SenseTime
SenseTime is a Hong Kong-listed AI platform company that evolved from facial recognition and surveillance infrastructure into a diversified AI and generative AI cloud services provider.
Publisher review
SenseTime is a Hong Kong-listed AI platform company that evolved from facial recognition and surveillance infrastructure into a diversified AI and generative AI cloud services provider. Founded in 2014 by researchers from the Chinese University of Hong Kong, the company pivoted strategically in 2024-2025 to position generative AI as its primary growth engine, driven by geopolitical constraints and commercial opportunity. The company operates through two main engines: SenseNova (a multimodal foundation model platform with cost-optimized variants including the open-source U1 model) and SenseCore (proprietary AI infrastructure built on domestic Chinese chips like Huawei's Ascend).
SenseTime targets Chinese enterprises and government agencies across multiple verticals: SenseFoundry Enterprise handles digital transformation for mid-market firms; SenseAuto powers autonomous vehicle systems and cabin monitoring; SenseFoundry Smart City serves municipal governance. Revenue in 2025 reached 5 billion RMB, with H1 2025 generative AI sales surging 72.7 percent year-over-year to 252 million USD. The company achieved profitability on an EBITDA basis in the second half of 2025, its first positive result since listing in December 2021.
However, SenseTime operates under significant constraints: U.S. sanctions since 2019 restrict NVIDIA GPU access (addressing allegations of surveillance technology use in Xinjiang), forcing the company to engineer around chip embargoes using domestic alternatives. The Pentagon designated it a "Chinese military company" in December 2024. For enterprise buyers outside China, integration depends on cloud availability and API pricing structures that remain partially opaque in public documentation. The company's trajectory reflects both serious technical capability and the reality of Chinese AI development in a multipolar infrastructure landscape.
How it works
-
SenseNova Multimodal Models
Foundation models (including open-source SenseNova U1 in 2B/8B variants and proprietary v6.5) optimized for cost efficiency, marketed as lower-cost alternatives to leading Western models for text, image, and vision-language tasks.
-
SenseCore AI Infrastructure & Cloud APIs
Proprietary cloud platform integrating compute, algorithms, and deployment services; powers SenseNova inference; delivers API access for enterprises within China and select markets.
-
SenseFoundry Enterprise Platform
B2B software suite for enterprise digital transformation, including IoT devices, authentication services, quality control, and infrastructure monitoring applications for commercial and industrial use.
-
SenseFoundry Smart City
Municipal software platform for traffic management, environmental monitoring, emergency response, and public safety, targeting government agencies in Chinese municipalities.
-
SenseAuto Autonomous Driving Stack
Integrated system for autonomous vehicles covering cabin monitoring, advanced driver assistance (ADAS), vehicle-to-everything (V2X) communication, and robotics (RoboX platform).
-
SenseCare Medical Imaging AI
Specialized models for diagnostic imaging analysis covering lung CT, cardiac, and liver analysis; serves hospitals and healthcare institutions in China.
-
SenseEarth Remote Sensing Intelligence
Computer vision application for satellite and aerial imagery analysis, addressing environmental monitoring and geospatial intelligence use cases.
Strengths and trade-offs
Strengths
- Rapid pivot to generative AI revenue (72.7% surge in H1 2025) demonstrates agility and market fit in domestic Chinese enterprise AI market.
- Achieved EBITDA profitability in H2 2025 despite U.S. sanctions and chip embargoes, signaling sustainable unit economics.
- Vertically integrated from infrastructure (domestic chip partnerships) through cloud platform to models and industry-specific applications, reducing external dependencies.
Trade-offs
- Sanctioned by U.S. government since 2019 and designated a "Chinese military company" by Pentagon in December 2024, creating regulatory barriers and limiting international expansion and customer base.
- Heavy reliance on domestic Chinese chip alternatives (Huawei Ascend, Cambricon) due to NVIDIA export restrictions; ecosystem lock-in risk and long-term cost/performance uncertainty versus Western GPU supply chains.
- Pricing, API documentation, and cloud service availability remain largely opaque in public sources; commercial terms and integration pathways for non-China customers unclear compared to competitors.
Pricing context
SenseTime does not publicly disclose detailed pricing tiers or per-token/per-API-call rates for SenseNova models or SenseCore cloud services in English-language documentation. The company operates as a B2B vendor selling enterprise platform software and cloud services to Chinese corporations and government agencies on commercial contract terms. H1 2025 generative AI revenue reached 252 million USD, suggesting a mix of high-value contracts and growing API/platform adoption, but specific SKU pricing, freemium tiers, or usage-based pricing are not available to international buyers or documented publicly. Interested customers must contact SenseTime directly for pricing discussions.
Getting started with SenseTime
-
Register for SenseTime cloud
Visit the SenseTime official website and create an enterprise account. Provide your business email, company details, and intended use case. SenseTime operates on a B2B contract model, so expect a sales representative to contact you for verification and onboarding.
-
Request API credentials
Contact SenseTime's sales team through the provided channels to request access to SenseNova or SenseCore APIs. You will need to sign a commercial agreement and receive API keys, endpoint URLs, and documentation specific to your region and use case.
-
Configure model deployment
Select the appropriate SenseNova model variant (e.g., U1 open-source or proprietary v6.5) for your task. Set up the inference environment using the provided API keys and configure parameters such as input/output format, batch size, and cost-optimization settings.
-
Run a test inference
Send a sample request to the SenseNova API endpoint using a tool like cURL or Python. Verify the response includes the expected output for your text, image, or vision-language task. Check latency and accuracy against your requirements.
-
Integrate into production
Embed the SenseNova API calls into your application code, handling authentication and error retries. Monitor usage and costs through the SenseCore dashboard. Schedule regular reviews with your SenseTime account manager to adjust contract terms as needed.
Frequently Asked Questions
What is SenseTime and what does it do now?
SenseTime is a Hong Kong-listed AI platform company that originally focused on facial recognition and surveillance. It has pivoted to generative AI and cloud services, offering multimodal models through SenseNova and AI infrastructure via SenseCore for Chinese enterprises and government agencies.
How did SenseTime pivot from facial recognition to generative AI?
In 2024-2025, SenseTime strategically repositioned generative AI as its primary growth engine, driven by geopolitical constraints and commercial opportunity. The company now operates SenseNova, a multimodal foundation model platform, and SenseCore, its proprietary AI infrastructure built on domestic Chinese chips.
What are SenseTime's main products and platforms?
SenseTime's key products include SenseNova for multimodal AI models, SenseCore for cloud infrastructure, SenseFoundry Enterprise for digital transformation, SenseAuto for autonomous driving, SenseFoundry Smart City for municipal governance, SenseCare for medical imaging, and SenseEarth for remote sensing.
How have U.S. sanctions affected SenseTime's business?
U.S. sanctions since 2019 restrict SenseTime's access to NVIDIA GPUs, forcing reliance on domestic alternatives like Huawei's Ascend chips. The Pentagon designated it a Chinese military company in December 2024, creating regulatory barriers and limiting international expansion and customer base.
What is SenseTime's financial performance and profitability?
SenseTime's 2025 revenue reached 5 billion RMB, with H1 2025 generative AI sales surging 72.7% year-over-year to 252 million USD. The company achieved EBITDA profitability in the second half of 2025, its first positive result since listing in December 2021.
How can I get pricing for SenseTime's AI services?
SenseTime does not publicly disclose detailed pricing for SenseNova models or SenseCore cloud services in English. The company operates as a B2B vendor selling enterprise platform software and cloud services on commercial contract terms. Interested customers must contact SenseTime directly for pricing discussions.
Alternatives
How SenseTime compares
Direct head-to-head against 3 competitors. Picked by 7wData.
SenseTime
- Pricing
- SenseTime does not publicly disclose detailed pricing tiers or per-token/per-API-call rates for SenseNova models or SenseCore cloud services in English-language documentation. The company operates as a B2B vendor selling enterprise platform software and cloud services to Chinese corporations and government agencies on commercial contract terms. H1 2025 generative AI revenue reached 252 million USD, suggesting a mix of high-value contracts and growing API/platform adoption, but specific SKU pricing, freemium tiers, or usage-based pricing are not available to international buyers or documented publicly. Interested customers must contact SenseTime directly for pricing discussions.
- Target
- SenseTime is a Hong Kong-listed AI platform company that evolved from facial recognition and surveillance infrastructure into a diversified AI and generative AI cloud services
- Strength
- Rapid pivot to generative AI revenue (72.7% surge in H1 2025) demonstrates agility and market fit in domestic Chinese enterprise AI market.
- Watch for
- Sanctioned by U.S. government since 2019 and designated a "Chinese military company" by Pentagon in December 2024, creating regulatory barriers and limiting international expansion and customer base.
Baidu AI Cloud
- Pricing
- Custom pricing based on compute and storage usage; no public per-user tiers.
- Target
- Chinese enterprises needing integrated AI and cloud infrastructure for smart city, auto, and finance.
- Deployment
- Public cloud, hybrid, on-premises.
- Strength
- Largest China LLM market share at 19.9% with 150,000 Qianfan clients.
- Watch for
- Tight Chinese government ties may limit international expansion and raise compliance risks.
Megvii
- Pricing
- Custom pricing per project; no public standard tiers.
- Target
- Chinese government and enterprise clients for facial recognition and smart city solutions.
- Deployment
- On-premises, edge, cloud.
- Strength
- Strong facial recognition IP with deep integration into Chinese public security systems.
- Watch for
- US blacklist since 2019 restricts access to American technology and markets.
IBM Watsonx
- Pricing
- Custom pricing; watsonx.ai starts at $140/user/month for enterprise tier.
- Target
- Global enterprises requiring hybrid cloud AI with strong data governance and agentic workflows.
- Deployment
- Public cloud, hybrid, on-premises.
- Strength
- Enterprise-grade data governance and agent orchestration via watsonx.orchestrate.
- Watch for
- Complex pricing and integration; smaller market share (0.04%) vs. hyperscaler AI platforms.
User reviews
No user reviews yet. Be the first to write one.
Sources
Reporting on this tool draws on these publicly available sources.
- www.sensetime.com — 2025 revenue (5+ billion RMB), H1 2025 generative AI revenue (252 million USD, 72.7% surge), EBITDA profitability in H2 2025
- en.wikipedia.org — Founding year (October 2014), founders Tang Xiao'ou and Xu Li, Hong Kong headquarters, U.S. sanctions history (2019 onward), Pentagon military designation (December 2024), GPU infrastructure scale, facial recognition history
- thebambooworks.com — Strategic pivot to generative AI as primary growth driver, '1+X' organizational structure, SenseCore AI cloud certification (February 2026)
- www.cnbc.com — Cost-efficiency positioning in Chinese AI market, competitive landscape in global AI race
- www.biometricupdate.com — U.S. sanctions history, reliance on Huawei Ascend chips as NVIDIA alternative due to export restrictions
- www.washingtonpost.com — U.S. government investment ban (2021) over allegations of Uyghur surveillance, government position on SenseTime
- idtechwire.com — Pivot from facial recognition / surveillance to generative AI, Smart City product history