Sensetime Group, Inc. Class B · HKEX:20

Sensetime Group, Inc. Class B — FY2025 Earnings Presentation

31 slides from Sensetime Group, Inc. Class B's earnings presentation, March 2026

  1. Slide 1: 2025 Annual Results Presentation

    S e n s e Ti m e G r o u p I n c . ​ 2025 Annual Results Presentation TO CREATE A BETTER AI-EMPOWERED FUTURE THROUGH INNOVATION 2026.03

  2. Slide 2: Disclaimer

    Disclaimer​ The information in this presentation has been prepared by representatives of SenseTime Group Inc. (the "Company", and together with its subsidiaries, the "Group") for use in presentations by the Group for information purposes. No part of this presentation should form the basis of, or be

  3. Slide 3: FY2025: Revenue Hits Record High with Clear Trend of Profitability Improvement

    FY2025: Revenue Hits Record High with a Clear Trend of Profitability Improvement RMB 5,015 Million  58.6 %  85.0% RMB 4,870 Million Group Revenue Losses YoY Narrowed EBITDA Collection Record High Clear Trend of YoY Narrowed Record High Loss Reduction Adjusted EBITDA Improves for Four Consecutive P

  4. Slide 4: Generative AI Drives Growth & Business Structure Optimized Steadily

    Generative AI Drives Growth & Business Structure Optimized Steadily BUSINESS SEGMENT PERFORMANCE 5,015m RMB 302m -5.9% 3,772m 1,083m Generative AI 321m +3.4% Computer Vision 1,047m 3,630m X Businesses +51.0% 2,404m 2024 2025 4

  5. Slide 5: Generative AI - "Three-in-One" Strategy

    Generative AI - "Three-in-One" Strategy Large Models SenseNova Multimodal Large Model Series Pushing the boundaries of original innovation AI Applications AI Infrastructure Cultivating high-value scenarios Accelerating full localization Securing vertical applications Deepening scenario-based deploym

  6. Slide 6: Deep integration of language and vision serves as a pathway to pushing intelligence

    Deep integration of language and vision serves as a highly efficient pathway to pushing the boundaries of intelligence 6

  7. Slide 7: Large Models - Language-vision Integration to Push AI Boundaries

    Large Models Large Models - Language-vision Integration to Push AI boundaries Applications AI Infra Digital-physical fusion & unified interaction Full-chain processing, planning and execution Native fusion of text, image and video Massive data aggregation & expanded context * Created by SenseNova-U1

  8. Slide 8: Large Models - Language-vision Integration to Push AI Boundaries

    Large Models Large Models - Language-vision Integration to Push AI boundaries Applications AI Infra Multimodal Agent: Significant improvement in task completion quality and optimized token consumption Task Completion Success Rate(%) Token Consumption(%) 100.0 94.3 54.9 48.6 Language Model Multimodal

  9. Slide 9: Large Models - Language-vision Integration to Push AI Boundaries

    Large Models Large Models - Language-vision Integration to push AI boundaries Applications AI Infra 2024.04 2024.07 2024.12 2025.04 2025.12 2026.Q2 5.0 5.5 5.5+ 6.0 6.5+ First commercial First to achieve Deep integration of Integration of multimodal Interleaved text-image + New version to come, mult

  10. Slide 10: Models - SenseNova-SI Spatial Intelligence Model (Nov 2025)

    Large Models Large Models - SenseNova-SI Spatial Intelligence Model (Nov 2025) Applications AI Infra Spatial Intelligence Model SenseNova - SI 1.3 Ranked No.1 in Comprehensive Evaluation Across 8 Major Spatial HuggingFace Downloads Intelligence Leaderboards* 3x the Downloads of Benchmark Model Sense

  11. Slide 11: Models - World Model: Kairos - SenseNova (Dec 2025)

    Large Models Large Models – World Model: Kairos – SenseNova (Dec 2025) Applications AI Infra The industry's first open-sourced embodied world model integrating "multimodal understanding, generation, and prediction” No. 1 in Comprehensive World Model Capabilities Generated with World Model Kairos 3.0

  12. Slide 12: Models - SenseNova NEO Architecture (Feb 2026)

    Large Models Large Models – SenseNova NEO Architecture (Feb 2026) Applications AI Infra NEO-unify: Building End to End Native Multimodal Unified Models Underlying reconstruction Three core innovations Matching flagship evolved beyond the allowing seamless performance with 1/10 of mainstream spliced

  13. Slide 13: AI Applications - Industry Deep Dive, Vertical Breakthrough

    Large Models AI Applications - Industry Deep Dive, Vertical Breakthrough Applications AI Infra From the perspective of scenario needs, the order of AI application implementation Complex Quadrant II Quadrant I Downstream user demands AI+ Video AI Game AI+ High level of professionalism, involve multim

  14. Slide 14: AI Applications - From Technological Leadership to Value Realization

    Large Models AI Applications - From Technological Leadership to Value Realization Applications AI Infra Office Video AI+ AI+ Software Generation Marketing Education TOP10 300,000+ 20x 500+ Frost & Sullivan creators improvement in academic 2025 Most Practical Agents* store operation efficiency instit

  15. Slide 15: AI Applications - Next-generation AI-native Consumer Applications

    Large Models AI Applications - Next-generation AI-native Consumer Applications Applications AI Infra 15

  16. Slide 16: AI Applications - Next-generation AI-native Consumer Applications

    Large Models AI Applications - Next-generation AI-native Consumer Applications Applications AI Infra Kapi Camera App Store Charts in multiple countries No.1 Kapi Budgeting Top 10 Most Popular AI Agents* * Frost & Sullivan: 2025 China AI Agent (Intelligent Agent) Best Practice Application List (Septe

  17. Slide 17: SenseCore - Co-optimizing Computing Power and Algorithms

    Large Models SenseCore - Co-optimizing Computing Power and Algorithms Applications AI Infra Dual-driven growth Transitioning domestic Compute-energy through efficiency and compute from “usable” synergy redefining AI compute optimization to “user friendly” growth frontiers LightX2V Compute-energy Lar

  18. Slide 18: SenseCore - Full-stack Domestic Ecosystem and Global Deployment

    Large Models SenseCore - Full-stack Domestic Ecosystem and Global Deployment Applications AI Infra First Launch in the Middle East domestic computing cluster Total operational computing power 40,400 PFLOPS+ Industry’s first integration Partnered with Huawei 910C liquid-cooled cluster 10+ domestic GP

  19. Slide 19: SenseCore - Leading in Both Market Scale and Technical Standards

    Large Models SenseCore - Leading in Both Market Scale and Technical Standards Applications AI Infra Ranked first among native AI cloud providers1 One of the first in China Computing platform service IDC large model inference core dimensions Awarded the “Excellent Level” certification in capability c

  20. Slide 20: Generative AI - Empowering Diverse Industries to Achieve Commercial Objectives

    Generative AI - Empowering Diverse Industries to Achieve Commercial Objectives in Scale ··· 20

  21. Slide 21: Computer Vision - Breakthrough in Profitability

    Computer Vision - Breakthrough in Profitability Period of Large-Scale Commercial Returns Highest Standard in the Industry Profitability Milestone Stable Cash Flow Achieved CV2.0 positive cash flow CV2.0 achieving net profit * for two consecutive years * 1 Industry Leader Ranked No.1 for nine consecu

  22. Slide 22: Computer Vision - The Global Momentum of Chinese Innovation

    Computer Vision - The Global Momentum of Chinese Innovation Kuala Lumpur Miral Group, UAE Krungthai Bank, Thailand International Airport Qiddiya, Saudi Arabia AI culture and tourism AI financial solutions AI video management platform AI CV large model comprehensive solutions customer analysis platfo

  23. Slide 23: X Businesses - Unlocking Ecosystem Value and Operational Synergy

    X Businesses - Unlocking Ecosystem Value and Operational Synergy …… Industry’s first end-to-end spatial intelligence Industry-leading product matrix First mass-produced home robot with “Ambient data — World model — Embodied Global benchmark for AI healthcare OpenClaw integration interaction” 23

  24. Slide 24: Financial Review

    Financial Review -

  25. Slide 25: FY2025 Financial Overview

    FY2025 Financial Overview Revenue Gross Profit EBITDA Net Loss (RMB million) (RMB million) (RMB million) (RMB million) 5,015 (471) 3,772 (1,782) 3,406 2,056 (3,143) 1,501 1,620 (4,307) (5,450) (6,495) 2023 2024 2025 2023 2024 2025 2023 2024 2025 2023 2024 2025 (10.6%) 10.8% 32.9% 44.1% 42.9% 41.0% (

  26. Slide 26: Operating Expenses

    Operating Expenses R&D Expenses G&A Expenses Selling Expenses Operating Expenses (RMB million) (RMB million) (RMB million) (RMB million) Opex (excl. SBC) Opex (excl. SBC) Opex (excl. SBC) Opex (excl. SBC) SBC SBC SBC SBC 6,250 5,795 178 5,570 4,132 324 81 292 3,775 3,466 143 150 1,511 6,072 1,464 5,

  27. Slide 27: Working Capital

    Working Capital Turnover Days – Period Ending Method(1) Net Impairment Losses on financial assets and contract assets(2) (RMB million) 1,835 Cash Conversion Cycle 781 334 228 129 287 2023 2024 2025 399 % of revenue 53.9% 20.7% 5.7% Collection of trade receivables 231 (RMB million) 4,871 4,623 183 16

  28. Slide 28: Capex and Cash Balance

    Capex and Cash Balance Capex(1) Total Cash Reserves(2) Equity and Bond Investment(3) (RMB million) (RMB million) (RMB million) Equity Investment 14,247 Bond Investment 44.6% 29.9% 69.6% 12,752 11,409 7,510 6,527 6,363 1,330 1,636 1,271 3,488 6,180 1,518 4,891 5,093 1,129 2023 2024 2025 2023 2024 202

  29. Slide 29: Strategic Outlook

    Strategic Outlook Continued commitment Deepening Industrial Efficient optimization of to the path of Intelligence unit intelligence costs native multimodal SenseNova redefining the Securing entry points in the Full promotion of domestic AI Paradigm Agent-native era substitution Consolidating “profit

  30. Slide 30: Q&A

    Q&A -

  31. Slide 31: THANKS!

    THANKS! TO CREATE A BETTER AI-EMPOWERED FUTURE THROUGH INNOVATION​ 2026.03

SenseTime Group — FY2025 Earnings Presentation on 6ix