China enterprise GenAI deployment is moving toward scale, with cloud and model suppliers the clearest beneficiaries

Overview

This report is a China AI thematic research note published by UBS Global Research, based on the first China AI Business Survey from UBS Evidence Lab. The sample consists of 102 Chinese IT executives and data engineers across multiple industries, surveyed in December 2025. The report's core conclusion is that Chinese enterprise GenAI adoption has moved beyond the pilot stage, with production deployment and scaling now underway. Management commitment and budget willingness are strong, and future demand is expected to continue expanding toward enterprise productivity, AI agents, cloud infrastructure, and the model layer.

Core views

The report argues that China enterprise AI adoption has four main themes: first, the GenAI deployment rate has reached 78%, making the scaling trend clear; second, AI agents are the next adoption focus and are expected to take longer to scale than ordinary GenAI, although open-source agent frameworks such as OpenClaw may lower the deployment barrier; third, AI budget growth is significantly faster than overall IT budget growth, indicating strong enterprise commitment and showing that budget is not the main bottleneck; and fourth, cloud and the model layer are the most direct beneficiaries, with Alibaba, Tencent, Kingsoft Cloud, and MiniMax highlighted, while the SaaS segment remains weighed down by AI disruption risk.

Analysis framework

The report mainly uses enterprise surveys, cross-market comparisons, and value-chain mapping. UBS combines the China AI Business Survey with U.S. AI business surveys, consumer AI surveys, and QuestMobile active-user data to compare China and the United States in use-case preferences, budget growth, ROI concerns, localization progress, cloud-provider selection, and model usage preferences, then maps the findings to covered companies and sector investment implications.

Methodology notes

• survey_researchUBS Evidence Lab China AI Business Survey

  enterprise AI adoption survey

  The survey was conducted in December 2025 with 102 Chinese respondents across multiple industries, skewing toward medium and large enterprises and mainly senior technical decision makers. It was used to assess trends in GenAI, AI agents, budgets, localization, and provider selection.

• cross_market_comparisonChina vs US AI Business Survey Comparison

  comparison of China and U.S. enterprise AI adoption

  The report compares differences in enterprise AI use cases and challenges between China and the United States: China leans more toward customer experience and visual analytics customer-facing scenarios, while the U.S. places greater emphasis on copilots, productivity gains, and developer tools.

• valuation_methodSOTP / Price-to-ARR / EV-to-sales

  covered-company valuation methods

  The disclosure section notes that Tencent and Alibaba use SOTP valuation, MiniMax uses Price/ARR, and Kingsoft Cloud uses EV/sales multiples, but the main body of the report does not provide a unified target price or a single rating change.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Alibaba

  Beneficiary

  Strengths

  A leading domestic cloud provider with full-stack AI capabilities; Qwen ranks among the top three enterprise models by usage.

  Weaknesses

  Disclosed risks include regulatory changes, macro pressure, offline retail competition, system outages, short-term profitability pressure from long-term investment, platform complexity, and corporate governance.

  Comparison

  Leads adoption in the China cloud market; Qwen enterprise usage is 28%, below Tencent Hunyuan and DeepSeek but still in the leading tier.

  Risks

  Data and content regulation, macroeconomics, competitive pressure, IT system stability, long-term investment execution, and profitability pressure.

• Tencent

  Beneficiary

  Strengths

  Tencent Cloud leads enterprise adoption, and Tencent Hunyuan is the most-used model among surveyed enterprises, with both cloud and model offerings well recognized.

  Weaknesses

  Execution in new businesses, integration of invested companies, and traffic acquisition and content promotion costs may affect returns.

  Comparison

  Tencent Cloud adoption is more balanced across company sizes, with penetration of about 30% among large enterprises; Hunyuan usage is 37%, slightly above DeepSeek at 36%.

  Risks

  Rising competition, new business execution, investment integration, traffic and brand promotion costs, IT system maintenance, international expansion, intellectual property, management, and regulatory risks.

• Kingsoft Cloud

  Beneficiary

  Strengths

  AI workloads account for more than 40% of total revenue, giving it higher leverage to enterprise AI demand growth.

  Weaknesses

  Relative revenue growth versus the industry and market share pressure, with uncertainty around the profitability inflection point.

  Comparison

  Compared with large diversified cloud providers, Kingsoft Cloud is a more leveraged exposure to high AI workload adoption.

  Risks

  Revenue growth below the industry, market share decline, failure to reach net income breakeven, weaker cloud demand, support from Xiaomi and Kingsoft Group falling short of expectations, and GenAI development and enterprise adoption slowing more than expected.

• MiniMax

  Beneficiary

  Strengths

  Growing popularity among domestic and international enterprises and developers, benefiting from the commercialization trend in AI models and applications.

  Weaknesses

  Commercialization and margins still need to be proven, and competition and execution challenges could weigh on monetization.

  Comparison

  Compared with models from internet giants, MiniMax better reflects the growth optionality of an independent model company.

  Risks

  Macro and geopolitical uncertainty, limited AI adoption, competition and execution pressure, model training data risk, user-generated content governance, and disruption to key service providers.

• SaaS and enterprise software

  Under pressure

  Strengths

  Deeper enterprise productivity use cases may create long-term opportunities to reshape software demand.

  Weaknesses

  AI disruption risk continues to weigh on the segment, and investor confidence recovery requires earnings resilience and valuation support.

  Comparison

  Cloud and model providers supply the infrastructure and benefit more directly; SaaS needs to prove value retention in the AI era.

  Risks

  AI may replace or reshape software functions, enterprise IT capabilities may be insufficient, ROI remains uncertain, and valuation pressure persists.

• Doubao, Qwen, Yuanbao, and other consumer AI products

  Theme-related

  Strengths

  Doubao leads the China consumer AI chatbot market, while Qwen and Yuanbao saw active-user gains after the Spring Festival promotion period.

  Weaknesses

  Competition in China consumer AI chatbots remains fragmented, and growth has slowed recently.

  Comparison

  The U.S. consumer AI market is closer to a duopoly of ChatGPT and Gemini, while China remains more fragmented.

  Risks

  Dependence on promotions, user retention, product upgrade pace, competitive consolidation, and uncertainty around new traffic entry points.

Key data

• GenAI deployment rate78%51% of respondents have deployed in at least one function, and 27% have already scaled in production across multiple business units.
• Expected time for PoC-stage or earlier enterprises to reach scale9.3 monthsApplies to GenAI deployments that have not yet scaled.
• Expected time for AI agents to reach scale1.7 yearsThe report believes open-source agent frameworks such as OpenClaw may accelerate adoption.
• Potential productivity gain from AI agents20%Average productivity improvement expected by respondents over the next two years.
• Average 2025 GenAI budget for Chinese enterprisesRmb66mAbout 12% of total IT budget.
• Expected growth rate of China GenAI budget in 202621.0% YoYAbove overall IT budget growth of 5.0% YoY and also above U.S. GenAI budget growth of 17.7% YoY.
• AI budget allocation23% personnel, 21% AI infrastructure, 21% AI models, 18% AI software, 17% project-relatedShows that Chinese enterprises are investing across multiple fronts in the early stage of AI adoption.
• Localization rate at the model layer82%Higher than AI infrastructure software at 60% and AI hardware at 29%.
• Expected localization in 2026AI hardware 51%, AI infrastructure software 61%, AI model layer 81%Shows that localization across the AI value chain will continue to advance.
• Cloud provider selection factorsCompute performance 60%, security and compliance about 48-52%, cost 43%For large enterprises, cost importance is only 30%, indicating that performance and reliability come first.
• Enterprise model usage ratesTencent Hunyuan 37%, DeepSeek 36%, Alibaba Qwen 28%The survey was conducted at the end of 2025, so model preferences may change rapidly as the competitive landscape evolves.
• Model selection factorsPerformance and accuracy 53%, ecosystem/technical support 43%, security/compliance 42%, cost 25%The model layer also shows a preference for performance and ecosystem over cost.

Impact & implications

In terms of investment implications, the report believes that Chinese enterprises have strong willingness to invest in AI, budget growth is fast, and localization is advancing clearly, which will drive continued AI demand growth. Cloud providers and model suppliers, as the core infrastructure providers for enterprise AI, will benefit most directly, especially Alibaba with both cloud and full-stack AI capabilities, Tencent with recognition in both cloud and models, Kingsoft Cloud with a high share of AI workload revenue, and MiniMax with rising popularity among domestic and international enterprises and developers. By contrast, software and SaaS need clearer earnings resilience and valuation support to rebuild investor confidence.

Risks

• Uncertain ROI remains the top challenge for GenAI adoption among enterprises in both China and the U.S., with 35% of Chinese respondents citing it as a major constraint.
• Chinese enterprises still face complex system integration, a shortage of qualified talent, and data restrictions, reflecting the need to improve internal IT capabilities and digital infrastructure.
• AI adoption and model preferences are changing rapidly, so the survey results from late 2025 may shift as the competitive landscape changes in 2026.
• Intensifying competition in the cloud and model markets may affect pricing, margins, and market share.
• The SaaS segment faces the risk of AI feature replacement or business-model restructuring.
• Regulation, data compliance, content governance, geopolitics, and export restrictions on advanced hardware may affect AI training, deployment, and commercialization.
• Competition in consumer AI products remains fragmented, and uncertainty persists around user growth, retention, and the formation of super-app entry points.

What to watch

• The pace at which China enterprise GenAI moves from single-function deployment to scaling across multiple business units.
• The rollout of AI agents over the next 1-2 years, and whether open-source agent frameworks such as OpenClaw lower the deployment barrier.
• Whether GenAI budgets grow as expected by 21.0% YoY in 2026 and continue to materially outpace overall IT budget growth.
• Whether localization rates for AI hardware, AI infrastructure software, and the model layer continue to rise.
• Changes in enterprise share for AI workloads among AliCloud, Tencent Cloud, Huawei Cloud, and Baidu Cloud.
• Shifts in the ranking of enterprise usage and ecosystem support for models such as Tencent Hunyuan, DeepSeek, and Alibaba Qwen.
• Whether China enterprise AI adoption expands from customer experience and visual analytics into productivity, workflow automation, and developer tools.
• Whether SaaS companies can demonstrate earnings resilience and valuation support in the AI era.
• Active-user trends and industry consolidation for consumer AI products such as Doubao, Qwen, and Yuanbao.