--- title: "Xiaomi Corp (Trans): No guidance, no order updates" type: "News" locale: "en" url: "https://longbridge.com/en/news/296232967.md" datetime: "2026-08-18T13:02:28.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/296232967.md) - [en](https://longbridge.com/en/news/296232967.md) - [zh-HK](https://longbridge.com/zh-HK/news/296232967.md) --- # Xiaomi Corp (Trans): No guidance, no order updates **Below is Dolphin Research's Trans of Xiaomi Group's FY26 Q2 earnings call** **I. Core takeaways** 1) Shareholder returns: Since 2026, Xiaomi has repurchased approx. HK$11.7bn. This already exceeds the full-year total last year, and the company will keep buying back in the open market to enhance shareholder value. 2) Outlook: Management gave no specific revenue or profit guidance. They flagged near-term pressure in Q3 from elevated memory costs, a slower demand recovery, and intense competition, while keeping R&D spend on the prior trajectory for this year and next. 3) Key quarterly metrics. The details are as follows. Total: revenue of RMB 108.9bn; GPM 19.8%; Adj. NP RMB 6.2bn. Smartphone × AIoT: revenue RMB 84.0bn; GPM 20%. Smartphones: revenue RMB 42.1bn, 38.7% of total; GPM 8.5% vs. 9.3% in 1H; global shipments 31.2mn units. IoT & lifestyle products: revenue RMB 31.3bn; GPM 20.1%. Internet services: revenue RMB 9.0bn; GPM 76.8% (+140bps YoY). Global MAU 770mn in Jun (+4.8% YoY); Mainland China MAU 198mn, a record high (+7% YoY). Smart EV, AI & other innovation: revenue RMB 24.9bn (+17.1% YoY), 23% of total; auto sales RMB 23.9bn, other related RMB 1.0bn. GPM 19.2%; OP loss RMB 2.6bn. 4) R&D and capex: Q2 R&D expenses RMB 9.2bn (+18.9% YoY); 1H R&D RMB 18.2bn (+25.6% YoY) with AI accounting for nearly 30%. Q2 capex RMB 3.6bn, with smart EV and AI/innovation at 65.8%. 5) Segment disclosure: MiMo LLM monetization currently sits under the 'smart EV, AI & other innovation' segment within the RMB 1.0bn 'other related biz.', still small in scale. It may be disclosed separately when appropriate. **II. Details from the call** **2.1 Management remarks** 1) Smartphones. Xiaomi proactively optimized its mix and raised prices in Q2, pushing ASP to a record high; shipments fell YoY but, per Omdia, it remained top-3 globally for the 24th consecutive quarter. Shipments ranked top-3 in 53 countries/regions and top-5 in 67. In SE Asia, LatAm, and Middle East it ranked No. 2 with shares of 19.3%, 16.2%, and 13.5% respectively; in Europe and Africa it ranked No. 3 with 16.5% and 10.9% share. In Mainland China, high-end models accounted for 32.1% of regional smartphone sales, a new high. Memory costs are at historical highs, but Xiaomi offset through mix upgrades, software optimization, and better ops, delivering Q2 GPM of 8.5%. 2) IoT & lifestyle products. China revenue declined YoY off last year's high base from state subsidies, while overseas revenue grew strongly on channel expansion and a broader category set. As of Jun 30, over 640 new retail stores were opened overseas. Tablet shipments rose to No. 4 globally, while TWS earbuds and smart bands ranked No. 2 globally. Xiaomi will attend IFA Berlin in Sep to showcase its phone–car–home ecosystem. Major appliances will begin a full-scale entry into Europe. 3) Smart EV. Q2 deliveries were 104,199 units, the sixth straight quarter of YoY growth; in 1H, the SU7 series ranked No. 1 by sales among BEV sedans priced above RMB 200k in Mainland China. As of Aug 17, 2026, cumulative SU7 deliveries topped 500k. In Jul, Xiaomi held an Auto Tech event and unveiled its first range-extended architecture, the 'Xiaomi Kunlun Tech Architecture'. Based on Kunlun, it launched the SkyNomad range-extended SUV line, positioned as a smart variable-space SUV. Pre-sale prices: N90 MAX at RMB 299,900 and N70 MAX at RMB 259,900; early deposits are strong, with official launch in Sep. In Jun, the Xiaomi YU7 GT Track Edition set a 10:29.483 autonomous lap record at Nürburgring Nordschleife, developed by its Europe R&D center and domestic ADAS team to improve mass-production safety under extreme scenarios. 4) MiMo base LLM. The MiMo-V2.5 series launched in Apr. On Aug 2, CCTV reported MiMo-V2.5 ranked No. 1 on OpenRouter's weekly global model invocation list, with tokens rising from 1.5tn to 10.5tn in two months, a 6x+ jump. Management attributes this to strong multimodal perception, agent and coding capabilities, competitive pricing, and full-series open source. In Jun, Xiaomi open-sourced MiMo Code (an exploratory AI coding assistant), released MiMo Claw GA, and partnered with Kingsoft Office on an end-to-end productivity solution. It also released Miloco 2.0, an open-source whole-home smart solution centered on MiMo, moving devices from command execution to 'thinking'. With rising usage, API calls and token bundles started contributing revenue this quarter; Xiaomi will keep iterating the base model to reach first-tier performance at the same parameter class. 5) AI phones and OS. Xiaomi believes AI phones must embed AI into the OS; the newly released HyperOS 4 is a major start, upgrading from in-app Q&A to an assistant that learns habits, understands intent, and gets tasks done. Powered by the in-house MiMo model and AI-native OS changes, 'Super Xiaoai' upgraded to 2.0 with new 'Dynamic Island' and 'Inspiration Bar' interactions, offering quick and expert modes. Expert mode integrates deeply with system apps to complete complex tasks across apps and devices. An expert-mode points-based membership was introduced: each user receives 1,000 free points per month. Heavy AI users can choose flexible subscriptions, with monthly promos starting at RMB 19. 6) Embodied AI robots. In Jun, the robotics team won two championships at CVPR 2026 and ICRA 2026 WBC. At the Xiaomi Auto plant, after a quarter of tuning, success rates for dual-sided tasks and self-tapping insert feeding rose to 98%. The team is piloting new workstations in the GA logistics area, with a 90% success rate for center-console side-cover tote folding. In Jul, Xiaomi released Xiaomi-Robotics-U0 and Xiaomi-Robotics-1: U0 is the first unified generative model in embodied AI that handles all four task types, targeting industry data bottlenecks. Robotics-1 was pre-trained on 100k hours of real-world data and further trained with cross-ontology data, ranked No. 1 in multiple sim benchmarks, and open-sourced on Aug 5. 7) Brand, ESG, and next-stage strategy. In Jul, Xiaomi made the Fortune Global 500 for the 8th straight year, ranking No. 232, up 65 spots YoY and a new high since first inclusion in 2019. It ranked No. 2 in Kantar's 2026 Top 50 Chinese Global Brand Builders. Xiaomi Titanium 2.0, released in Jul, cuts CO2 emissions by approx. 93% vs. conventional primary aluminum, has obtained EPD certification and registration, improves toughness with unchanged strength, and meets stringent auto structural standards. It is in mass production in the new-gen SU7 and Xiaomi YU7 series. Following disasters in Guangxi, Hubei, Gansu, Liaoning, Jilin, and Hebei in Jul, Xiaomi Foundation donated RMB 10mn in cash. As of Jun 30, 2026, the Xiaomi Innovation Joint Fund had granted RMB 274mn to support 182 teams in Mainland China. While Q3 still faces high memory costs, slow demand recovery, and intense competition, short-term pressure does not change its long-term strategy. Xiaomi will continue to invest in core tech across AI, chips, OS, and embodied AI. **2.2 Q&A** **Q: After memory price hikes in 1H, has the smartphone biz found a dynamic balance among pricing, shipments, and GPM? Can you sustain stability via price hikes, tighter volumes, and mix upgrades over the next few quarters?** A: Several factors matter. First, the magnitude and pace of memory cost inflation; we expected a long upcycle, but the surge in Q4 last year and Q1–Q2 this year exceeded our expectations and is concerning, and not all costs can be passed through. We had to streamline the lineup and adjust mix, which takes a longer planning and launch cycle. Entry-level models are hit hardest: based on Q2 costs, even models priced above RMB 1,500 were affected by memory alone, excluding other costs, implying entry-level phones need to sell at RMB 2,000 for standard storage, not oversized storage. These moving parts require alignment across functions. After Q2 adjustments, shipments fell YoY but we remained No. 3 globally; ASP hit a record high, up almost RMB 300 QoQ, and GPM held at 8.5%, better than feared. Into Q3–Q4, memory costs are still rising but at a slower pace and are manageable; earlier, when increases were steep, the impact was significant. In Q3, the Redmi flagship, typically launched in Oct, was pulled forward to Aug, with Pro Max and Camera Pro launched together. Camera Pro is performing well, with 62%+ mix of 16GB and above, and ASP a little over RMB 5,000. Sales are ongoing, but vs. the prior gen, growth is 60%; overall, the business is under control. **Q: How are SkyNomad orders so far? What differentiates it vs. rival SUVs, and what resonates most with buyers?** A: SU7, YU7, and SkyNomad target distinct user groups, and we have achieved clear differentiation with minimal overlap with SU7 users. SU7 and YU7 emphasize driver experience, while SkyNomad focuses on space, aiming to deliver an excellent spatial experience with many use-case breakthroughs. Post-launch deposits were strong, and our analysis shows families and users with more passengers will be key buyers, with an average age a few years older than SU7 buyers. The flexible, variable space draws significant attention; even before hands-on experience, it has sparked imagination, and the car already supports many scenarios; we are optimistic on orders. **Q: Please outline the AI monetization roadmap across near/mid/long term, and provide updates. Of the RMB 16bn AI budget mentioned at the start of the year, how much was spent in 1H and Q2, and any change to the full-year plan?** A: We launched MiMo-V2.5 in 1H, and it has been well received globally. By late Jul, weekly token calls ranked No. 1 globally, and monthly calls also ranked No. 1, indicating developers are voting with usage; we have systemic advantages in capability, efficiency, and cost control. Next, we will roll out new apps to broaden access to MiMo, including our first desktop app for PC, and another new release now in training. Monetization-wise, we are still in a heavy investment phase and will proceed steadily, focusing on the phone–car–home ecosystem. With HyperOS 4.0, MiMo is integrated into the OS; Super Xiaoai 2.0 is a key step, and further progress will deepen MiMo's integration with phones and OS. At the base layer, we have opened some APIs and token bundles that are contributing revenue, but this business has just started; monetization is not the near-term priority, and we will keep iterating models and deploying them in OS scenarios. **Q: The smart EV and innovation segment GPM fluctuated YoY and QoQ. Can you break down the drivers and AI's impact? With N70/N90 MAX pre-sale pricing well received, what will drive segment GPM in Q3–Q4?** A: Three factors drove the fluctuation. YoY, Q2 last year included SU7 Ultra deliveries, which had relatively better margins, making last year's Q2 a higher base; this quarter, Ultra's contribution was weaker. QoQ, we delivered more next-gen SU7s (launched in Mar), and the SU7/YU7 delivery mix differed from Q1; we had flagged higher costs for the new SU7, so as SU7's mix rose, GPM fell QoQ. Third, LLM revenue recognition began and negatively impacted the segment's GPM. SkyNomad N70/N90 will officially launch in Sep; early feedback is positive, but final pricing is not set. Margins will depend on final pricing, which will influence SkyNomad's GPM in Q3–Q4. **Q: Raw materials and inventories rose sharply in 1H. How much was strategic stocking (e.g., memory) vs. supporting current profit and GPM? With finished goods down quickly, could margins face pressure in coming quarters?** A: Two points on inventory. First, raw materials rose QoQ from close to RMB 30bn last quarter to just over RMB 40bn in Q2, driven by forward stocking—especially memory—as well as higher unit prices, not just volumes. Second, finished goods declined as expected, given 618 stocking and subsequent sales; this is straightforward. A further factor: next-gen SU7 deliveries started in early Feb, with some production brought forward before quarter-end, so overall inventories fell in Q2, also reflecting the auto business. **Q: IoT overseas revenue jumped in Q2. Are overseas retail stores expanding as planned and on track? With more categories and EV exports from 2027, are you preparing for site selection, factories, and large-store expansion abroad? What is the outlook for IoT and EV overseas?** A: IoT overseas has vast headroom and is on track. Xiaomi has 630+ Mi Home stores overseas and will keep expanding next year; these will anchor IoT and premium phones, covering about 30% of overseas premium smartphone sales, laying a strong foundation. For IoT, especially major appliances, we were previously a blank in overseas markets; growth will accelerate in 2H and expand significantly next year. Market access barriers vary by country, requiring heavy compliance work; we will accelerate entry into more countries. EV exports are also on track for 2H 2027. We visited many countries this year; partners and dealers are highly interested in representing Xiaomi, seeing us as a differentiated, scarce Chinese brand capable of premium products. They view Xiaomi as a tech company building EVs, not a traditional automaker. In discussions, we mostly engaged top local dealers—top 10 in each market—with 7–8 often approaching us proactively; we also visited many countries and clients. **Q: Smartphone shipments were constrained by memory costs in Q1–Q2, yet internet services kept growing. If memory pressure eases or stabilizes, how should we view growth and GPM for internet services?** A: Internet services include flows tied to shipments and to the installed base. The installed base hit records with 770mn global MAU and nearly 200mn in Mainland China, underpinning revenue stability. Our push into premium phones has made progress, lifting the share of high-end users in MAU; premium users contribute higher ARPU than low-end users. Ads have been healthy in recent quarters; the recent 'price war' in food delivery spurred ad spending, which supported mobile ads—together these factors strengthened stability. **Q: Beyond base models, you mentioned new apps, MiMo Claw, and smart home appliances. When will full integration of AI across phone–car–home hardware be visible?** A: Our AI plan starts with the base LLM as the heaviest investment to serve the phone–car–home ecosystem; this is the bottom layer. Above that are several lines. First, smartphones: we are rebuilding Android with AI, and HyperOS 4 is a big step, enabling agents to execute many actions. Second, home: Miloco is the key lever, with many scenarios landing; some cloud components still carry higher cost but will improve as costs decline. Third, autos: smart driving. Fourth, robots: multiple versions already released. All lines are deeply integrated with MiMo, developing both independently and convergently to form a stronger full-ecosystem model. **Q: We can already see AI-related revenue. Any color on trajectory? Will R&D spend be adjusted under memory cost pressure?** A: AI revenue sits in the 'smart EV, AI & other innovation' segment. MiMo monetization has just started, and revenue remains small; it may be disclosed separately at the right time, but for now it remains within that RMB 1.0bn bucket. Operating conditions still face significant memory-cost pressure. R&D this year and next will follow the previously shared trajectory with no adjustment; we are resolute on this. **Q: What is the expected market share for Xiaomi smart EVs for the full year and 2H?** A: We cannot disclose specific percentages. Our goals focus on two main lines with SkyNomad built on Kunlun, and targets tied to smart EV technologies and intelligent driving. **Q: Will EV ASP keep rising? How is profitability shaping up?** A: ASP moves with product mix. YU7 deliveries were higher in Q1, SU7 higher in Q2, and SU7 is priced below YU7, causing ASP swings; last year's Q2 had a higher Ultra mix, and Ultra's ASP is far above average. ASP is an outcome rather than a target; it does not directly equate to margin. Lower ASP does not necessarily mean lower GPM, which is an important nuance. **Q: Q2 smartphone ASP rose 21.9% YoY to a record, with an increased mix above RMB 3,000, though high-end saw some declines. If memory costs keep rising, how will the global smartphone landscape evolve? What are the product-mix strategies for 2H and next year?** A: On memory pricing, I doubt it will stay at extremely high levels long term, though it should remain well above the prior trough. Current prices are 5x last year's Q2 level, which is unlikely to persist; where it settles is uncertain, likely somewhere in between. Globally, costs and pricing will reach a new equilibrium over time, but near term it's hard to be precise given push-pull dynamics; for example, Apple held prices, then recently signaled price hikes for the next flagship, while others are split on raising or cutting. Over the long run, everyone will incorporate memory cost moves into pricing, driving a dynamic balance; smartphones are near-essential, so I am relatively optimistic. **Q: With the robotics plant now in production, are robots positioned mainly for cost reduction or as a future revenue driver? Any quantifiable targets?** A: Robotics will need time to achieve large-scale maturity, so short-term targets are hard to set. We will keep investing with a positive long-term view and pursue integration with existing businesses, which is why we are investing. We have not considered moves in food delivery. With in-house factories, many application scenarios will emerge; capabilities such as LLMs will converge with robotics and also create synergies with chips and OS, while robotics advances will feed back into model capabilities; we expect broad synergies but will stay pragmatic without overly specific targets for now. **Risk disclosure and statement for this article:**[**Dolphin Research Disclaimer and General Disclosure**](https://support.longbridge.global/topics/misc/dolphin-disclaimer) ### Related Stocks - [01810.HK](https://longbridge.com/en/quote/01810.HK.md) - [81810.HK](https://longbridge.com/en/quote/81810.HK.md) - [HXXD.SG](https://longbridge.com/en/quote/HXXD.SG.md) - [XIACY.US](https://longbridge.com/en/quote/XIACY.US.md) - [688111.CN](https://longbridge.com/en/quote/688111.CN.md) ## Related News & Research - [Kingsoft Office, Kingsoft Corp subsidiary, posts H1 net profit attributable to shareholders up 236.94% to RMB 2.52 billion](https://longbridge.com/en/news/296344953.md) - [Kingsoft FY26 H1 profit attributable to owners more than doubles to RMB 1.64 billion; revenue rises 6% to RMB 4.93 billion](https://longbridge.com/en/news/296340938.md) - [Xiaomi SU7 series tops 500,000 cumulative deliveries](https://longbridge.com/en/news/296067965.md) - [Xiaomi EV delivers 31,267 units in July, 39% of annual target](https://longbridge.com/en/news/295642276.md) - [Luxeed RX to open pre-orders on August 20 with 4 LiDAR sensors](https://longbridge.com/en/news/295761523.md)