iOS revenue share. Other efficiency gains also helped offset near-term AI spend; G&A declined both YoY and QoQ, with total headcount down by 1,000 QoQ. Core operating profit for main businesses was RMB 66.1bn (GP minus opex, excluding investment gains), up 12% YoY. While margin improved YoY, profit growth is likely to come under pressure this year, and the market now expects only single-digit growth (implying a slight margin dip). On the commonly watched metric, Q1 non-IFRS net profit was RMB 67.9bn (+11% YoY), with margin at 34.6% (+50bps YoY). **3) Buybacks no longer a strategic priority, but cash can still support during pressure points:** Since the South African major shareholder started a long-term selldown in 2022, Tencent’s buybacks drew attention, especially from conservative insurers. With Tencent past its growth bottleneck and macro conditions evolving, plus the major shareholder slowing sales (avg. 1.4mn shares per month in Q1 vs. 15mn a year ago), market cap is increasingly driven by growth expectations. In the broader AI cycle, buybacks are taking a back seat. This mirrors global mega-cap tech, as Google and Meta paused buybacks. From management’s perspective, allocating to AI vs. buybacks comes down to ROI. Given AI’s disruptive potential and risk, investment clearly ranks higher now. Q1 buybacks were HKD 7.6bn; despite CNY holidays and the results blackout, this was a sharp YoY reduction. Even so, Tencent’s robust cash flow suggests it still has room to deploy capital during periods of market stress. **4) Legacy engines keep minting cash, but more AI-led monetization is needed:** On revenue, Q1 grew 9%, with ads and games still carrying the load. As the base gets bigger, sustaining growth gets harder. (1) Ads grew 20%, beating expectations despite a weak macro, driven by Video Accounts and search performance. AI-related sectors, games and e-commerce led the way, supported by an upbeat AI cycle, game seasonality in Q1 and incremental demand from WeChat Shops. (2) Games rose 8%, with China up 6% and overseas up 13% (14% ex-FX). On paper this was below expectations, but evergreen titles Honor of Kings and Peacekeeper Elite held up well, with contributions from Delta Force events and Wuthering Waves (Kuro to be consolidated by end-2024). The miss vs. expectations largely reflects a late CNY and the late-quarter launch of the high-profile mobile title Rock Kingdom, pushing some gross billings into next quarter. Deferred revenue at Q1-end reached RMB 141.3bn (+15% YoY), accelerating vs. Q4, implying underlying billings growth north of 10%. However, the weak launch of Honor of Kings: World in Apr will weigh on near-term growth expectations. There is still room to improve, supported by the Kings IP; the key is how fast internal adjustments can be executed. (3) FinTech & Biz Services were roughly in line, with FinTech growing low single digits and Biz Services up 20%, driven by AI cloud demand and higher tech commissions from WeChat Shops. Price hikes likely had limited in-quarter recognition; the earnings call should give more color on cloud strategy. **7\. Full financials at a glance** **Dolphin Research View** **Tencent’s capex will clearly step up this year, but** the stability and diversity of its core franchises should allow a more balanced approach — not exhausting cash flows or fully sacrificing profits. Annualizing Q1 suggests capex could run above RMB 120bn; on rough math, cash flows still offer cushion. Net cash was RMB 146.8bn at Q1-end. We roughly estimate FY operating cash flow at RMB 330bn; after RMB 35bn media content and lease liabilities, RMB 45bn dividends, RMB 15bn interest, and RMB 50bn net external investments (higher than prior years, reflecting AI bets such as DeepSeek), there would be about RMB 185bn left for capex payments and buybacks in-year, before considering on-book cash balances. Notably, management’s AI investment is not just capex. Given chip constraints and ROI, Tencent will also rent external compute, which runs through opex rather than depreciation. Excluding personnel costs from R&D, underlying infrastructure tech spend rose 61% YoY in Q1, accelerating to 28% of total R&D (vs. 20% a year ago). This outpaced total depreciation growth, which was just above 20% YoY. **Pre-results chatter around a 'RMB 1tn over three years' AI plan spooked investors. Our read from industry checks is that this refers to a combined capex + opex value framework:** for 2026, assume RMB 120bn capex plus RMB 36bn AI opex in Q4 run-rate terms annualized over a five-year typical depreciation horizon, which would be equivalent to peers spending roughly RMB 300bn in capex. **Under this accounting for investment, the actual cash flow pressure is not as severe.** Still, Tencent needs belt-tightening elsewhere to keep profit and cash flow in balance, such as tighter basic opex control and a reduced buyback run-rate. This is the trade-off. (1) Offsetting profit erosion will rely on natural GPM tailwinds in existing businesses and proactive opex controls. But with heavy capex rolling into depreciation each year, the required offsets will accumulate. Legacy businesses alone are unlikely to be enough. Direct AI monetization may not add much near-term profit, but it expands the long-term growth runway and can bolster market confidence now, particularly among conservative investors assessing Tencent’s willingness to accept near-term profit dilution for AI. While cloud may not be the strategic core today, industry momentum argues for more monetization. If capex guidance moves meaningfully higher, Tencent Cloud’s role in the AI strategy could be upgraded. In Mar–Apr, Tencent Cloud raised prices multiple times and pushed WorkBuddy, CodeBuddy and free trials for the HY3.0 model, suggesting a potential shift in posture. With deep partnerships and stakes in leading independent model developers (DeepSeek collaboration and fundraising, stakes in Minimax and Zhipu), plus Tencent’s own HY3.0 with clear price-performance advantages, Tencent Cloud has ample growth potential. (2) On buybacks, using the RMB 185bn combined budget for capex and buybacks, if capex exceeds RMB 120bn, then buybacks likely drop to RMB 50–60bn without tapping on-book cash and short-term investments. This implies 2026 will not see last year’s steady HKD 1bn per day; repurchases may be more opportunistic, stepping up when the stock comes under acute pressure. Details should be clarified on the call. Near-term results will visibly look 'worse' and are not comparable with the last two years. But at sub-13x P/E, the market seems to embed punitive assumptions that it is too early to expect AI payoff. These include fear from the 'RMB 1tn over three years' narrative, disappointment around Honor of Kings: World dampening 2026 game growth confidence, and a bearish chain of logic that Tencent could structurally lag in AI, weakening the WeChat entry point, with AI reshaping game development and eroding large studios’ historical R&D advantages. At the same time, the market may be overlooking two points: **(1) AI’s primary disruption today is on the productivity side,** driven by urgent user needs and willingness to pay, while C-end time-share penetration remains early. Tencent’s core is more C-end with a stable competitive landscape, buying it time and funding to adjust strategy. **(2) Tencent itself is changing.** In the prior two years, Tencent arguably underweighted AI and relied on bottom-up, horse-race style innovation. This AI wave requires concentrated resources — especially data and infrastructure — and a willingness to relax near-term ROI thresholds to solve cross-department coordination and resource allocation, which is why smaller players or independent labs sometimes move faster. Last year’s org changes signal Tencent’s intent to adapt, but more is needed. Under resource constraints, it must break group-level organizational bottlenecks (e.g., Yuanbao’s role in the WeChat ecosystem and its relationship with future in-app agents). We look for more focus and org moves this year. **Detailed analysis follows** **I. WeChat ecosystem steady; QQ continues its structural decline** WeChat MAUs reached 1.432bn in Q1, with a net add of 14mn QoQ, a normal seasonal uptick. QuestMobile data show WeChat time spent still rising, with stable share in the industry. QQ MAUs were 516mn in Q1, up 8mn QoQ thanks to CNY, but user attrition persists YoY. The trend remains down. ![图表, 折线图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/bdFCYKURbZEDiwdYVeVYn9B9kyXvPmPi.jpg?x-oss-process=style/lg) ![图片包含 图表描述已自动生成](https://pub.pbkrs.com/cms/2026/0/SakqyiHP68AmTzgQDKCsNR7KnfgziDaC.jpg?x-oss-process=style/lg) In AI chatbots, despite heavy promotion early in the year, the only one with meaningful retention has been Doubao. Qianwen and Yuanbao both faded after paid traffic ended, though Qianwen held up better than Yuanbao. Post HY3.0, Yuanbao’s reputation improved on some reasoning tasks, appearing ‘smarter’ than Doubao and clearly better than 2.0. But organic adoption is still slow as most users stick with Doubao via direct traffic from Douyin, albeit with low daily time spent. This indirectly suggests AI’s disruption is currently more on the productivity side, where users have stronger willingness to pay and seek higher intelligence and value. VAS paid users fell by 1mn QoQ to 266mn in Q1. Adds came mainly from Tencent Music (we estimate +1mn), while Tencent Video likely lost users. Though official paid numbers were not disclosed again, QM data show long-form video time under pressure. **II. Games: underlying performance fine, with some billings deferred to next quarter** Online games revenue was RMB 64.2bn in Q1 (+7.9% YoY), with China +5.8% and overseas +13% (+14% ex-FX), overall below expectations. Under the hood, evergreen titles did well, with incremental contributions from Delta Force and Wuthering Waves; the main issue was a late CNY and the late-quarter launch of Rock Kingdom, which deferred billings into Q2. Deferred revenue reached RMB 141.3bn at Q1-end (+15% YoY), a clear acceleration vs. Q4. However, the weak start for Honor of Kings: World in Apr dampens growth expectations. There is still room to improve given the Kings IP, but across the pipeline there are fewer true mass hits, so sustaining growth will likely rely on operating newer titles such as Rock Kingdom and refreshing evergreen franchises rather than producing the kind of surprise seen in the last two years. VAS billings rose 11% YoY in Q1. Excluding music at 6% and likely negative growth in video subscriptions, domestic game billings should have grown in the teens as management suggested, still ahead of the industry. ![图表, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/kYLxHP2UBPZV85JLzqXmW1mDMVT9DSsL.jpg?x-oss-process=style/lg) ![图表, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/1k6SomTiEzxHQBxTygHHKF5YaN8kr1HC.jpg?x-oss-process=style/lg) **III. Ads: resilient and strong despite macro headwinds** Despite a weak consumption backdrop, ad revenue kept growing near 20%, beating expectations. Management cited AIM+ tools (now a 30% penetration of ad budgets), Video Accounts (time spent +30%) and stronger WeChat search. **IV. FinTech & Biz Services: stable low growth, looking for Tencent Cloud’s story** FinTech & Biz Services grew nearly 9% YoY in Q1, with a slight QoQ acceleration. Biz Services rose 20% (including WeChat Shops commissions), a touch slower than 22% in Q4, implying FinTech likely grew low to mid-single digits (+~5% YoY by our estimate). Given the macro, a visible recovery is still hard to see. ![图表, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/WxKdWU8p9TkEnxwhw5L6Qryv8jr3Q46D.jpg?x-oss-process=style/lg) ![图表, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/gTGkcXUdGoXFt3kvWPxah78YnFRSXPXf.jpg?x-oss-process=style/lg) That said, frequent Tencent Cloud price hikes in recent months, rapid user traction for WorkBuddy, and the launch of value-for-money HY3.0 increase our expectation that management will say more about cloud’s role, helping break the near-term valuation overhang. We will watch this closely. **Because HY3.0 had a prior free trial (about two weeks), it quickly climbed to No.1 in tokens consumed by individual users upon launch.** Even after paywalls returned, HY3.0 still holds a value edge and remains No.1 in daily tokens on OpenRouter. This speaks to both performance and price-performance. ![图表, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/kfM3m4cW1CP3bT5muYgJuRN2UnLXSbU2.jpg?x-oss-process=style/lg) **V. Investment gains: asset disposals and fair value uplift** We track investment gains using the prior definition (pre-2025): other gains, net (including investment income) and share of profit of associates/JVs. Combined investment gains rose QoQ in Q1, mainly from asset disposals and fair value movements in financial assets, offsetting intangible impairments at investees. Share of profit recognized was nearly RMB 3.6bn, down 21% YoY. Given volatility, we focus on core operations and de-emphasize investment swings in our assessment. ![图表, 瀑布图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/SfUNKtzUDMfiTetGAPoePRk4wbJnkcL5.jpg?x-oss-process=style/lg) ![图表, 条形图, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/g71gU878JfqHYpnzNhfJ9kbrFMbUWehi.jpg?x-oss-process=style/lg) **VI. Profit drag from AI is being partly offset by business mix shifts** Adj. net profit was RMB 67.9bn in Q1 (+11% YoY), in line with expectations. Core operating profit (GP minus opex, stripping out share of associates and sundry items) grew 12% YoY, slightly ahead of expectations. Despite heavy Q1 marketing for Yuanbao and pre-launch user acquisition for new games, higher game GPM and tighter G&A (headcount -1,000 QoQ) lifted margin by nearly 100bps YoY. But this trend is unlikely to persist as higher depreciation from elevated capex rolls in. ![图表, 条形图, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/coMqWzS2SUa1TQyvGywwbDE7FcXCsqHH.jpg?x-oss-process=style/lg) ![图表, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/sx3B38CGVPqiofG9kdryKFvQK5fMtWMs.jpg?x-oss-process=style/lg) ![图表低可信度描述已自动生成](https://pub.pbkrs.com/cms/2026/0/vTTKMuUqcCFncr3xzNvxN1UbVodYYPBG.jpg?x-oss-process=style/lg) ![图表中度可信度描述已自动生成](https://pub.pbkrs.com/cms/2026/0/SM8SYMcZuKsMd7DqV9UQJquMTtvcrEbk.jpg?x-oss-process=style/lg) ![图表描述已自动生成](https://pub.pbkrs.com/cms/2026/0/YH3ELpVRWyjacanVhiXUuBXcQ32Cw9TS.jpg?x-oss-process=style/lg) Capex booked was RMB 31.9bn in Q1, up sharply QoQ, while cash capex outflow was RMB 37.0bn, again showing prepayments. While Tencent’s conservative style prioritizes ROI, three straight quarters of prepayments suggest the bottleneck lies on the compute supply side. ![图表, 瀑布图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/HtEd8LfK89USGHTEcJNWRAQETwmaSvKX.jpg?x-oss-process=style/lg) **VII. Major shareholder selling slows further; buybacks shrink** A quick look at buybacks and selldowns. In the two months since the last results, major shareholder selling continued to slow, with monthly sales down to 1.4mn shares. As of yesterday, Prosus held 22.68% of Tencent, down 6bps vs. the prior results date. On company buybacks, Q1 totaled HKD 7.6bn, down sharply YoY. Daily repurchases fell to HKD 300mn in Mar from HKD 640mn previously; apart from Apr 8, the day before the Q1 blackout began when HKD 1bn was repurchased, buyback intensity did not notably rise despite stock weakness. ![图表, 直方图描述已自动生成](https://pub.pbkrs.com/cms/2026/0/gggFNqT1VSYWvZ7JeCNMYxbPGZvKHh87.jpg?x-oss-process=style/lg) ![图表描述已自动生成](https://pub.pbkrs.com/cms/2026/0/sRgr8ZF2ADA1JhYZwQdzkAHdLLr2VMVp.jpg?x-oss-process=style/lg) **Dolphin Research on 'Tencent Holdings'** **Earnings season (last 12 months)** Nov 13, 2025 call recap: [Tencent Holdings (Trans): prioritizing AI, but trimming full-year capex guidance](https://longbridge.cn/topics/36331836?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=3c23cc7f-6199-4b8d-9c46-94608fd59010) Nov 13, 2025 earnings take: [Tencent: not playing Meta’s heartbeat game, aiming for steady returns](https://longbridge.cn/topics/36331836?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=3c23cc7f-6199-4b8d-9c46-94608fd59010) Risk disclosure and disclaimer: [Dolphin Research disclaimer and general disclosures](https://support.longbridge.global/topics/misc/dolphin-disclaimer) ### Related Stocks - [00700.HK](https://longbridge.com/en/quote/00700.HK.md) - [META.US](https://longbridge.com/en/quote/META.US.md) - [03908.HK](https://longbridge.com/en/quote/03908.HK.md) - [AAPL.US](https://longbridge.com/en/quote/AAPL.US.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [80700.HK](https://longbridge.com/en/quote/80700.HK.md) - [TCEHY.US](https://longbridge.com/en/quote/TCEHY.US.md) - [TCTZF.US](https://longbridge.com/en/quote/TCTZF.US.md) - [601995.CN](https://longbridge.com/en/quote/601995.CN.md) - [AAPB.US](https://longbridge.com/en/quote/AAPB.US.md) - [AAPD.US](https://longbridge.com/en/quote/AAPD.US.md) - [AAPU.US](https://longbridge.com/en/quote/AAPU.US.md) - [AAPX.US](https://longbridge.com/en/quote/AAPX.US.md) - [AAPY.US](https://longbridge.com/en/quote/AAPY.US.md) - [APLY.US](https://longbridge.com/en/quote/APLY.US.md) - [GGLS.US](https://longbridge.com/en/quote/GGLS.US.md) - [GGLL.US](https://longbridge.com/en/quote/GGLL.US.md) - [HTCD.SG](https://longbridge.com/en/quote/HTCD.SG.md)