---
title: "Xiaomi (Trans): Paid tokens >30%; confident of hitting 550k vehicles target"
type: "Topics"
locale: "en"
url: "https://longbridge.com/en/topics/41121434.md"
description: "Below is Dolphin Research's Trans of Xiaomi Corp.'s FY26 Q1 earnings call. For our take on the results, cf. 'Xiaomi: Questioning Life? The worst is behind us'.I. $XIAOMI-W(01810.HK) core takeaways recap. 1) Shareholder returns: YTD 2026, the company has repurchased approx. HK$8.4bn in the open market, already exceeding last year's total buybacks. 2) Overall performance: Q1 revenue of RMB 99.1bn, Adj. net profit of RMB 6.1bn; overall GPM at 22%..."
datetime: "2026-05-26T15:05:50.000Z"
locales:
  - [en](https://longbridge.com/en/topics/41121434.md)
  - [zh-CN](https://longbridge.com/zh-CN/topics/41121434.md)
  - [zh-HK](https://longbridge.com/zh-HK/topics/41121434.md)
author: "[Dolphin Research](https://longbridge.com/en/news/dolphin.md)"
---

# Xiaomi (Trans): Paid tokens >30%; confident of hitting 550k vehicles target

**Below is Dolphin Research's transcript summary of Xiaomi Group's FY26 Q1 earnings call. For our take on the results, see '**[**Xiaomi: Questioning Life at the Bottom? The Worst Is Over**](https://longbridge.cn/en/topics/41112391?channel=SH000001&invite-code=294324&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=c42e93e0-c27f-4870-bf87-a61a100af9ca)**'.**

**I.** $XIAOMI-W(01810.HK) **Key takeaways**

1\. **Capital returns**: YTD 2026, the company has repurchased approx. HKD 8.4 bn in the open market. This already exceeds the full-year total in 2025.

2\. **Group results**: Q1 revenue was RMB 99.1 bn with adj. net profit of RMB 6.1 bn. Overall GPM was 22%.

3\. **R&D and capex**: R&D expenses were RMB 8.95 bn (+33.4% YoY), and capex was RMB 3.27 bn (+20% YoY). Innovation areas including smart EV and AI accounted for 45.6%.

4\. **Losses in EV and other innovative businesses**: The smart EV, AI and other innovation segment posted an OP loss of RMB 3.1 bn in Q1. This was mainly due to delivery phasing, the fade of purchase tax subsidies, and higher AI investment.

5\. **AI investment guide**: Management initially guided AI investment of approx. RMB 16 bn for the year. They will adjust dynamically based on business progress and ROI, and the budget could be raised further.

**II.** $Xiaomi Corporation(XIACY.US) **Call details**

**2.1 Management highlights**

1\. **Smartphones**

a. Q1 revenue was RMB 44.3 bn, contributing 47% of group revenue. Global shipments were 33.79 mn units, and ASP hit a record RMB 1,310 (+8.2% YoY). In mainland China, premium models accounted for 23.5% of smartphone volume.

b. Per OMDIA, global share was 11.3%, ranking top 3 worldwide for 23 straight quarters. Xiaomi ranked No.2 in LatAm with 17.4% share (+200bps), and No.3 in Europe/SEA/Middle East/Africa at 17.2%/16.9%/13%/9.2%. It placed top 3 in shipments in 47 countries and regions, and top 5 in 65.

**c. Xiaomi proactively moderated mid/low-end shipments and channel inventory to absorb a 'super-long cycle' surge in memory costs, avoiding simple pass-through. Through mix upgrades, software optimization and operational discipline, smartphone GPM stayed at a relatively healthy 10.1% in Q1.**

2\. **IoT**

a. Q1 revenue was RMB 24.7 bn. Domestic revenue declined YoY on a tough state-subsidy comp last year, while overseas channel expansion and category broadening drove record-high overseas revenue with double-digit YoY growth. Overseas contributed about 40% of IoT revenue.

b. Global category ranks: No.2 in TWS, No.3 in wearables, and No.5 in tablets. Xiaomi is exiting price wars, pursuing a dual-track strategy of premiumization in China and scale overseas.

c. GPM was 25.2%, up 510bps QoQ, showcasing cross-business hedging against single-line volatility. IoT helped offset pressure on smartphone margins this quarter.

3\. **Internet services**

a. Q1 revenue was RMB 9.5 bn (+4.3% YoY) with GPM of 76.1%. Advertising revenue was RMB 7.1 bn (+7.8% YoY), the main growth driver.

b. Global MAU reached 750 mn in Mar 2026 (+3.8% YoY), and mainland China MAU hit a record 196 mn (+8.1% YoY).

4\. **Smart EV, AI and innovation**

a. Segment revenue was RMB 19.9 bn (+6.9% YoY), contributing 20% of group revenue. Smart EV sales were RMB 19.0 bn, with RMB 0.9 bn from related businesses. Xiaomi delivered 80,856 vehicles in Q1 (YU7 only; next-gen SU7 in transition), and ASP after tax was about RMB 235k. GPM of 20.1% fell QoQ mainly due to purchase tax subsidy fade (RMB 10k–15k per car), demo car sales, and higher battery/commodity costs.

b. Next-gen SU7: cumulative orders exceeded 26k by Apr 23, 2026. Total reservations for the new 7 series (incl. YU7) surpassed 80k in the first 48 days. As of Apr 30, 2026, the 7 series delivered 232k units cumulatively over 10 months.

c. On May 21, Xiaomi launched the performance SUV YU7 at RMB 389,900, setting an SUV lap record at the Nürburgring. The standard YU7 is priced at RMB 233,500. Over 50% of buyers opted for the RMB 429,000 high-spec trim, lifting ASP and margins.

5\. **AI foundation models and on-device AI**

a. In Apr 2026, Xiaomi released the MiMo-V2.5 family (Pro 2.5 TTI Series and 2.5 ASR). MiMo-V2.5-Pro ranked No.1 on Artificial Analysis' global open-source LLM comprehensive intelligence list, top 5 on the overall global LLM board, and No.1 among open-source models on the Agent index.

b. By end-Mar, MiMo-V2-Pro topped OpenRouter daily/weekly/monthly charts. Weekly token consumption reached 4 tn in the last week of Mar, and post-free-trial weekly retention was ~35%. On Apr 3, Xiaomi launched the MiMo Token Plan (Lite/Pro/Max), with Pro/Max contributing over 50% of revenue.

c. OpenRouter data on May 12 showed MiMo accumulated 1.45 tn tokens over the prior month, leading globally among open-source Agent models. Under the MiMo 100-tn Token Plan, Xiaomi aimed to distribute tokens free to developers worldwide in 30 days, with nearly 80 tn issued by May 12.

d. For AI phones, 2026 is a pivotal year. Xiaomi MiCo passed the first authoritative evaluation by CAICT for mobile agents, and has expanded to tablets, PCs and smart displays. HyperOS will evolve with Agent at the core.

e. Smart driving: In Mar, Xiaomi unveiled its end-to-end model architecture, and in May, released and open-sourced the Xiaomi semi-automated driving model built on XLA, unifying VLM and latent-space reasoning across three technical tracks.

f. Embodied robotics: On Apr 27, Xiaomi released Xiaomi-Robotics-0 and conducted its first live demo at CMG Investor Day. Management views it as the ultimate platform integrating AI, chips, OS and manufacturing.

**2.2 Q&A**

**Q: Q1 AIoT margin was strong and new products sold well. Can revenue maintain a high growth pace over the next two quarters, and what are the highlights in new products and overseas?**

A: In Q3 last year, we anticipated a super-long upcycle in memory with large price hikes, which would hit memory-intensive products like tablets. AIoT is critical to the group, helping offset memory cost pressure, so we raised AIoT's strategic weight from Q3 onward with differentiated domestic and overseas playbooks.

In China, we are committed to premiumization. This year's launches earned strong reviews on design, quality and deep AI assistant integration, and demand has been robust, with ACs praised for fast cooling and airflow, and refrigerators for freshness, aided by renovation demand.

Overseas, our share remains low but the runway is large. Since last year, we solved localization in many countries, broadened the product matrix, and deepened partnerships with multiple e-commerce platforms, driving a doubling of overseas AIoT revenue. Looking ahead, share in categories like earbuds and watches remains modest in China, and the overseas market is over 2x China's size with different competitive dynamics, so IoT headroom is substantial. Financial addendum: **Q1 is not a peak season, but overseas AIoT posted double-digit YoY growth to a record high, and overseas contributed about 40% of IoT revenue**.

**Q: How do next-gen SU7 and YU7 buyers differ from prior models, and what are the implications for volume, revenue and profit?**

A: We simultaneously launched two YU7 trims, the Standard and GT. Conceptually, SU7 is closer to Model 3, while YU7 still has room for differentiation, and **our teardown of Model Y shows the base version makes up 70% of its sales, where we face some cost disadvantages**.

For daily use, 600 km range already covers most scenarios, while longer range requires a cost trade-off, so we offered the Standard and GT combo. The GT set the Nürburgring SUV lap record and is a performance benchmark, and over half of locked-in buyers chose the fully loaded RMB 429k trim.

Our research shows many YU7 buyers are converting from Mercedes and BMW. Initial GT monthly capacity is just over 2,000 units, with relatively short lead times, while the 600 km Standard fits daily commuting and matches its target budget band. With a RMB 20k price spread between trims, GT carries higher margins; volumes still need to be observed, but we are confident in GT's contribution.

**Q: Xiaomi's AI progress surprised the market this year. Beyond the RMB 16 bn AI budget, can you share MiMo ops metrics cadence, and the roadmap/timeline for MiCo and the next-gen AI OS?**

A: We are all-in on AI, rapidly rolling out MiCo across tablets, wearables, EVs and smart speakers. The key is unifying user data and elevating UX, which depends on the foundation models, upper-layer frameworks and HyperOS evolution, and all teams are pushing in sync, including this year's OS release.

On Token, the paid conversion is very high, with over 30% of tokens deployed from paid tiers and more than 50% from Max. Users welcome new versions, and Pro/Max account for the majority of revenue. The RMB 16 bn AI budget will stay dynamic with business progress, and we will allocate more inference capacity if model iteration gains user approval; the budget may rise, subject to clear ROI.

**Q: Q1 EV deliveries fell from the 2025 peak, making the 550k full-year target look tight. What is the new model roadmap, and can you quantify one-offs like purchase tax and cross-year VAT subsidy effects on margins?**

A: In Jan–Feb, we only delivered YU7, while we paused SU7 sales for two months to clear the prior-gen and make room for the new SU7, a fast-iteration industry practice that clarifies our intent to customers. Feedback has been very positive, but this limited Q1 deliveries to about 80k.

Deliveries rebounded significantly in Apr, next-gen SU7 has short lead times, and actual monthly deliveries topped 30k. We will launch a larger, all-new platform model in 2H with strong competitiveness, and **we are confident in achieving the 550k full-year target**.

**On margins, the largest hit came from the purchase tax subsidy fade, plus VAT subsidy issues on prior-year orders not yet delivered, with a per-car impact of RMB 10k–15k, dragging ASP and margin**. **Second, demo car sales began this quarter, also pressuring ASP; third, battery, memory and commodity inflation is an industry-wide issue**, though our higher ASP mitigates margin impact, keeping our GPM relatively high vs. peers. Lower Q1 deliveries also hurt fixed-cost absorption. As next-gen SU7 and both YU7 Standard and GT ramp, sequential volume and ASP should rise, with supply-chain cost-downs providing further offset.

**Q: Under memory cost pressure this year, how will smartphones balance volume, price and margin?**

A: The super-long memory upcycle is tracking our prior view: long in duration, large in magnitude, and Q3 contract prices are still rising. **We expect the pace to moderate in Q3, but the absolute base is high, so pressure remains significant**, affecting not just smartphones but the broader CE industry.

In Q1 we did several things: **first, we dynamically balanced price, cost, volume and margin**, as blunt price hikes or cuts are unwise in the short term. **Second, while volume fell, ASP rose ~10% and GPM expanded 200bps**, validating our approach. Third, we launched high-end MAX and others, and we will not simply pass costs linearly to consumers; we must price to real demand. Fourth, domestically we held up well, with VCA-activated share around 14%, because we resisted early price hikes and aimed to be the last to raise, only making small moves in mid-Apr, and users understood. This framework prepares us for the whole super-cycle while capturing the AI phone opportunity.

**Q: Please update on EV overseas rollout timing and sales network expansion.**

A: Our overall strategy is largely unchanged from last quarter. **We plan to start opening overseas stores in Q3–Q4 2027, entering developed markets first, then emerging, starting from high-end and expanding down to upper mid-end**.

The team is fully engaged in go-global prep. Overseas operations are complex, involving legal/compliance, product localization, and network and call-center planning, so the prep workload is heavy.

**Q: MiMo now deploys over 1 tn tokens per day. What are conversion and retention under the Token Plan, and do you have internal commercialization KPIs?**

A: After launch, we observed several patterns: **first, paid tokens account for over 30%; second, the Max tier accounts for over 50% of deployed tokens; third, overseas tokens exceed 50%**. We have not disclosed retention, but it is high.

We are still in the model-iteration phase without a closed commercial loop, and many new models are hitting the market. We aim to keep accumulating data and iterating during token use, and the Token Plan will adjust with market dynamics. We are in data collection and model iteration, far from treating the Token Plan as a major standalone business for now.

**Q: Overseas AIoT units and mix are rising quickly. How do you see growth over the next three quarters, especially year-end, and any new guidance?**

A: Over the next few years, IoT will be our key lever to offset memory inflation in smartphones and must maintain fast growth. In China, we are sticking to premiumization, and recent products met internal targets, with earbuds sales 2x target and clear gains in product strength and market acceptance.

Internationally, the runway is huge but our share is still low, and we must keep developing for overseas needs. Many barriers are not easy to overcome, and I have personally spent a lot of time on this over the past two years. Third, channels: we have been investing in e-commerce since last year; overall, overseas IoT still faces headwinds, but I believe growth can persist for a long cycle, with 3–4x room in overseas AIoT.

**Q: With new LLMs every 2–3 months, will MiMo follow the market cadence? As operators enter and start price wars, how are you positioned, and how will AI-driven efficiency show up in OpEx?**

A: We do not target a 'new version every three months' cadence and will not release for its own sake. MiMo must serve and be coupled with our businesses, which are grounded in real scenarios and data, and without a business loop it is hard to assess model value.

So we will keep strengthening the model while also selling tokens to the market, a dual approach that differentiates us. We are not in the business of selling tokens per se, so we will not release just to release. **Internally, we encourage employees to use MiMo more to lift productivity and generate data to feed model iteration, advancing both simultaneously**.

**Q: The EV, AI and innovation segment posted a RMB 3.1 bn OP loss in Q1. As quarterly EV volume breaks into six digits and AI spend rises, how should we think about full-year profitability for the segment?**

A: The biggest difference vs. prior quarters is the drop in deliveries, as we did not deliver SU7 in Q1 and focused on YU7 to make room for the new SU7, directly weighing on revenue. **From Q2, EV deliveries will definitely rise QoQ; second, AI investment will also increase**.

The segment's operating outcome will be driven by three variables: **volume, margins and AI spend**, and we will manage the balance among them.

**Q: Deliveries peaked in Q2 2025 then declined. What happened, and how will you restore growth momentum?**

A: Initially, we expected higher sales but lead times were long, and some users were poached by competitors during the wait. From the lineup perspective, we benchmark Model Y, with 600 km range fully sufficient for city commuting, plus an 800 km variant as a supplement.

We have added another model to the mix, now YU7 Standard and YU7 GT. Feedback since last weekend's launch has been very positive, and we will progressively converge on our expected volumes.

**Q: Q1 margins held up QoQ. Given higher memory costs, how will smartphone and CE margins trend?**

A: Upward cost pressure will persist, and we must balance scale, profit and margins. More importantly, **when costs rise, we must ask whether consumers will still pay more, and blunt pass-through is not viable**.

We need to rebase products on the new cost structure so users still see fair value, while staying true to our value proposition, and proactively adjust product mix and quality. Scale will likely decline, but higher ASP can partially offset volume, and further GP margin work can keep gross profit manageable. This is our overall approach for smartphones.

<End\>

**Risk disclosures and statements:**[**Dolphin Research Disclaimer & 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)

## Comments (12)

- **一飞冲天8888 · 2026-05-27T06:06:39.000Z**: How to short Xiaomi
- **JS_____ · 2026-05-27T02:22:17.000Z**: Well done
- **jamike · 2026-05-27T01:10:40.000Z**: Whether the key point is real or just boasting, I don't know
- **Neozzz · 2026-05-27T00:25:40.000Z · 👍 2**: After watching the interview with Xiaomi's AI head, I added a short position.😉
- **长驰 · 2026-05-26T23:54:55.000Z**: Shareholder return is only 84... Issued an additional 40 billion at a high price.And there's AI innovation. I highly doubt it's money spent on purchasing equipment, and with an investment of 8.9 billion, do you think it can support a market cap of 100 billion? Can it iterate quickly and reduce costs
  - **惘闻** (2026-05-27T00:26:59.000Z): You're shorting, why are you talking so much? This is the stock market!
  - **长驰** (2026-05-27T00:48:17.000Z): I'm just bearish, this is the stock market, not your home
  - **惘闻** (2026-05-27T00:54:06.000Z): Anyone can talk big, haha...
- **爱存钱的小海豹 · 2026-05-26T15:57:12.000Z**: 🙂‍↕️If there's only one possibility for the situation to reverse tomorrow, it's that I'm gone. If it continues to be sluggish tomorrow, the only possibility is that you're talking nonsense again.😤
- **FarmerD · 2026-05-26T15:17:06.000Z**: So good! A 20-centimeter gain at the bottom tomorrow.