--- title: "XPEV (1Q26 Trans): Humanoid robot to begin mass production by year-end" type: "Topics" locale: "en" url: "https://longbridge.com/en/topics/41222773.md" description: "Q2 overall GPM is expected to be roughly flat vs. Q1. Approx. 20.6%." datetime: "2026-05-28T14:41:12.000Z" locales: - [en](https://longbridge.com/en/topics/41222773.md) - [zh-CN](https://longbridge.com/zh-CN/topics/41222773.md) - [zh-HK](https://longbridge.com/zh-HK/topics/41222773.md) author: "[Dolphin Research](https://longbridge.com/en/news/dolphin.md)" --- # XPEV (1Q26 Trans): Humanoid robot to begin mass production by year-end **Below is Dolphin Research's**$XPeng(XPEV.US) **Q1 2026 earnings call Trans. For our take on the print, see '**[**XPeng: Inflection Point In Sight — Can the Eastern Tesla Story Dance On?**](https://longbridge.cn/zh-CN/topics/41211888?app_id=longbridge&utm_source=longbridge_app_share&share_track_id=0fb69e4a-37b0-4a5e-98a1-35863ccd17c7&invite-code=4NOXYT&locale=zh-CN&community_badge=1&profile_following_followers_activities=1)**'** **I. Key takeaways from the print** 1\. **Q2 2026 guidance**: deliveries of 100,000–106,000 units (+59.5% to +69.1% QoQ). Revenue of RMB 19.6–20.8bn (+50.4% to +59.6% QoQ). **Q2 overall GPM expected roughly flat vs. Q1 (~20.6%).** 2\. **Q1 2026 revenue and profit**: total revenue RMB 13.03bn (-17.6% YoY, -41.4% QoQ), driven by lower deliveries. Overall GPM 20.6% (vs. 15.6% a year ago, 21.3% in Q4), vehicle GPM 12.1% (vs. 10.5% YoY, 13.0% QoQ). OP loss RMB 1.87bn; net loss RMB 1.78bn (vs. net profit RMB 380mn in Q4). As of Mar 31, 2026, cash and cash equivalents were RMB 42.09bn. 3\. **R&D stepped up materially**: Q1 R&D expense RMB 2.91bn (+46.8% YoY), mainly for new models and AI-related development. SG&A was RMB 1.88bn (-3.2% YoY, -32.5% QoQ), primarily on lower dealer commissions. 4\. **Technology services revenue**: full-year 2026 tech, services and IP licensing revenue guided to be flat vs. 2025. From Q2, XPeng will start volume delivery of the Turing SoC to Volkswagen. 5\. **Intl profitability well ahead of domestic**: latest monthly intl sales are approaching 20% of total (vs. ~10% for full-year 2025). **Intl vehicle sales deliver significantly higher GPM and net profit contribution than domestic.** **II. Call details** **2.1 Management highlights** 1\. **Corporate strategy** a. In Q1, the Chinese corporate name officially changed from 'XPeng Motors' to 'XPeng Group', marking a strategic shift from an intelligent EV maker to a physical AI company. b. Physical AI at scale is the top corporate priority. Three key applications — VLA 2.0 autonomous driving, Robotaxi and the humanoid robot — will be the core growth engines for revenue and profit. c. Management believes the scaling law has been validated in autonomy and robotics, and will continue to ramp AI investment. 2\. **Vehicle biz.** a. Q1 deliveries were 62,680 units, a seasonal trough. Management is balancing scale with operating quality rather than chasing volume for its own sake, and expects sequential deliveries to rise sharply from Q2. b. In Apr, the 2026 XPeng MONA M03 launched, including the MAX trim with Turing AI SoC and the Ultra SE supporting VLA. With that, Turing AI SoC upgrades are now across the lineup. Over 85% of MONA M03 buyers opted for MAX or Ultra SE; the model has led A-segment BEV sedan sales in China for 19 consecutive months. c. On May 20, the flagship GX debuted as an L4-era flagship and China’s first pre-installed, mass-produced Robotaxi with full hardware redundancy. **Over 80% of initial orders are for the Ultra flagship priced above RMB 350k, and the BE flagship lead time exceeds 30 weeks. GX carries the highest margin in the lineup, with most SKUs beating margin expectations.** d. Three new models are planned for 2H26, all with Turing SoC and VLA 2.0, offering both range-extended and BEV options and designed for global markets. e. From this year, new-model pricing and trims will prioritize commercial value, targeting sustainable volume over short-term spikes. 3\. **Intl biz.** a. With the X7 starting overseas deliveries in Apr, monthly intl deliveries surpassed 6,000 for the first time. **From Q2, intl revenue is expected to exceed 20% of total.** b. Targets: **by Q4, sustain 10,000+ monthly overseas deliveries; full-year overseas deliveries to more than double YoY.** c. Manufacturing footprint: plants in Indonesia and Malaysia (mainly for local demand); Magna’s Austria facility for Europe. All three will expand capacity this year and add new models. Most Europe volume is expected to be locally built. d. The Munich R&D center is the fastest-growing. Five-year goal: 50% of revenue and profit from overseas. 4\. **Autonomous driving (VLA 2.0)** a. In Apr, ADAS mileage penetration on VLA 2.0-enabled models exceeded 50% for the first time, showing advanced autonomy as a core user feature. b. The second VLA release will come in Q3 (Aug) with a major step-up in capabilities, stronger generalization and much lower manual intervention. Combined with VRM, it will enable a butler-like voice interaction, rolled out via three OTAs (Aug and year-end). c. VLA 2.0 is under test in Europe, with overseas deliveries targeted next year pending regulatory approvals in multiple countries. d. Long-term goal: deliver L4-level software capability on L2 hardware, implying a step-change in the business model. 5\. **Robotaxi** **a. Deployment is currently limited to Guangzhou (licensed). After validating tech, product and model there, XPeng will scale across China and globally.** **b. A dedicated, cost-efficient Robotaxi model is planned for 2027 to further validate the model, with large-scale commercialization expected from 2028.** **c. Model: XPeng will provide the tech platform and take commissions, without operating fleets. Localized operations will be run by multiple partners at home and abroad.** d. Recent tightening of China’s AV regulation has not adversely affected development cadence. 6\. **Humanoid robot (IRON)** a. Mass-production IRON hardware and software are progressing well, entering the ETG systems integration phase. Automotive-grade safety and reliability standards will be applied, with many existing auto suppliers also providing robot components. **b. The next-gen in-house dexterous hand has been completed, offering greater agility and much lower cost.** **c. Targets: SOP by year-end; initial pilot deployments in XPeng-owned stores; deliveries to enterprise customers in China and overseas from 2027.** d. A next-gen IRON will be showcased in Q3 with multilingual interaction, full-body motions and stepwise autonomy in executing professional tasks. e. Use cases: initially as greeter and product specialist in showrooms. Later, open to ecosystem partners for more scenarios such as retail checkout and pro services. **f. Commercial model: hardware ASP above autos with higher hardware margins, plus software/AI model licensing revenue. Payback periods are shorter overseas than in China, implying greater commercial value.** g. XPeng is the only robotics company in China with full-stack in-house capabilities from SoC and physical AI foundation models to data generation, pre-train/finetune, and from joints and dexterous hands to motion control. **2.2 Q&A** **Q: How are GX orders trending, what is the steady-state sales target, and what is the margin outlook?** A: GX is outperforming expectations across the board. The BE flagship’s lead time is now over 30 weeks and still extending; Ultra flagship orders exceed 80% of total and are rising. Notably, MAX is only about 5% of the mix, slightly below prior expectations; the range-extended variant started slower but is catching up quickly in Northwest and North China as we step up marketing. We are confident in GX’s performance in the RMB 300k+ premium segment. Margins are strong: as the group’s flagship, most SKUs are ahead of margin expectations, with only one SKU slightly below. From this year, we are prioritizing commercial value in trims and pricing to drive sustainable sell-through rather than a front-loaded spike, and will emphasize product management and modular supply-chain quality to secure supply, support the ramp and stabilize deliveries. **Q: Progress of Robotaxi operations in China and overseas, any regulatory impact from recent tightening, and how will B2B Robotaxi affect B2C sales?** A: The recent tightening of China’s AV oversight has not affected our development pace. We see true large-scale Robotaxi opportunities after 2028 and are methodically preparing for that, advancing VLA 2.0 deployment and testing in both China and overseas. In China, we are moving fast in Guangzhou (licensed), iterating on mass-produced models. **We plan to launch a cost-efficient model in 2027 dedicated to validating and showcasing the Robotaxi business model in China, and we are confident in achieving high success.** XPeng will not operate fleets; we will provide the tech platform and collect commissions, with multiple partners running localized operations globally. Since announcing the plan, we have received extensive inbound from potential partners at home and abroad who are closely tracking our progress. On B2C impact: Robotaxi testing and R&D will benefit B2C, as the VLA model yields various ADAS strategies (e.g., speed modes, reduced-intervention modes) that directly enhance retail user experience. Over time, B2B (Robotaxi ops) and B2C (retail sales) will form distinct segments, a split that will sharpen as regulations open up. **Q: What is the capability roadmap for VLA 2.0 over the medium to long term?** A: In 2026, VLA (or VLA+VRM) will hit two milestones. **First, in Q3 (Aug), we will release the second VLA iteration with higher intelligence, stability and generalization, and a marked drop in manual intervention/takeover rates. VLA 1.0 focused on baseline safety, engineering quality and basic UX, with a limited sense of headroom; the Aug release should bring a step change. Coupled with VRM, we will deliver not only ADAS strengths but a butler-like voice experience, rolled out in three OTAs (Aug and year-end).** **Longer term, we aim to achieve L4 software capability on L2 hardware, which would fundamentally change the business model. This is the direction we are pushing toward; further details to come in due course.** **Q: Where does XPeng’s humanoid robot have a cost advantage vs. peers, and what is the long-term export strategy for 2027?** A: Scaling a humanoid robot is a different challenge from autos — safety, hardware reliability, maintainability, supply chain and data privacy are all different. **Most products on the market are consumer-grade in quality, while we target automotive-grade production quality and reliability. Many peers have not seriously addressed supply-chain readiness at scale and the data-security trade-offs between edge and cloud.** We began a full redesign in 2025 leveraging full-stack in-house hardware and software to unlock commercial potential more systematically. **Batteries and other key components are fully in-house; today’s cost stack resembles autos, and we will continue to explore more scalable and economical approaches.** **For overseas, like our vehicles, the robot has been designed as a global product from day one, with compliance on hardware, software and data privacy. The commercial opportunity overseas could exceed China’s given labor substitution economics. Our VLA model generates ~200mn user events per hour; if processed in the cloud, that would be ~100GB per hour, so we architected for localized deployment from day one. As our China R&D system advances, we expect bigger international breakthroughs.** **Q: Plans to expand Robotaxi beyond Guangzhou and license progress?** A: Deployment is currently confined to Guangzhou where we hold an operating license. Our strategy is to go deep in Guangzhou first to validate tech, product and model, then expand nationwide and globally with multiple partners running local operations. Since announcing the plan, we have received substantial inbound interest from potential partners globally who are watching closely. **We believe that after launching a dedicated cost-efficient Robotaxi in 2027 and delivering a more complete solution, our commercialization capabilities at home and abroad will improve significantly.** **Q: What is the guidance for Q2 overall GPM and vehicle GPM?** A: **Q1 gross profit in absolute terms was similar to Q4, while we faced cost pressure from higher memory costs and battery raw materials that began to flow through in Q1 and are likely to persist through the year. GX launched in Q2 and will start deliveries in subsequent quarters; as our highest-margin model, the product mix should improve through Q2 and 2H. Netting these factors, we expect Q2 overall GPM to be roughly flat vs. Q1.** **Q: What are the long-term growth drivers in overseas markets, and how does local manufacturing compare with exports on profitability?** A (He Xiaopeng): Intl biz. is one of our four core strategies. **Over the next five years, we target 50% of revenue and profit from overseas. With four new models coming in 2H, our overseas coverage will expand materially, and scale benefits should be more pronounced from 2027–2028. Hardware, software, sales channels and localized charging and aftersales have incorporated overseas profitability considerations for years.** A (Brian Gu): Latest monthly data show intl volume near 20% of total (vs. ~10% in 2025), a meaningful step-up. Intl vehicle sales — on both GPM and net contribution — are well ahead of domestic, and remain so even with tariffs and cost inflation this year. I am in Austria with Magna to secure capacity for Europe and global demand. **I expect intl contribution to hold around 30% for the year, a key driver of overall growth and profitability.** **Q: What is the cost composition of overseas local production and how do localization rates vary by region?** A: We operate two Southeast Asia plants (Indonesia and Malaysia) primarily for local markets, and a partner facility with Magna in Austria for Europe. All three will expand capacity and add new models this year. Southeast Asia is mainly local-demand focused; Austria is the core for Europe with most European sales expected to be locally built. Regions without local manufacturing will continue with exports for now, while we explore capacity expansion and deeper localization to meet rising local-content requirements as our share grows. **Q: What are the humanoid robot’s application scenarios and pricing strategy?** A: Our humanoid differs from most in being designed for human-dense environments and interaction. Initial deployment will be in our own stores where the robot acts as a greeter and product specialist to introduce vehicle features before handoff to sales, lifting throughput significantly. As capabilities mature, we will open to ecosystem partners in verticals like retail checkout or smart shopping assistants, co-generating training data for pre-train and finetune to build scenario-specific models. On commercialization, our plan is still under validation. **The robot’s skeletal architecture is similar to cars, but ASP will be higher with superior hardware margins; given substantial on-board compute, software licensing and cloud monetization also have large upside, making the overall model more attractive than autos.** For end customers, payback is the key metric; overseas payback is significantly shorter than in China, so the economics are stronger and the overseas outlook larger. We are focused on the purchase and usage logic to ensure the model is compelling for all parties. **Q: As XPeng rebrands as a physical AI platform company, how will the biz. model evolve, and what is the commercialization timeline across segments and the future revenue mix?** A: Today, most revenue comes from hardware and the ecosystem. Globally, few automakers have managed to build true software platforms and networks on top of hardware volume — XPeng has that potential, spanning software fees, software revenue and lateral network effects. Looking ahead, we see the agent era driving an ant-colony effect — reinforcing network effects across distributed and centralized clusters. These compounding effects (scale, brand, margin) will strengthen our moat. Near to mid term, we will first drive brand premium, scale and GPM uplift. Global expansion and higher margins should improve overall profitability, which in turn funds sustained R&D to reinforce tech moats and form a virtuous cycle. **Q: How will 2026 services revenue trend, and can the Volkswagen partnership extend beyond China?** A: **We guide 2026 tech, services and IP licensing revenue to be flat vs. 2025. From Q2, we will deliver the Turing SoC to Volkswagen at scale; we see such tech commercialization as highly attractive. With a broad stack of proprietary technologies, we are open to expanding commercialization with more markets and partners.** **Risk disclosure and statement:**[**Dolphin Research Disclaimer and General Disclosure**](https://support.longbridge.global/topics/misc/dolphin-disclaimer) ### Related Stocks - [TSLA.US](https://longbridge.com/en/quote/TSLA.US.md) - [09868.HK](https://longbridge.com/en/quote/09868.HK.md) - [XPEV.US](https://longbridge.com/en/quote/XPEV.US.md) - [TSDD.US](https://longbridge.com/en/quote/TSDD.US.md) - [TSLL.US](https://longbridge.com/en/quote/TSLL.US.md) - [TSLQ.US](https://longbridge.com/en/quote/TSLQ.US.md) - [09366.HK](https://longbridge.com/en/quote/09366.HK.md) - [07766.HK](https://longbridge.com/en/quote/07766.HK.md) - [07366.HK](https://longbridge.com/en/quote/07366.HK.md) - [TSLR.US](https://longbridge.com/en/quote/TSLR.US.md) ## Comments (1) - **定投标普20年 · 2026-05-28T17:22:50.000Z**: XPeng's crappy car is already selling at this price, yet it still has a 20% gross margin. How much cost-cutting must be going on?