过去一年,中国AI软件私有化市场交出了一份不错的成绩单:计算机视觉92.5亿元,语音语义118.6亿元,机器学习平台稳定增长。但真正值得关注的,不是这些数字和名词,而是一个更根本的结论:中国私有化AI市场已经走出“技术演示”阶段,进入以场景深耕、工程化交付、多模融合为核心的深水区竞争。 在这个阶段,单纯的大模型能力或算法排名不再决定胜负,谁能把行业知识、私有化数据与可规模化的产品能力真正结合,谁才能在下半场胜出。与此同时,市场仍高度分散——多数赛道CR3低于50%——这既说明竞争激烈,也意味着格局远未锁定。

以下,我们基于IDC最新数据,分三个细分市场拆解这一趋势。

计算机视觉:92.5亿元大盘,CV2.0时代加速到来

2025年中国计算机视觉AI软件私有化市场规模达到92.5亿元。在这一成熟市场中,头部厂商凭借场景深耕能力占据领先地位。

市场格局上,商汤科技以19.5%的份额居首。作为国内计算机视觉领域的开创者,商汤凭借深厚的算法积累与超大规模训练能力,在城市安全、零售、汽车等场景保持领先。海康威视以16.7%的份额位列第二,依靠硬件与软件深度整合的产品矩阵以及遍布全国的渠道网络,在安防与工业视觉领域具备不可忽视的规模优势。创新奇智(9.3%)主攻工业制造场景的视觉质检与缺陷检测,已在汽车、电子制造等头部客户中实现批量复制;电信AI公司(7.5%)与大华股份(7.1%)则分别凭借运营商生态和安防硬件深度,在公共安全与智慧园区场景持续渗透。

值得关注的是,计算机视觉市场正经历从“CV 1.0”到“CV 2.0”的深刻变革。传统的计算机视觉以感知为核心,依赖针对特定场景训练的专用模型,一个场景一套算法,部署成本高、泛化能力有限。而随着视觉大模型的崛起,CV 2.0正在重新定义这一市场——从多模型到统一大模型解决多场景问题,从单模态感知到图文多模态理解,从闭集识别到开集推理,从单纯的“看”到“看懂、会搜、能生成”。CV 2.0呈现出几个核心特征:一是统一大模型替代多模型,大幅降低部署和运维成本;二是多模态融合,实现跨模态对齐与“万物检索”能力;三是生成式视觉,从感知延伸至创作;四是端侧与边缘智能,视觉Agent开始落地。这一轮转型将进一步拉大不同厂商之间的技术代际差距,市场格局可能在近年内出现新一轮洗牌。

语音语义AI软件:118.6亿元,大模型重塑竞争格局

大模型的引入使得语音语义AI从“听得清、听得懂”向“答得好、能办事”演进,智能体能力的增强成为厂商差异化竞争的新焦点。2025年中国语音语义AI软件私有化市场规模达118.6亿元,是三大细分领域中体量最大的赛道,也是大模型技术渗透最为深入的私有化场景。自然语言处理(NLP)、语音识别与合成,已成为政务热线、金融客服、医疗记录、企业智能办公等场景的标配基础能力。

市场格局中,科大讯飞以15.6%的份额领跑,凭借在教育、政务、医疗三大核心赛道超过二十年的深耕积累,以及星火大模型的本地化部署能力,科大讯飞在语音语义私有化市场建立起高壁垒的护城河。百度智能云(14.2%)依托文心大模型在语言理解与生成领域的领先性能,以及在政务、金融等行业的广泛布局,紧随其后。阿里云(10.4%)与腾讯云(8.9%)则以云厂商的综合生态优势在企业级NLP私有化部署中持续渗透,尤其在大型企业的混合云场景中具备一体化交付的优势。整体来看,语音语义赛道的其他厂商占比超过50.9%,市场仍处于高度分散状态,区域系统集成商与垂直行业方案商构成了市场的长尾主体。

机器学习平台软件:向大模型工程化平台演进

对于私有化部署市场而言,机器学习平台的核心价值在于帮助企业构建自主可控的AI能力。在数据安全合规要求较高的金融、政务、能源等行业,私有化机器学习平台成为企业训练行业专属模型、沉淀AI资产的关键基础设施。2025年机器学习平台软件私有化市场保持稳定增长,成为企业构建AI能力的重要基础设施。

范式以30.4%的市场份额位居第一,凭借AutoML自动化机器学习技术和在金融、零售等行业的深度积累,持续领跑市场。华为云(25.0%)依托全栈AI能力和政企客户资源,稳居第二。星环科技(2.5%)作为大数据与AI融合的代表厂商,也在积极拓展机器学习平台市场。值得注意的是,该市场“其他”厂商占比高达42.1%,显示市场仍处于相对分散状态,竞争格局尚未固化。

机器学习平台市场呈现出几个显著特征:一是从传统ML向大模型工程化平台演进,涵盖大模型微调、RAG知识库、Agent开发的全栈AI工程化平台;二是AutoML与低代码成为标配,降低AI开发门槛、提升模型生产效率成为平台竞争的关键;三是云厂商与AI厂商差异化竞争,云厂商依托基础设施和生态优势,AI厂商则凭借算法能力和行业解决方案取胜。

结语:私有化市场的下半场,拼的是“可规模化的行业深度”

综合来看,中国私有化AI市场正呈现出几个明确趋势:垂直场景成为增长核心驱动力,而非通用API调用;CV 2.0重塑视觉市场,推动从感知走向理解与生成;技术融合加速,多模态与Agent成为下一竞争高地;竞争格局持续演变,头部厂商份额仍相对分散,未来12个月有望进一步洗牌。

IDC预测,2026年中国AI软件私有化市场仍将保持强劲增长势头,尤其是随着大模型私有化成本的进一步下降与工具链的成熟,中型企业市场将成为新的增量战场。在这场持久战中,谁能把行业Know-How真正转化为可规模化交付的产品力,谁就能在私有化市场的下半场赢得先机。

本文IDC相关报告:

IDC《中国AI软件市场半年度追踪,2025H2》

如果您希望获取更详细的市场份额数据、细分场景分析,或针对贵企业的定制化竞争策略建议,欢迎与IDC团队联系。

请点击此处与我们联系。

Hangzhou has always defied expectations. Once celebrated for the serene beauty of West Lake, it reinvented itself as the birthplace of China’s digital economy through the rise of Alibaba. Now, it’s doing it again — emerging as one of China’s most serious players in artificial intelligence and embodied intelligence.

That’s why IDC Directions 2026 is coming to Hangzhou for the first time.

In 2025, the added value of Hangzhou’s core digital economy industries reached RMB 678 billion — accounting for 29.5% of the city’s GDP, with annual growth of 9.3%, according to the Hangzhou Municipal Bureau of Statistics. That momentum outpaces the broader national trend for digital sectors, underscoring the city’s leading position as a regional innovation hub. This isn’t a city on the rise. It’s a city already there.

Why Hangzhou, and Why Now

IDC Directions Beijing and Shenzhen have long been the go-to forum for China’s ICT community — drawing hundreds of industry participants, delivering senior analyst insights on AI, cloud, cybersecurity, and emerging technologies, and offering one-on-one sessions where ICT leaders dig into their real challenges and growth opportunities. The 2025 edition alone attracted nearly 300 industry leaders, digital experts, and investors when it launched in Shenzhen.

Now, that same calibre of conversation is coming to the Yangtze River Delta — and Hangzhou is the right city for it.

Three things set Hangzhou apart:

  • A proven innovation ecosystem. Hangzhou is home to a dense mix of startups, scaled enterprises, and specialized industry clusters. It’s not an emerging hub — it’s an active one, with real deal flow and real decision-makers already operating here.
  • Leadership in tomorrow’s technologies. Robotics, intelligent computing, and smart home industries are growing fast here, and Hangzhou’s engineering and R&D capabilities have already been showcased on a national stage at the CCTV Spring Festival Gala. This is a city that builds things.
  • Strong institutional backing. The Hangzhou municipal government’s Future Industry Cultivation Action Plan (2025–2026) is already in motion, and major national gatherings — including the China (Hangzhou) Embodied Intelligent Robot Industry Conference and the National Artificial Intelligence Industry Development Conference — have firmly established the city as a destination for serious tech collaboration.

What to Expect at IDC Directions 2026 Hangzhou

What we can already say: Curated event sessions will center squarely on Hangzhou’s flagship strength sectors—robotics, smart home ecosystems, and intelligent computing—while linking local innovation to the transformative trends reshaping China’s broader ICT industry agenda.

The event will feature keynote and breakout sessions delivered by IDC China’s most senior analyst team, with deep expertise spanning enterprise AI, emerging technology, digital economy, and industry transformation across core verticals.

Featured discussion themes will dive into tangible, high-impact topics: the rapid commercial adoption of AI Agents, full-stack cloud and computing platform evolution fueled by open-source models such as DeepSeek, and projected industrial AI investment in China set to hit RMB 900 billion by 2028.

Beyond insightful content, the event will gather a high-caliber audience pool including enterprise CxOs, top tech solution vendors, industry innovators, and institutional investors from Hangzhou and across the Yangtze River Delta—covering Shanghai, Nanjing, Suzhou and surrounding tier-one innovation cities. It offers a chance to network with regional decision-makers, benchmark industry best practices, and align long-term business strategies with local growth momentum as well as the strategic priorities outlined in China’s 15th Five-Year Plan.

Join Us in IDC Directions 2026 in Hangzhou

IDC Directions 2026 Hangzhou is where the region’s most important technology conversations will happen. Whether you’re sharpening your AI strategy, exploring opportunities in robotics and embodied intelligence, or building the partnerships that will define the next phase of your business — this is where you need to be. Register now!

Elly Hao - Senior Marketing Manager, Demand Generation - IDC China

Elly Hao is Senior Marketing Manager, Demand Generation, at IDC China, where she drives data-backed, integrated marketing initiatives that deliver qualified leads and support business growth. A seasoned cross-functional collaborator, she has overseen key programs including the annual FutureScape campaign. This year, Elly is leading the marketing efforts for IDC Directions Hangzhou, a key strategic growth initiative.

Telecom operators are entering a new phase of digital infrastructure and AI monetization. That was the clear signal at FutureNet World in April 2026. The conversation is no longer about whether AI belongs in telecom. That question has been answered. The real question is which operators can turn experimentation into measurable commercial value reflected in P&L statements.

What is emerging is not simply a new set of AI use cases. It is a broader redefinition of what telecom providers can offer. For enterprise customers, the product is shifting from raw connectivity to orchestrated outcomes. The winners will not necessarily be the operators with the biggest AI infrastructure footprints or the most GPU capacity. They will be the ones that can combine connectivity, cloud, edge, automation, and governance into trusted, SLA-backed services.

That is the framing from IDC’s panel on April 23, 2026, “Unlocking New Revenue Opportunities by Monetizing AI and Digital Infrastructure” at FutureNet World London. The panelists were senior telecom executives: Emilio Varas Jiménez, Customer Fulfilment Head of AI and Operations Improvement, Vodafone; Franck Morales, Senior Vice President, Marketing and Business Development, Orange Wholesale International; Natali Delić, Chief Strategy and Digital Officer, Telekom Srbija; and Martin Rueckert, Chief AI Officer, Tallence AG.

AI monetization starts at home

External AI monetization has to be built on internal proof points. Before operators can credibly take AI capabilities to enterprise customers, they need to show they can deploy them inside their own business.

This matters because enterprise buyers are increasingly skeptical of AI promises not backed by operational experience. Telcos that can demonstrate AI-driven improvements in field service operations, assurance, employee productivity, and customer experience will be in a stronger position to package that expertise into enterprise services.

For telecom executives, this means the first monetization opportunity is often not a customer-facing AI product. It is the ability to refine, prove, and operationalize AI internally and then externalize that know-how as a service.

From connectivity to business outcomes

The enterprise buying center is changing. Businesses are looking beyond bandwidth or basic cloud access. They want guaranteed performance, sovereign routing, managed AI-enabled operations, and real-time decisioning environments that align with business risk and revenue goals.

This is especially visible in financial services, manufacturing, and retail, where latency, compliance, and uptime have direct commercial consequences. In these environments, value comes from combining multiple layers: network slicing, APIs, edge compute, AI orchestration, and managed service delivery.

Telecom providers are being pulled higher up the stack. The role of the operator is evolving from connectivity provider to orchestrator of digital infrastructure outcomes. 

Orchestration is becoming the real differentiator

Enterprises do not want to decide what runs in the public cloud, what belongs at the edge, and what must stay on-premises. They want those decisions handled for them, governed, compliant, cost-predictable, and reliable.

That changes the basis of competition. Operators that can orchestrate workloads across hyperscaler, edge, and on-premise environments build a durable market position. Those that cannot risk turning AI infrastructure into a commodity layer with limited pricing power.

The long-term value in AI and digital infrastructure will accrue to operators who can integrate, operate, and govern AI-enabled services at scale.

Edge AI will favour smaller, purpose-built models

For most real-time telecom and enterprise use cases, the requirement is not maximum model size. It is deterministic performance, low latency, and auditability.

That favors smaller, domain-specific models and, in some cases, non-transformer architectures, particularly in industrial automation, remote diagnostics, and real-time network decisioning.

This has major implications for investment strategy. Operators that assume the future of edge AI depends on pushing large language models closer to the endpoint may be overestimating both enterprise demand and the technical fit. In many scenarios, the commercial opportunity will come from deploying the right model, not the largest one.

Agentic AI remains one of the most talked-about areas in the market, but many enterprise pilots are still failing to reach production. [Source: attribute to panelist name or add IDC data reference.] The problem is not only technical capability. It is trust.

For agentic AI to be backed with an SLA, enterprises need confidence that decisions are bounded, explainable, and auditable. In regulated and mission-critical environments, free-form reasoning is not enough. Determinism matters. Governance matters. Standards alignment matters.

Telecom providers looking to monetize agentic AI should focus on domain-constrained deployment models. The path to commercial scale is likely to come from tightly scoped, standards-aligned agents that can operate within controlled decision environments.

Sovereign AI infrastructure brings opportunity and risk

Sovereign AI demand is real, particularly in regulated sectors, but that does not mean every operator should rush to build large-scale local AI factories.

There is significant capex risk in overbuilding. If utilization remains low or infrastructure cycles shorten faster than expected, operators could face stranded assets within three to five years. [Per panel discussion, April 23, 2026. Add IDC data reference if available.]

The more sustainable approach is hybrid and multi-cloud by design: combining hyperscaler or neo cloud computing, edge resources, and targeted sovereign deployments where regulation or national security requirements justify them. The key is to align infrastructure investment with verified demand rather than hype-driven positioning.

The bottom line

Telecom AI monetization between 2026 and 2028 will be defined less by model ownership and more by execution discipline. Operators that can prove value internally, orchestrate hybrid environments effectively, deploy trusted and auditable AI, and match infrastructure investment to real demand will be best positioned to capture new revenue and higher margins.

The market is past experimentation. The next phase belongs to operators that can industrialize AI as a commercial capability.

As enterprises move from AI pilots to full deployment, a harder problem is coming into focus. The bottleneck isn’t compute or tooling. It’s judgment. Without it, AI doesn’t make organizations smarter. It just makes them faster at making bad decisions.

Most workforce strategies haven’t caught up. When companies talk about the human skills employees need to work alongside AI, the conversation tends to collapse into the same shortlist: critical thinking, creativity, collaboration, growth mindset and so on. These are useful concepts, but they’re too vague to act on. And they are far removed from how human-AI work actually plays out.

Imagine an agent that has already triaged three incidents by the time a team lead logs in. What does critical thinking look like at that moment? What does creativity mean when a workflow designer must rearchitect a customer support process that can escalate, hand off, and learn from every interaction? Terms like “creativity” and “critical thinking” aren’t useful because they just aren’t actionable. Enterprises can’t assess them, teach them, or track whether anyone is getting better at them.

Breaking down the vague

The IDC Human Skills Framework for Agentic AI identifies eight clusters of human capability that organizations need to build now. Each cluster unpacks into specific, trainable subskills, the kind IT leaders can assess, teach and track.

Take critical thinking. The framework breaks it down into problem framing, assumption spotting, hallucination detection, trade-off analysis and metacognition with AI, which is the habit of asking whether AI is quietly shaping your conclusions in ways you haven’t noticed. These are more than just soft skills, or human skills. They are operational disciplines.

The same logic applies to decision-making with AI. Leaders need to be able to map accountability for AI-driven choices, recognize when outputs introduce disparate impact, and make privacy-by-design decisions before agents touch sensitive data. The framework calls these out as distinct, learnable behaviors. They matter because humans must remain in the lead. But if they lack the analytical judgement to know when to accept AI outputs or push back on them, AI becomes just a mechanism for making bad decisions faster.

Human judgement is a guardrail

Most organizations think of AI guardrails as a technical problem, something the model team or the platform vendor handles. The IDC framework pushes back on that assumption.

Human judgment is a guardrail. When an agent recommends a configuration change and a senior engineer signs off without questioning the logic chain, that’s a failure of critical thinking. It has nothing to do with the model. When a team can’t explain an AI-assisted decision to a skeptical regulator, that’s a trust-building failure, not a communications problem.

The framework is built on the premise that humans need to be explicitly trained to catch the things AI will miss. And they must do it consistently, not just when they happen to be paying attention.

The hybrid role gap

Organizations are already creating hybrid roles that sit across IT, operations and the business. IDC is seeing the rise of workflow orchestrators, risk monitors, human-agent collaboration leads. These roles typically get staffed with strong technologists. The problem is that the skills those roles actually require (facilitation, change management, cross-functional sensemaking, storytelling) often aren’t part of a technologist’s development path.

The framework gives HR and L&D leaders a map for closing that gap intentionally. Because it’s modular, it can plug into existing competency models rather than requiring organizations to start from scratch.

Saying no is a skill, too. Employees need the discipline to decline AI-driven solutions that are too risky, too irrelevant and don’t fit the context. That is a trainable behavior that appears in the framework under the strategizing AI use cluster, alongside opportunity-sensing and pilot-to-scale judgment.

The inclusion is deliberate. Organizations stuck in perpetual proof-of-concept mode, or chasing automation for automation’s sake, need leaders who can push back on bad AI ideas as clearly as they can champion good ones. That requires analytical judgment.

What to do with it

The framework isn’t meant to be read once and filed. IDC’s guidance for technology buyers, appended to the research, emphasizes keeping it alive. As AI agents grow more capable and new use cases emerge, the skills and training examples need to update alongside them. A static competency model will age out fast.

The practical starting point: map a small set of critical behaviors to specific AI use cases. Avoid long competency lists. Equip managers to coach employees on AI use day to day in the flow of work. That’s a lot more effective than just pointing people toward courses. Build cross-functional squads that experiment with AI workflows and share what they learn.

The central bet the framework makes is that organizations succeeding with agentic AI won’t be the ones with the most sophisticated models. They’ll be the ones whose people know what to do when the model gets it wrong.

Gina Smith, Ph.D

Gina Smith, Ph.D - Senior Research Director, IT Skills for Digital Business

As a Senior Research Director at IDC, Gina Smith produces research in the IT education and skills sector. Her responsibilities include primary research, analysis, and the production of market insights worldwide. The New York Times bestselling author of Apple cofounder Steve Wozniak’s…

Enterprise software has always been built around the assumption that there will be a human sitting at the keyboard. Every design decision, every workflow, every pricing model has been optimized for people interacting with interfaces. That assumption is collapsing faster than most vendors anticipated. Vendors who don’t adapt won’t just lose market share. They’ll become infrastructure.

AI agents don’t care about your UI. They don’t appreciate an elegant dashboard or a well-designed navigation menu. What they need is data access, API depth, and integration reliability. That’s a fundamental shift in what enterprise software is actually for. And the numbers suggest it’s happening faster than most vendors anticipated.

The moment is now

When Anthropic launched the Model Context Protocol (MCP) in November 2024, it recorded roughly 100,000 downloads in its first month, a solid start for a new integration standard. When OpenAI adopted MCP in March 2025, monthly downloads jumped to 22 million within weeks. A more than two-hundred-fold increase in a matter of days. That’s not the adoption curve of an interesting experiment. That’s a standard taking hold.

Several vendors moved quickly. Salesforce launched the Agentforce Sales app inside ChatGPT in open beta in December 2025. Block, Pfizer, and Cloudflare all started running agent-driven workflows in production. The theoretical became operational, and the competitive dynamics of enterprise software reset while some vendors were still debating whether agents were real.

Own the agent or feed it

The most instructive thing happening right now isn’t any single vendor move. It’s the contrast between how different vendors are positioning for an agent-driven world.

Salesforce built Agentforce in 2024 as a proprietary agent harness: running inside the Salesforce ecosystem, optimized for sales and customer service, priced as a premium product. The logic was clean. Own the orchestration layer, control the experience, deepen customer commitment. Then in 2025, they pivoted toward MCP. An Agentforce MCP client entered beta. Salesforce-hosted MCP servers went into beta at Dreamforce in October 2025, with general availability in 2026. The Agentforce Sales app went live inside ChatGPT in December 2025. This is textbook hedging: they want customers in Agentforce, but they know external agents are coming regardless, so they’re making sure Salesforce is in the workflow either way.

Contrast that with a different posture visible across several ERP and back-office vendors: positioning the system of record as the agent destination, not the agent source. The pitch is openness. Rather than racing to own the AI layer, these vendors are investing in clean APIs and MCP compatibility so external agents can call their systems without friction, whatever model or orchestration layer the customer chooses. They’re not betting on owning the agent. They’re betting on being indispensable to it.

Two coherent bets on the same future. One side believes they can own the orchestration layer. The other believes the durable value sits in being the cleanest, most accessible system of record. Both could be right. What matters is that neither is waiting to find out.

From sandbox to production

Adoption at the enterprise level matters because it removes the theoretical from the conversation.

Block built an internal AI agent called Goose, running on MCP, that orchestrates workflows across internal systems and connects data warehouses to internal service registries. Cloudflare built an internally governed MCP platform that lets teams expose internal resources to agents without creating security exposure, with governance built at the protocol layer rather than bolted on afterward. Enterprises across financial services and life sciences are using MCP servers for controlled agent access to sensitive data under the same architecture. This pattern keeps surfacing: agents at the orchestration layer, enterprise systems as the data and execution layer, humans reviewing outputs rather than executing workflows.

These aren’t proofs of concept. They’re production deployments at sophisticated organizations. That pattern is consistent enough to call it a new architecture for enterprise computing, not just a trend.

The UX moat is gone

If agents become the primary orchestration layer (and the current trajectory points there), competitive advantage in enterprise software shifts from interface quality to data quality and API comprehensiveness.

A cross-application agent coordinating work across CRM, ERP, and HCM systems isn’t choosing platforms based on how they look. It’s choosing them based on how reliably they expose their capabilities and how trustworthy their data is. That’s existential for vendors whose moats are built on UX. And it forces a rethink of pricing, go-to-market, and product strategy from the ground up. If the primary user of your platform might be an AI rather than a human, who are you selling to? How do you price when consumption is driven by agent calls rather than seat count? These aren’t hypothetical future problems. Forward-looking vendors are working through them right now.

Make the call before the market does

Vendors best positioned for this shift are investing in API depth, clean data models, and MCP compatibility, and making a clear call on whether they want to own the orchestration layer or be the best-in-class system agents call. The vendors most exposed are still competing primarily on interface, without a coherent answer to the agent question.

The deeper divide: which vendors become the nerve center of the agent era, and which become the infrastructure those agents run on? Both are viable businesses. But they require completely different product strategies, pricing models, and customer relationships.

The window to make that choice deliberately is closing. Vendors who wait for the market to decide will find the decision has already been made.

Eric Newmark

Eric Newmark - Group Vice President & General Manager of IDC's SaaS, Enterprise Software, CX and Workplace Solutions Division

Eric Newmark is Group Vice President & General Manager of IDC’s SaaS, Enterprise Software, CX, and Workplace Solutions Division, which includes several teams of analysts covering SaaS, 18 enterprise application markets, software monetization, business platforms, marketplaces, and services firms focused…

There is no shortage of ways to get an answer quickly. A question into ChatGPT, a search across a few analyst portals, a summary pulled from a PDF someone emailed three weeks ago. In under ten minutes, you have something that looks like intelligence. 

The problem is what happens next. When that answer gets questioned, challenged, or presented to a board that wants to know where it came from. 

This is the AI answer gap: the widening distance between the speed at which AI tools can produce answers and the standard to which those answers must be held in real decision-making environments. It is not a technology problem. It is a structural one. And it is getting harder to ignore. 

Three failures hiding inside one problem 

The AI answer gap isn’t a single breakdown. It’s three distinct failures that tend to travel together, and that compound each other when they do. 

The speed-to-answer gap 

When a critical decision arrives: a market entry call, a board presentation, a competitive repositioning. How long does it actually take your team to ground it in trusted, defensible research? For most organizations, the honest answer is longer than it should be. 

In IDC’s conversations with enterprise technology leaders, the time between a strategic question and a confident, evidence-based answer is consistently measured in days or weeks, not hours. Teams spend that time searching: pulling reports, cross-referencing sources, building a coherent picture from fragments. 

The cost isn’t just time. It’s the decisions that get made on incomplete intelligence. 

The AI credibility crisis 

The intuitive fix is to use AI to speed up the research process. That introduces a second problem the first one obscures. 

Public AI tools are fast. They are also wrong in ways that are difficult to detect until the moment they matter most. Hallucinated citations. Outdated market data presented as current. Confident summaries of research that doesn’t exist. The pattern is consistent enough that “AI hallucination” has become a standard budget discussion item for enterprise technology teams. 

The deeper issue isn’t accuracy in isolation. It’s accountability. When a technology leader presents a recommendation to the CFO or the board, the question is never just “is this correct?” It’s “where did this come from, and can you defend it?” A fast answer with no traceable source fails that test regardless of whether it happens to be right. 

“An AI backed by IDC’s research gives me a lot more confidence in the answers.”
—
Phillip Langeberg, CTO, The Resorts Companies

Confidence isn’t just about accuracy. It’s about provenance. The ability to say: here is the answer, here is the source, here is the reasoning — and have all three hold up under scrutiny. 

The workflow intelligence barrier 

The third failure is the most underestimated. Even when high-quality intelligence exists: proprietary research, trusted analyst data, validated market sizing. It tends to live somewhere other than where decisions get made. 

A portal that three people have bookmarked. A PDF that gets forwarded by email. A report that someone read six months ago and summarized in a slide deck that may or may not reflect the current version. Intelligence that should be shaping decisions is instead sitting one context switch away from the people who need it. 

The result is a structural gap between the intelligence an organization has access to and the answers that actually inform its decisions. Closing the AI answer gap requires more than better data. It requires rethinking where intelligence lives. 

Why the gap is widening, not closing 

The intuitive assumption is that more AI tools mean better answers. The evidence suggests the opposite is happening. As the number of AI research tools available to enterprise teams has multiplied, so has the complexity of the research environment: more sources to reconcile, more outputs to verify, more decisions about which tool to trust for which type of question. 

Across IDC’s research with enterprise technology organizations, the pattern is consistent: the teams that make the fastest, most confident decisions aren’t the ones with the most tools. They’re the ones whose intelligence infrastructure is the most coherent, where trusted data, workflow integration, and traceability work together rather than separately. 

That’s a design problem. And it’s one that the current generation of AI research tools, built to maximize speed rather than defensibility, hasn’t been designed to solve. 

What closing the AI answer gap requires 

Addressing the AI answer gap doesn’t mean slowing down. It means building toward a different standard, one where speed and defensibility aren’t in tension. 

Three things distinguish intelligence infrastructure that closes the gap from infrastructure that widens it. First, answers need to be traceable: every output should be linkable to a source that can be examined, challenged, and defended. Second, intelligence needs to live where decisions happen, not in a separate system that requires deliberate effort to access. Third, the research base itself needs to be trustworthy at the source: proprietary, current, and built on a methodology that can withstand scrutiny. 

These aren’t aspirational standards. They’re the minimum bar for answers that are usable in a real decision-making environment. The organizations closing the AI answer gap are the ones treating them as requirements, not nice-to-haves. 

The path forward 

The AI answer gap is a structural problem, and structural problems require structural solutions. Adding faster tools to an incoherent intelligence infrastructure doesn’t close the gap. It accelerates it. 

IDC has spent six decades building the proprietary research base that enterprise technology decisions depend on. The next step is making that intelligence accessible where decisions actually happen, embedded in the workflows, the tools, and the moments where the AI answer gap currently lives. 

IDC Quanta is that platform. If you want to stay informed as it launches this summer, the link below takes you there. 

Ryan Smith - Content Marketing Director - IDC

Ryan Smith is the Director of Content Marketing at IDC, where he leads brand-level content and social media strategy, aligning research insights with compelling storytelling to engage technology decision-makers. With a background in both IT and marketing, Ryan brings a unique blend of technical understanding and creative strategy to his work. He’s also a seasoned storyteller, speaker, and podcast host who believes the right message, told the right way, can drive both trust and transformation.

When IDC CEO Lorenzo Larini chose China as one of his first major destinations as a new chief executive, he wasn’t following a script. He was reading the data. 

Four months into his tenure, Larini is heading to Beijing to see firsthand what IDC’s analysts have been tracking for years: a market in a fundamental shift that most Western executives are still processing from a distance. He asked to meet with robotics companies. He wanted to see the technology himself. For Kitty Fok, IDC’s Managing Director in China and a 20-year veteran of the market, that instinct signals something important. 

“The energy in China right now, the innovation, the change since COVID, it is something very different from six years ago,” says Fok. “I really want him to feel that after his visit.” 

If your mental model of China was formed before the pandemic, it is time for an update. 

What does it mean that China is now a technology innovation engine? 

China is no longer just the world’s manufacturing center. It is one of the most active technology innovation markets on the planet, and the products, platforms, and business models coming out of China are already reshaping global competition. 

That transformation was built on three structural forces IDC has tracked across its 77 dedicated China research programs: the scale of China’s manufacturing base, the size of its domestic population and the demand it generates, and the external pressures that have pushed Chinese companies to build solutions from the ground up. 

Chinese vendors now lead globally in smartphones and electric vehicles. Robotics is where this next wave becomes impossible to ignore. IDC’s newly launched research on embodied intelligence projects China’s robotics market growing from $1.4 billion today to $77 billion in five years. That is 94% compound annual growth, every single year. Chinese robots are already shipping to Asia Pacific, Europe, and North America. This is not a future trend. It is already happening. And when you build for a billion people first, export follows naturally.  

Now look at the software layer. Alibaba built one of the most sophisticated e-commerce and logistics networks in the world. TikTok changed how a generation consumes content. And last year, DeepSeek moved global markets. Not because it was built with the most advanced compute infrastructure available, but because it was built without it. 

“We couldn’t get enough GPU from Nvidia because of restrictions,” says Fok. “So we had to change. DeepSeek happened because we didn’t have enough compute power. We had to innovate and create a different algorithm.” 

Necessity is a very effective R&D strategy. Western companies not watching this closely are working from an increasingly incomplete picture. 

What is the Belt and Road Initiative and why does it matter for global business? 

China’s innovation story doesn’t stop at its borders. That’s where the Belt and Road Initiative comes in. 

Modeled on the ancient Silk Road, it is the Chinese government’s coordinated strategy to help Chinese companies expand globally. In practice, the government establishes diplomatic relationships in target markets, state-owned enterprises move in first and build infrastructure, and private Chinese companies follow with a path already cleared. 

“The government itself is meeting with governments along the Silk Road, creating a better environment,” says Fok. “The state-owned enterprises set up infrastructure. Then all the other Chinese companies can go along with that.” 

The result is a wave of Chinese companies entering markets in the Middle East, Africa, and Europe that operate differently from what Western incumbents are used to: greater pricing flexibility, faster customization, and business models most Western companies haven’t encountered before. If you sell into those markets, you have new competition. If you want to reach Chinese enterprises expanding globally, they are actively looking for IT partners who understand their destination markets. Both are opportunities most Western companies haven’t fully accounted for. 

What Western companies get wrong about China 

The most common mistake IDC has observed across decades of working with multinationals is treating China as a sales geography rather than a distinct market. 

“The first mistake is assuming that anything that works in the US can sell in China,” says Fok. “I have seen American companies translate their product catalog into Chinese and hope that it will sell. It does not work like that.” 

China is a hardware-centric market. Even security companies need to sell appliances, not just software. When AI arrived, buyers expected AI devices alongside AI services. The product, the messaging, and the go-to-market approach all need to be built for the market, not translated into it. 

The second mistake is chasing the wrong benchmark. The competitive density in China is real, but a multinational doesn’t need to win it all. 

The key is finding the right slice. IDC helps companies identify their real addressable market in China, not the theoretical total, but the segment where a multinational can realistically compete and win. 

What does IDC uniquely offer for China strategy? 

That ground-level intelligence is what IDC has been building for approximately 40 years in China. IDC was the first foreign company to receive a media license in the country. Kitty Fok has spent 20 of her 30 years at IDC working in the Chinese market. 

Today, IDC operates 77 dedicated China research programs and has more than 70 analysts based in China. IDC’s nearest Western competitor has roughly 20. That presence matters because China is not a single market. Its provinces operate at different levels of maturity, its industries are distributed unevenly, and understanding which segment is addressable for a foreign company requires people talking to buyers and vendors every week, not a global report applied from the outside. 

“Sometimes we need to talk to the headquarters of a multinational to help them understand why China needs a different strategy,” says Fok. “Why it needs to be more than a sales office. That conversation requires local knowledge, not just market data.” 

IDC’s China intelligence works in both directions: helping Western companies enter China with the right strategy, and tracking how Chinese enterprises are expanding globally so Western companies in those markets know what’s coming. 

The signal in Larini’s visit 

Which brings the story back to Lorenzo Larini. 

Japan and China are the two largest single markets for IDC in Asia, and Larini is visiting both within months of taking the role, alongside IDC CFO Tiziana Figliola.  

He’s not just scheduling meetings. He is looking to meet with robotics companies and see the technology firsthand.  

According to Figliola—who lived and worked in China prior to becoming CFO at IDC— making this visit with Larini is a signal of how seriously IDC is weighing China as a driver of its next phase of growth. “What Larini will find is not the China of 2019. The technology is different. The competitive landscape is different. The pace of change is different in ways that are hard to appreciate without being there,” said Figliola.

For Fok, Larini’s visit matters beyond IDC. Since COVID, many Western executives have quietly stepped back from China, and a six-year gap in direct engagement is a long time in a market moving this fast.

That is the invitation for any Western executive weighing a China strategy. Come and see it. And when you are ready to act on what you see, make sure the intelligence behind your decisions comes from people who have been watching this market for decades. 

IDC has 70 analysts in China right now. The data is there. The question is whether you are ready to use it. 

Ryan Smith - Content Marketing Director - IDC

Ryan Smith is the Director of Content Marketing at IDC, where he leads brand-level content and social media strategy, aligning research insights with compelling storytelling to engage technology decision-makers. With a background in both IT and marketing, Ryan brings a unique blend of technical understanding and creative strategy to his work. He’s also a seasoned storyteller, speaker, and podcast host who believes the right message, told the right way, can drive both trust and transformation.

Christina Cardoza - Content Marketing Manager - IDC

Christina Cardoza is a Content Marketing Manager at IDC, where she specializes in brand content and social media strategy. With a background in journalism and editorial leadership, she has a proven ability to transform complex technology topics into clear, actionable insights.

AI is reshaping how partner ecosystems create value. As automation absorbs traditional delivery work, partner-led growth is shifting toward influence, orchestration, and trusted advisory roles. Here’s what ecosystem leaders need to know for 2026. 

Artificial intelligence (AI) has moved from experiment to operating reality across the technology ecosystem. Models are scaling fast, agents are being embedded into platforms, and automation is absorbing activities that partners have historically delivered as billable services. For ecosystem leaders, this has created an uncomfortable question: where does partner value sit in an AI-driven world – and how does growth happen next? 

The short answer is that partners are not being disintermediated. But their role is being fundamentally rewritten. 

Across IDC’s 2026 partnering ecosystem research, partner advisory board discussions, and recent market analysis, a consistent pattern is emerging. AI is not eliminating partners – it is reshaping where, when, and how value is created across the ecosystem. This shift is happening faster than many organizations’ partner models, metrics, and engagement structures can keep up with. 

How AI Is Shifting Partner Value from Delivery to Decision Influence 

One of the clearest signals we see is that AI is absorbing many traditional partner value activities: configuration, basic implementation, tier-one support, and even elements of solution design. Agentic systems and AI-driven service layers are reducing friction and cost in these areas – but they are also compressing margins and undermining longstanding partner revenue assumptions. 

At the same time, customer expectations are changing just as quickly. Buyers are no longer navigating linear vendor-to-partner journeys, or making decisions through a single buying role. Instead, buying journeys are fragmented, non-linear, and shaped by multiple stakeholders across IT, business, finance, and operations. 

Influence is spread across a blend of direct and indirect touchpoints – platforms, marketplaces, peer communities, embedded AI recommendations, and trusted advisory relationships that cut across suppliers and traditional channel boundaries. 

In this environment, execution alone is no longer a sufficient source of partner differentiation. 

What customers increasingly need is help deciding where and how to apply AI, how to control risk and cost, how to curate models and agents, and how to operationalize them inside real business processes. That shifts partner value up the stack – from doing, to guiding; from delivering, to shaping outcomes. 

Why AI Is Raising the Bar for Partner Value in the Ecosystem 

Another consistent theme from partner advisory board discussions is that while AI lowers barriers to entry in some areas, successful AI adoption at customers is anything but “plug and play.” 

AI must be: 

  • Created and adapted for specific business contexts  
  • Curated across multiple models, tools, and platforms  
  • Governed for security, cost, compliance, and performance  
  • Controlled over time as usage, scale, and risk increase 

These are not one-off tasks. They require ongoing judgment, trust, and ecosystem coordination. In practice, this makes the partner role – particularly the trusted advisor role – more critical, not less, as the pace of change accelerates. 

The challenge is that many partner programs and metrics are still optimized for yesterday’s value creation logic: certifications, headcount leverage, implementation scale, or resale volume. As AI reshapes economics, these signals tell an increasingly incomplete story. 

AI-Driven Partner Transformation: What’s Changing in 2026 

All of this points to 2026 as a transition year. Not a collapse of partner models, but a period of accelerated partner transformation. 

We see clear divergence emerging: 

  • Partners who double down on execution alone face margin pressure  
  • Partners who invest in advisory, IP, ecosystem orchestration, and AI control layers are gaining influence  
  • Vendors and platforms that continue to treat all partners the same risk misallocating investment 

The opportunity now is to rethink ecosystem strategy through the lens of buyer influence, value creation, and AI-driven change, rather than legacy program structures. 

Ecosystem Strategy in 2026: Key Trends and Insights 

In our upcoming webinar, Ecosystem Strategy in 2026: Turning AI Disruption into Partner-Led Growth, IDC will bring together: 

  • 2026 partnering ecosystem trends  
  • Real-world partner advisory board insights  
  • Buyer journey and influence models  
  • Practical guidance on where focused ecosystem investment can drive sustainable growth 

This session is designed for leaders across cloud platforms, enterprise SaaS, GSIs, distributors, infrastructure vendors, and AI-driven ecosystems who are shaping partner and growth strategy for the years ahead. 

If you are navigating how AI is reshaping your ecosystem – and where partners fit next – this is a conversation you will not want to miss. 

If you have any questions about anything in this blog, please drop them in here.  

Stuart Wilson

Stuart Wilson - VP, WW Partnering Ecosystems, Alliances and Channels

Stuart Wilson is VP, Worldwide Partnering Ecosystems, Alliances & Channels at IDC. He leads IDC's research and advisory agenda focused on partnering ecosystems, strategic alliances, channels, marketplaces, and ecosystem-led growth. He currently leads IDC's EMEA Partnering Ecosystems research program and…
Andreas Storz

Andreas Storz - Research Director, Software Channels and Ecosystems

Andreas Storz is Research Director for IDC's Software Channels and Ecosystems program. Based in the US, Andreas focuses on the evolution of go-to-market models, new digital value chains and the wider impact on partner ecosystems, exploring how current and future trends will…

近日,国家网信办、国家发展改革委、工业和信息化部联合印发《智能体规范应用与创新发展实施意见》(以下简称《实施意见》)。这是国家层面首次针对智能体这一人工智能产品形态,给出覆盖技术底座、标准协议、场景应用、安全治理和产业生态的系统性政策框架。这意味着企业级智能体从过去的小规模技术试点阶段,正式进入以治理、协同和规模化为核心的新竞争周期。

从IDC的视角看,《实施意见》把技术底座、安全治理、应用牵引和创新生态构建成了一张全局路线图。技术与标准构成发展基础,分类分级与权限管控划定行为边界,五大类典型场景提供规模化落地入口,开源与供需平台则负责连接研发与市场。

在IDC预测中,2026年和2027年将是是中国企业场景中的活跃智能体数量增速最快的两年,单年同比增长超过200%,并将在2031年达到3.5亿个活跃智能体,2026-2031年CAGR达135.3%(《中国智能体规模预测,2026-2031》)。而《实施意见》这样的智能体国家顶层政策,也将进一步加速智能体走向规模化的应用。

活跃智能体数量增长反映的是企业部署广度,也会直接放大运行、权限、审计和治理压力。一个企业内部只有几个智能体时,治理方法更接近单个项目的管理;当Agent数量进入快速增长周期,治理就会变成IT架构问题。身份、权限、审计、工具调用、异常回退和跨系统协同,都会从项目交付细节变成平台能力要求。

《实施意见》放在这个阶段变化中看,正是把这些问题推到了产业议程前台,也让企业级Agent从小规模试点建设进入更正式的IT治理和产业协同框架。对企业来说,Agent会更快进入正式IT治理体系;对技术供应商来说,产品竞争也会从功能构建,延伸到运行、治理、评测和生态分发能力。

场景牵引推动Agent规模化复制

《实施意见》明确提出,要围绕科学研究、产业发展、提振消费、民生福祉和社会治理五大方向,打造19个典型应用场景。政策并没有停留在技术倡导层面,而是把智能体放进具体的任务环境里,通过典型场景牵引技术验证、产品迭代和应用落地。

智能体与上一阶段生成式AI应用的差别,就在于它需要进入业务环节端到端的完成任务。生成式AI更多围绕内容生成、知识问答和辅助分析展开,智能体则需要调用工具、访问系统、执行动作,并对结果产生影响。只有进入真实业务场景,任务边界、数据条件、系统接口、权限控制和结果责任才会被定义出来,才能真正落地出好用可用的智能体。

当前智能体在企业业务中的渗透率依然较低,仅有18%(IDC Syndicated Survey 2026: China AI Agents Market)。智能体没有在企业业务中真正规模化应用起来,核心在于缺少清晰的切入场景和可复用的落地路径。政策在这个角度提供了场景牵引,通过典型应用场景帮助智能体走向业务落地,并进一步形成规模化应用。

开发平台将会从构建工具延伸为运行和治理底座

《实施意见》在夯实发展基础部分提出,要提升智能体任务理解、任务规划、工具使用、长期记忆等核心能力,推动智能体互认互通和群体协同,同时完善底层框架、功能组件以及研发、测试、部署、运维工具链。

这与IDC观察到的,智能体平台厂商产品迭代的趋势高度吻合。2025年头部平台的产品形态已经从单纯的开发工具演进为涵盖规划、开发、测试、发布、观测、优化、治理的完整系统,企业也越来越关注智能体深入业务后,智能体是否是否可观测可治理,以及如何确保智能体执行任务的稳定性安全性(具体见IDC即将发布的《中国智能体开发平台市场份额,2025》中的重大市场变化章节);同时,智能体规模化进入企业之后,企业内部多种来源的智能体(应用内智能体、低无代码配置智能体、独立智能体、定制智能体)如何统一管理和协同也会成为企业下一个阶段的关键需求。

《实施意见》中关于工具链、互认互通、群体协同的部署,将把这些工程化能力进一步推向产品标配。这些要求会转化为智能体开发、测试、运行、观测和治理的一整套工程能力,并逐步沉淀为智能体开发平台的核心竞争力。

安全治理成为Agent进入业务流程的准入条件

《实施意见》用相当篇幅讨论产品准则、决策权限、行为管控、内生安全、供应链安全、应用衍生风险、分类分级治理与合规服务体系。安全合规既是政策关注的重点,也是企业实际落地中的关键门槛。IDC在2026年最新智能体调研中发现,62%的企业把数据权限与安全合规列为智能体跨系统执行的首要障碍;而在智能体平台引入目标上,58.7%的企业把治理与合规放在首位(IDC Syndicated Survey 2026)。

智能体只有安全可控,才能正在被企业放心的纳入业务流程中。因此智能体的权限治理、行为审计、决策追溯、安全护栏等能力需逐渐的沉淀为智能体或智能体平台的标准能力。在金融、医疗、政务、公共安全等高合规领域,这种能力更是智能体能否落地的前提条件。

标准化分发和生态将重塑智能体产业

《实施意见》在标准协议和创新生态部分提出,要建立智能体标准体系,加强智能体互联协议(AIP)等关键标准推广应用,探索建设智能体注册平台,提供数字身份管理、检索发现、能力声明等服务,并推动智能体软件商店和行业供需信息发布平台建设。

智能体标准体系的成熟会明显推进智能体,尤其是独立智能体的规模化进程。在IDC 2026年初的预测中,独立智能体是增长弹性最高的智能体形态之一,其爆发主要依赖标准协议和智能体生态的成熟(《中国智能体规模预测,2026-2031》)。AIP、注册平台、软件商店和供需平台逐步成型后,独立智能体的构建、分发和使用门槛都会随之下降,独立智能体尤其是像Xclaw这类智能体产品有机会更快进入企业规模化应用场景中。

行动建议

面向企业用户

  • 把安全治理能力列为智能体规模化前置条件。 企业应尽早建立场景分级、任务边界和权限管理机制。在金融、医疗、政务等高合规领域,应参考政策要求提前规划备案、检测、安全评估以及第三方权威评测,低风险场景也应建立合规自测和风险报告机制。
  • 建设智能体统一运营和协同体系。随着智能体的规模化应用,企业内部将同时存在多类智能体,因此企业在平台选型和架构规划中,需要提前考虑智能体注册、统一身份、统一权限、统一审计、统一成本管理和跨智能体协同能力。

面向技术供应商

  • 安全治理需成为产品基础能力。智能体安全治理相关的决策权限边界、行为管控、最小权限、全链路审计、异常检测和回滚机制等能力,会逐步成为企业采购时的核心指标。
  • 建立智能体全生命周期能力。智能体全生命周期涉及到的开发、测试、部署、运行、评测、优化和治理等均需建立起相应的平台级能力,为未来企业级智能体规模化运营提供支撑。
  • 构建异构智能体的统一管理能力。智能体规模化时代,企业智能体的来源多样,平台需要建立统一的智能体身份、权限、审计、成本和调度能力,帮助企业把分散的智能体管起来,建设企业级的智能体运行和治理入口。
  • 关注智能体标准协议和平台商店的建设。AIP、智能体注册平台、软件商店和供需平台会推动智能体的标准化分发。建议厂商提前规划产品路线,并参考政策要求开展第三方功能、性能、质量与合规评测。

分析师观点

孙振亚,IDC中国分析师:《IDC中国研究经理孙振亚表示,《实施意见》的出台,标志着中国智能体产业从技术和市场自发探索,进入政策框架下的规模化扩张阶段。政策一方面通过分类分级、决策权限、行为管控和合规服务体系,为企业部署智能体提供可预期的合规路径;另一方面通过标准协议、注册平台、软件商店、第三方评测和供需对接机制,为智能体走向跨系统协作和产品化分发铺设了基础设施。

与IDC同行,抢占智能体规模化的战略先机

《实施意见》的发布,标志着企业级智能体正式进入治理驱动、生态协同的新阶段。但对企业来说,仍然存在诸如“如何在自己的行业与业务中,识别高价值的智能体场景”、“如何在平台选型、安全治理、标准化对接等关键决策上少走弯路”等现实挑战。

IDC长期追踪全球与中国智能体市场的发展脉动,拥有覆盖技术供应商、企业用户、投资机构的多维研究体系。已发布《中国Agent基础设施平台/执行平台技术评估,2026》、《中国智能体规模预测,2026—2031》、《IDC Market Glance:中国AI Agent市场概览,1Q26》等智能体报告,即将发布《Agent企业最佳实践与场景精选(ROI视角)》、《中国智能体开发平台市场份额,2025》、以及针对Xclaw产品进行评估的《企业级通用Agent助手评估》等一系列重磅报告,并可为企业提供定制化的场景评估、平台选型与治理成熟度诊断服务。

欢迎您联系IDC中国智能体研究团队,获取最新研究成果,或与分析师进行交流。让我们助您在智能体规模化的浪潮中,走得更稳、更快、更远。

请点击此处与我们联系。

Zhenya Sun

Zhenya Sun - Research Manager

Zhenya Sun is a research manager for the IDC team focused on exploring the application of technology and industrial development of AI and AI agents. He is also responsible for providing clients with consulting services on technologies, products, and markets…

视角决定格局。​若仍将公有云安全视作单一的技术模块,便无法窥见其真正的战略全貌。

IDC 2025全球数据指明:公有云安全已跨越单纯的合规门槛,跃升为企业的数字生存底座。​ 它不再是隐匿于防火墙之后的附属插件,而是支撑AI规模化落地、护航业务出海、乃至抵御周期波动的核心底层资产。

三个信号值得认真对待:

  • 信号一:全球公有云安全支出超1100亿美元,增速超20%,是IT支出中最坚硬的赛道
  • 信号二:中国市场云安全收入增速普遍高于云基础设施本身增速,客户正在从为资源付费向为安全付费跨越
  • 信号三:随着AI业务场景的深化,云原生安全已成为大模型落地的隐性门槛,若缺乏深度集成的原生安全体系,企业很难安心地将高价值数据与应用托管其上

一、全球浪潮:安全正在重新定义云的价值

站在2026年的时间节点回看,公有云早已不再仅仅是企业的外挂资源,而是数字化生存的电力系统。

2025年,全球企业正在混合云和多云架构中疯狂奔跑,全球公有云安全市场连续多年保持每年20%以上增速,2025年支出飙升至1100亿美元。

在全球范围内,AWS、Microsoft Azure和Google Cloud等头部玩家正经历着前所未有的技术迭代。他们不再仅仅提供安全插件,而是通过硬件级隔离、身份边界重塑以及AI驱动的情报分析,将安全深度嵌入云原生底座。

与此同时,以Palo Alto Networks、CrowdStrike、Wiz和Zscaler为代表的安全巨头也在快速进化。他们正推动从单一工具向CNAPP云原生应用保护平台和代码到云(Code-to-Cloud)的统一平台转型,试图在多云环境中建立一道跨厂商、自动化的全栈防御屏障。

在这场技术长跑中,谁能率先解决复杂环境下的可见性真空,谁就将定义下一个十年的云安全标准。

二、中国实践:快速跟进与创新,但仍有差距

2025年,中国云计算市场规模持续扩大,云安全产品的营收增速普遍高于云基础设施本身的增速,显示出客户从买资源向买安全的意识跨越。

作为国内市场的领头羊,阿里云已将其安全产品线全面推向智能体时代,通过升级Agentic SOC,利用大模型实现了超过80%的安全事件自动响应。华为云演进为AI原生安全架构,通过其安全大模型实现对AI算力集群的实时监测。腾讯云充分发挥攻防领域的深厚沉淀,在内容风控、反欺诈、DDoS高防等领域形成了极高的市场占有率。

传统网安巨头正通过云地协同战略,在云安全管理平台C-SOC市场占据核心位置。深信服、奇安信等厂商通过与公有云厂商的深度技术绑定,成功实现了从卖防火墙硬件到卖云安全服务的转型。

但不可忽视的是,相比于全球安全厂商,中国安全厂商目前在公有云安全市场仍需发力。虽然在私有云和特定业务场景中展现了卓越的实战能力,但在全球化的SaaS安全标准、跨云的Agentless治理精度、以及安全左移的代码级闭环上,国内厂商仍有巨大的进阶空间。

三、商业启示:为什么公有云安全是好生意

海外巨头的成功路径为中国厂商提供了极具诱惑力的商业模板:

极致的用户粘性

安全业务之所以具备IT板块中最高的粘性,是因为它正逐渐从外部挂载的锁具演变为深度嵌入企业架构的神经系统。一旦部署,更换安全供应商不再是简单的卸载与重装,而是一场涉及底层策略重写、合规记录迁移以及全员操作习惯重塑的伤筋动骨式手术。

抵御周期的现金流

与传统网络安全硬件的订单式生存不同,Wiz和CrowdStrike的成功路径证明了SaaS化订阅模式在资本市场中的统治力。这种模式具有极强的抗周期属性,即便在经济下行期,企业可能会削减新员工入职或新业务线扩张,但绝不敢轻易关掉云端的安全防护。

掌握AI与全球化的通行证

在当前的全球技术格局下,拥有顶级的云安全能力已不再仅仅是为了防御黑客,它更是企业进军AI云和全球化市场的入场券。只有具备原生云安全能力的厂商,才能解决大模型在云端训练、推理过程中的数据投毒和提示词注入风险。中国厂商若能构建起符合国际主流标准的云安全体系,本质上就是为中国企业的全球化交付提供了一套合规标准化的底座。

四、IDC:不止于数据,更是决策指南

IDC已发布 《IDC中国半年度安全软件数据跟踪报告——公有云,2025H2》,其中包含公有云部署模式下的数据安全软件、终端安全软件、身份和访问控制软件、软件安全网关、安全分析和运营软件、漏洞管理软件等子市场的厂商数据跟踪。

上述商业逻辑的兑现,最终需要落地到真实的市场数据中加以验证。化繁为简,洞见未来。 本次报告透过公有云安全市场不同赛道中的厂商营收数据,为您剥离噪音、还原真相,助您在复杂的市场环境中精准锚定业务增长点与投资机遇。无论您是寻求赛道突围的企业,还是布局未来的投资者,皆能在此获取专属的行动指南。

具体而言,可以为不同角色的参与者提供以下支持——

云厂商:帮助您精准识别自身安全产品的市场位次,明确自身赛道定位与竞争位置。

安全厂商:帮助您应找准生态位,重点锚定高增长赛道,避免无效内卷;同时积极融入头部云厂商生态,通过技术集成与联运合作,借助云市场流量触达客户。

企业客户:为您提供了客观的供应商评价维度,帮助 CIO 识别哪些厂商在特定公有云安全领域具备长期投入和领先地位。

投资机构:报告是评估公有云安全景气度的核心参考,能够清晰展现公有云安全市场的天花板及各子市场的集中度,利用数据降低因信息不对称导致的投资偏差。

结语

公有云安全已经完成了从附加品到必需品的身份蜕变。在客户从买资源向买安全的意识跨越中,IDC 将持续通过系统性的研究,助力每一位参与者:看清坐标、捕捉先机、定义未来。

欢迎广大云厂商和安全厂商关注IDC公有云安全系列研究,如需进一步沟通或获取深度数据与战略洞察,请与IDC联系。

请点击此处与我们联系。

Joe Zhao

Joe Zhao - Senior Research Manager

Joe Zhao is a senior research manager of Enterprise Research for IDC China. He focuses on research and analysis of the China security market. Joe has more than 12 years of domestic and international work experience in ICT. Prior to…