Logo
Become a Client

How do you decide, in an afternoon, whether a drone program is worth a government’s time? For Andrew Shipe, that’s a real, recurring question.

As an Innovation Strategist on Sourcewell’s innovation team, Sourcewell Ventures, Shipe helps a government cooperative serving public agencies across the country figure out which ideas are worth pursuing, and which ones need to wait. Sourcewell itself is unusual as a government entity: it can’t tax to raise revenue, so everything it does has to justify itself as a service governments want. Ventures exists specifically to find opportunities beyond Sourcewell’s core cooperative-purchasing business, new ways to serve public agencies that the private market isn’t equipped, or motivated, to solve on its own.

With a small, focused team, Shipe doesn’t have the luxury of researching every idea that comes in. Every idea has to earn its place quickly, on the strength of the evidence behind it.

The problem with comparing the incomparable

The hardest part of that job is finding a shared yardstick. A county fleet services department’s problem doesn’t look anything like a fire department’s wildfire response, even though both are government agencies wrestling with urgent, resource-constrained work. There’s no obvious way to weigh one against the other. Without a consistent way to evaluate opportunities across such different contexts, good ideas were easy to misjudge, or simply lose to timing while something else got prioritized.

Shipe has come to rely on IDC for this kind of cross-context clarity: understanding where technology is going, and how different types of agencies are adopting it, so that a fleet department’s technology curve and a fire department’s technology curve become at least a little more comparable.

A vague pain point, made concrete

Nowhere did that gap between “everyone says this is a problem” and “we understand why” show up more clearly than with drones.

Agencies at the city and state level had been telling Sourcewell the same thing for a while: drone-based first-responder programs were powerful. They changed how a community could be kept safe. But they weren’t affordable, and nobody seemed to have a clear, specific answer for why. Shipe’s team didn’t have a good sense of where the cost was coming from, which meant they couldn’t tell whether it was a problem Sourcewell could realistically help solve.

Working with IDC and IDC Quanta, Shipe put real numbers behind what agencies had only been able to describe anecdotally, turning a vague complaint into a defined, structural problem a government partner could plan around.

The trust problem underneath every other problem

None of that research would have mattered if Shipe couldn’t trust the numbers behind it. Trust was exactly where every other AI tool he’d used had fallen short.

“That’s kind of always been one of my challenges with other AI tools,” he said, “what is real and what is hallucinated.” It’s a familiar frustration for anyone who has asked a chatbot for a statistic and then had to go verify it independently anyway. At that point, the tool hasn’t saved any time. It’s just moved the work later.

IDC Quanta changed that with inline footnotes tied directly to source articles. A list of links at the bottom of a response never told Shipe which claim came from which source, so he’d never been able to trust those either. IDC Quanta showed him exactly which sentence was backed by exactly which footnote. If a number looked interesting or surprising, he could open the source immediately, in context, and check it against the original article before he ever used it in front of a decision-maker.

“

The appeal of IDC Quanta, he said, was that it’s backed by specific articles, and I could read them and make sure that the numbers I’m getting are correct.

That capability sounds small, but it changes what you can trust the tool to do. Shipe could finally build a real decision on top of what it told him, instead of just using it to get started.

Faster research means more shots on goal

Drones were one example. The same problem (a real question with no fast, trustworthy way to answer it) comes up constantly for a team fielding ideas from every direction.

The Sourcewell Ventures team doesn’t have a shortage-of-ideas problem. Ideas arrive from so many directions that time to evaluate them is the real constraint.

That’s what made IDC Quanta useful beyond any single research question. Shipe could plug in a new idea and get enough grounded, source-checked information back to make a real call: yes, this has potential, worth spending more team time on. Or no, this isn’t the right fit for the market right now, worth setting aside. Multiply that across every idea competing for a small team’s attention, and the value goes beyond accuracy on any one question. It’s the ability to run through far more ideas than the team could have researched manually, without sacrificing confidence in the answers.

Shipe called it a genuine leap forward from every other AI tool he’d used. The reason had nothing to do with flash. It was the first tool he felt he could build a real decision on top of.

Shipe’s team still gets more ideas than it has hours to research. IDC Quanta just means more of those hours go toward the ideas worth the time.

Learn more about IDC Quanta for Tech Leaders

Christina Cardoza - Content Marketing Manager - IDC

LinkedIn
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.

随着生成式AI和智能体加速进入企业生产环境，AI正在从辅助工具逐步转变为参与业务决策和执行的关键基础设施。与此同时，Prompt注入、数据泄露、模型供应链、智能体权限滥用和工具调用等新型风险不断出现，传统网络安全体系已经难以完全覆盖AI应用带来的安全挑战。

在AI应用规模化落地的推动下，AI安全市场也进入快速发展阶段。IDC预测，到2030年中国AI安全市场规模将达到340亿元，未来五年复合增长率超过50%。在市场快速增长的同时，不同AI安全技术的成熟度和落地节奏也开始出现分化，企业面临的核心问题正在从“是否需要AI安全”进一步转向“应该优先建设哪些AI安全能力”。

23项AI安全技术进入不同发展阶段，技术路线开始分化

为了更清晰地判断不同AI安全技术的发展阶段及未来潜力，并帮助用户进行技术选型，国际数据公司（IDC）近日发布了《中国AI安全技术发展路线图，2026》（Doc#CHC54918225 ，2026年10月）报告，该报告对23项新兴及重要的AI安全技术进行了系统评估，并从技术采用阶段、市场热度、风险级别、采用速度以及市场培育时间等维度，对不同AI安全技术的发展状态进行了可视化呈现，360数字安全集团、安华金和、阿里巴巴、安恒信息、长亭科技、硅基流动、火山引擎、华为、君同未来、绿盟科技、领信数科、蚂蚁数科、奇安信、启明星辰集团、深信服科技、腾讯、网宿科技、网易智企、新华三、悬镜安全、亿格云、亚信、中国电信、竹云(按公司首字母顺序排列）入选各技术点推荐厂商，2026年IDC中国AI安全技术发展路线图详见下方：

报告显示，当前AI安全技术尚未成熟，整体市场处在多条技术路线并行发展的阶段。不同技术在市场需求、技术成熟度和企业采用节奏上存在明显差异。这意味着，AI安全市场正在从“技术概念期”逐步进入“技术成熟期”。对于企业而言，未来AI安全建设的关键并不是简单追求技术覆盖，而是根据自身AI应用成熟度和业务风险，判断哪些技术需要立即部署，哪些技术需要持续观察，以及哪些技术值得提前布局。AI安全建设也不再是简单增加一套安全产品，而需要逐步形成覆盖AI资产、身份权限、数据、模型、智能体、供应链、运行时以及安全运营的完整治理体系。

AI安全正在进入一个新的阶段：从保护AI，走向治理AI；从发现风险，走向理解和验证风险；最终从被动防御走向AI驱动的主动安全。具体的技术发展趋势如下：

1. AI安全建设开始从“单点防护”走向体系化治理新阶段

随着企业AI资产快速增长，传统围绕单个模型、单个应用进行安全防护的方式已经难以覆盖企业的AI安全防护需求。企业首先需要解决“用了多少AI、有哪些模型和智能体、连接了哪些数据和工具、存在什么风险”等基础问题，并进一步实现风险持续监测和治理。因此，AI安全正在从过去的单点检测与防护，逐渐向资产发现、风险识别、持续监测和统一治理演进。AISPM有望成为连接AI资产、风险和安全运营的重要基础能力，并推动AI安全从传统的产品能力向体系化治理能力转变。

2. 智能体安全进入快速发展阶段，AI安全的治理对象正在发生变化

从AI安全技术发展路线图中可以看到一个非常明显的变化：AI安全正在从“模型安全”走向“智能体安全”。报告中非人类身份管理、智能体零信任授权、智能体运行时安全、智能体安全网关、智能体通信协议治理和Skills安全检测等技术，正是这一变化的直接体现。IDC认为，随着Agent进入更多核心业务流程，智能体将逐渐成为AI安全治理的核心对象。未来企业需要建立覆盖智能体开发、部署、运行和退出全生命周期的安全治理体系，而不仅仅是在模型外围增加安全检测能力。

3. AI供应链正在成为新的安全边界

过去，企业更多关注软件、代码和基础设施供应链，而AI应用进一步引入了模型、训练数据、开源框架、Agent Skills、MCP Server、第三方工具以及外部AI服务等新的组成部分。这意味着，一旦其中某个环节存在安全问题，都可能通过模型调用、Agent执行或者工具链传递到最终业务系统。因此，AI供应链安全正在成为AI安全体系的重要组成部分。未来企业需要建立覆盖模型、数据、代码、工具和基础设施的全链条安全治理能力，并逐步完善AI物料清单、供应链风险评估和运行时监测等能力。

4. AI安全正在从“上线前评估”走向“持续风险管理”

AI系统与传统IT系统最大的不同之一，在于其运行状态具有较强的动态性。模型可能持续更新，Prompt会发生变化，Agent任务会动态调整，数据访问和工具调用关系也会不断变化。在这种情况下，一次性的模型安全评估很难覆盖AI系统整个生命周期的风险。IDC认为，AI安全正在逐渐形成一条完整的治理链路：AI资产发现 → 持续风险评估 → 运行时监测 → 持续检测 → 安全响应 → 审计与合规。这意味着AI安全的治理模式正在发生变化：从过去关注“上线之前有没有风险”，转向持续回答“运行过程中风险有没有发生变化”。未来，AI安全产品也将进一步从单点能力向平台化方向演进，逐渐形成围绕AI资产、风险、威胁和治理的统一安全运营体系。

5. AI安全正在从“技术中心”走向“业务中心”

过去，AI安全更多关注模型是否存在越狱、Prompt注入等安全问题。但随着AI越来越深地进入企业业务流程，仅仅判断模型“是否安全”已经不够。企业还需要知道：这个AI服务于什么业务？它访问了哪些数据？拥有哪些权限？调用了哪些工具？执行的行为是否符合业务流程？一旦发生异常，又会对业务造成多大影响？因此，AI安全正在从单纯的技术防护进一步向业务风险治理演进。未来的AI安全平台需要将AI资产、数据、身份、权限、业务流程和风险上下文关联起来，在业务场景中判断什么是正常行为、什么是异常行为。这也意味着，AI安全的最终形态可能不再是一组孤立的安全产品，而是一个能够理解AI、理解业务并持续治理AI风险的平台。

6. 从“发现风险”走向“验证与自主修复”

AI能力本身的发展，也正在改变安全防护的技术范式。随着攻击者越来越多地利用AI生成攻击Payload、实施Prompt注入、寻找漏洞和自动化攻击链，传统依赖人工分析和规则匹配的安全防御方式将面临效率瓶颈。因此，“AI对抗AI”将成为AI安全未来的重要技术范式。AI不仅是需要被保护的对象，也将逐渐成为安全检测、攻击模拟、威胁分析、风险验证和自动化响应的重要执行力量。在此基础上，可验证AI、AI红队、AI检测与响应等技术的发展，将推动安全能力从单纯的风险识别向更高程度的自动化和自主化发展。

IDC 中国网络安全领域研究经理王一汀表示，当前，生成式AI正式进入规模化应用阶段，AI安全也随之从模型与应用防护，向智能体身份权限、AI供应链、运行时安全和全生命周期安全治理等方向演进。技术范式上“AI对抗AI”将成为核心发展方向。治理模式上，模型评估将走向持续风险治理，并进一步融入业务场景进行风险管控。除此之外，可验证AI、AI自我修复等前沿技术不断涌现，推动AI安全从“发现风险、阻断风险”向“理解风险、验证安全、自主修复”迈进。

总体来看，AI 安全正在进入一个新的建设周期。技术路线虽然开始分化，但企业行动路径已经逐渐清晰：先摸清 AI 资产和智能体身份，再管住权限、运行时、数据和供应链风险，最终形成覆盖全生命周期的持续治理能力。对CIO和安全负责人而言，当前的关键不是等待技术完全成熟，而是根据自身AI应用成熟度和业务风险，确定哪些能力需要立即部署、哪些需要持续观察、哪些值得提前布局。

进一步交流

如果您的企业正在推进 AI 安全体系建设，或在智能体安全、AI 供应链、运行时治理等方面遇到选型与落地难题，欢迎与 IDC 中国网络安全研究团队进一步交流。IDC 可提供技术路线解读、市场趋势分析、厂商能力评估及定制化研究服务，帮助企业在快速变化的 AI 安全市场中明确优先级、降低选型风险。请点击此处与我们联系。

Sophia Wang, CISSP

Sophia Wang, CISSP - Research Manager

Analyst Profile
Sophia Wang is a Research Manager in IDC China. She is responsible for the analysis and research of China's cybersecurity market. Her primary focus is on China's cybersecurity appliance and services market and operational technology (OT) security market. Additionally, she…

最近Meta的Muse特别火。9月在美国正式上线之后，讨论度一直居高不下。

如果从智能体的能力来看，Muse并不是一个全新的产品。它依然采用典型的智能体架构，基于用户的指令或需求，思考规划之后，调用相关的工具或技能完成特定任务，这些也都是当前智能体产品普遍在做的事情。

但为什么是Muse？为什么国外同样在做的智能体产品，像OpenAI的ChatGPT Work、Anthropic的Claude Cowork，并没有像Muse这样引起现象级的关注和传播？

国际数据公司（IDC） 在近期发布的《中国企业级通用 Agent 产品技术评估》中，从记忆、任务执行、生态连接等维度对主流智能体产品进行了评估。这些维度同样适用于理解 Muse 和国内个人助理 Agent 的差距。

为什么是Muse

为什么是Muse，核心原因主要还是Muse的场景定位和这类产品有一定的区别。ChatGPT Work、Claude Cowork以工作场景为主且用户大多是职场人员，Muse的定位是纯粹的个人生活助理，它的目标就是帮个人完成大量日常生活中的事情。

Muse的使用案例里，更让人印象深刻的是那些人做起来既耗时又麻烦的事情。比如Muse可以帮用户给保险公司打电话完成理赔；可以根据美食视频列出采购清单并直接采购；可以帮用户以更好的价格售卖二手物品；帮用户和商家讨价还价等等。

这类琐事虽然看起来很小，但需求更普遍，也更贴近普通人的日常生活中的核心痛点。用户感受到的是AI真的可以帮助他解决生活中麻烦的事情，这种体验很容易被分享，形成传播效应。

生态是个人助理智能体的基础

要达到一个好用的个人助理Agent，当前需要具备一个非常基础的能力，就是整个服务生态的成熟度。这里讲的服务生态，是智能体完成用户日常任务所需要的各类服务系统和工具能否被智能体安全可控的调度使用。包括购物、订票、订餐、出行、支付系统，账号登录体系，以及电话、邮件、日历这些日常工具。这些系统构成了日常任务的完整闭环，如果有缺失，用户就不得不手动介入做填写点击和上下文的搬运工作，体验就会大打折扣。

在 IDC 的评估框架中，生态连接能力是衡量通用 Agent 成熟度的关键维度之一。智能体能否安全可控地调度外部服务，直接决定其任务边界。

Muse在这方面做了很多事情，在上线初Muse就已经接入了Spotify、OpenTable、Ticketmaster、Plaid等合作伙伴，覆盖音乐、餐饮、演艺票务和金融账户。支付环节Muse可以通过Stripe的Link生成一次虚拟卡号完成结账，其它包括PayPal和1Password账号系统也都接入了Muse。同时Meta也给Muse打通了WhatsApp、Facebook、Instagram这些其旗下的社交应用。

同时对于还没有合作的服务或生态，Muse也可以通过浏览器操作的方式，像人一样操作网页打通更多场景和生态。但浏览器操作也有它的边界，平台如果不认可，这种模拟用户的操作随时可能被限制，比如亚马逊就在Muse上线不久屏蔽了它的购物操作。所以生态的合作关系，以及整个服务市场For Agent的成熟度，真正决定了个人助理智能体的任务边界。

国内尝试：千问APP与豆包手机

国内我们也看到不少厂商做过相关尝试。例如千问早在今年1月就把淘宝、支付宝、闪购、飞猪、高德等阿里系业务打通，上线了400多项AI办事功能；字节跳动的豆包手机助手也尝试从操作系统底层打通App应用的生态，但是他们都还没有做得非常成功。

千问的核心问题是，它还没有真正的围绕千问APP去打造生态，还是把千问APP作为流量入口的思路，去给阿里的应用导流。所以用户在使用千问APP时仍然会有严重的割裂感，它不了解用户购物的偏好，仍然需要用户在弹出的淘宝卡片中完成最终的下单和支付，非阿里系的生态的能力也不完整。

豆包手机最开始做得是非常有野心，在25年末最初版本的能够基于系统级权限，像人一样直接操作手机上的各类应用，跨App比价点外卖、订机票，甚至帮用户回复微信消息。从跨应用自主执行任务的形态来看，它已经很接近Muse今天的样子。但随后就遭到了各个应用厂商的强烈抵制，微信出现账号异常下线，淘宝、京东和多家银行App也相继采取了风控和限制措施。豆包很快下线了操作微信的能力，并对金融类应用等场景的操作也都做了限制。

而Muse则和商家等服务方建立起了商业合作的生态，共同为用户提供服务。Muse通过推荐和分发为商家带来流量和订单，Meta则从Muse促成的交易中获得收益。

IDC 的评估显示，国内主流智能体产品在记忆和常规端到端任务上表现已经不错，差距更多体现在生态壁垒和跨平台协同上。

产品未来：从被动式助理到Agentic管家型助理

管家型助理和被动式助理的区别在于，很多常用的需求和生活习惯，你不需要告诉管家，它能够记住，并在合适的时间、合适的场合自主地帮你解决。而被动式助理还需要你告诉它要做什么。在当前我们看到，Muse已经有了一点管家的影子，它会记住用户只提过一次的细节，比如用户的购物偏好，朋友的饮食禁忌，在很多场景下不需要用户主动开口重复，就能够直接先给出合理建议。

这些能力需要产品深度了解用户日常生活的信息，也就是日常上下文才能真正实现，这一方面需要有完善的记忆系统的支持，另一方面，像Meta打通WhatsApp、Facebook、Instagram等社交应用，以及和智能眼镜等消费级智能硬件融合，也是为了更好的无缝打通用户日常生活中的上下文，以支持智能体基于用户日常的行为偏好来完成任务。

国内市场的智能体产品离Muse并不遥远

上面提到的智能体的关键能力，生态、记忆和端到端的任务能力，在IDC发布的《中国企业级通用Agent产品技术评估》报告中，都有相关评估。报告的测试场景虽然以企业场景为主，但这些能力，对面向个人助手场景的智能体同样适用。

从IDC评估结果来看，目前中国市场主流的智能体产品已经具备相关能力：

  • 在记忆方面，大部分产品已经能够很好的支持包括用户记忆在内的各种分层分类的记忆体系，可以保存用户偏好、任务上下文和关键需求。
  • 在任务方面，常规的端到端任务完成能力也非常不错，而像需要基于浏览器操作或桌面操作的的这类复杂任务，部分智能体也有较好的表现。
  • 在生态方面，有一些产品已经走出办公场景，和生活服务的生态做打通。比如WorkBuddy与美团合作，可以帮用户领取优惠券并发起交易，通过微信支付的AI专属卡完成支付。其它如百度搭子、纳米Work、Teleagent、商汤小浣熊等Agent产品也具备一定的生活服务能力。

所以中国的这些通用智能体产品离Muse并不遥远，差距主要还是在这些服务生态的壁垒上。

IDC未来建议

提高生态开放性，建立生态开放标准

当前中国市场仍以超级App为主，大量服务、数据和交易都封闭在各家平台内部。豆包手机的遭遇也说明，单靠技术手段绕过平台很难走通。行业需要建立面向智能体的开放标准，包括智能体的身份如何标识、能调用哪些服务、交易收益如何分配、出了问题由谁负责，让平台、商家和智能体厂商都能找到参与的理由，打造为智能体服务的应用生态。

构建安全可控的个人记忆系统

个人日常生活的记忆，需要更安全、更私密，同时也更智能地记录下来，形成用户的偏好、习惯和生活状态。Muse的做法可以参考，记忆和凭证都放在用户专属的虚拟机里，智能体可以使用密码和支付信息，但看不到它们，用户也可以随时让它忘掉某些内容。

打造Agentic管家型助理而非被动式助理

打造能够主动帮助用户的产品，结合用户的记忆、日程、位置、消费记录和生活节奏，帮助用户端到端地完成需求，并基于用户的习惯和偏好，洞察用户潜在的需求。比如出差前一周提醒订票，能够直接并给出符合用户习惯的航班和酒店作为参考；常买的日用品快用完时能够提前下单，只让用户做最终确认动作，最大程度减少用户的重复信息录入和操作交互。

从 Muse 到管家型助理，个人助理 Agent 的竞争正在从任务式交付，转向生态、记忆与管家式服务的综合比拼。IDC 长期跟踪企业级与消费级 Agent 的能力演进，在《中国企业级通用 Agent 产品技术评估》中已建立覆盖记忆、任务执行、生态连接等维度的评估框架，并正在推进消费级 Agent 趋势洞察。对企业而言，IDC 可以帮助判断自身产品所处的能力阶段，识别生态壁垒与提升路径，评估生态合作与开放标准策略，并提供行业对标、趋势研判和定制化研究服务。无论您是 Agent 厂商、平台方还是服务生态伙伴，都欢迎与 IDC 分析师团队交流，获取评估框架、参与研究访谈或预订相关报告。

更多研究请参考：

IDC TechAssessment：中国企业级通用Agent产品技术评估报告#（Doc#CHC54893926 ,2026年9月）

IDC Perspective：消费级Agent趋势洞察（1Q27发布）

进一步交流

从 Muse 到管家型助理，个人助理 Agent 的关键门槛在生态、记忆与端到端任务能力。IDC 相关研究已覆盖企业级通用 Agent 评估，并正向消费级场景延伸。如您想了解所在产品的成熟度、生态壁垒与提升路径，欢迎联系 IDC 分析师团队，获取评估框架或预订相关报告。请点击此处与我们联系。

Zhenya Sun

Zhenya Sun - Research Manager

Analyst Profile
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…

大模型时代，应用与智能体正加速进入核心生产环节，行业对云基础设施的需求发生深刻变革。企业开始用推理效率、服务链路和合规能力来衡量基础设施，一个贯通算力、数据、模型服务与安全治理的AI云市场应运而生。国际数据公司（IDC）正式将这一市场定义为“AI云市场”，并首次发布《中国AI云市场追踪报告，2025》并给出中国AI云市场规模数据：2025年，推理需求集中释放，企业级MaaS的Token调用量一年增长约16倍，推动中国AI云市场规模达到684亿元。

定价锚点从“资源”转向“业务结果”，云基础设施需要重新定义

过去十年，云计算的定价锚是资源——核时、GB月、带宽峰值。企业采购的逻辑也很直接：比价格、比SLA、比资源池规模。这套锚点在大模型时代正在急剧松动。

根据IDC调研发现，企业决定是否采购AI基础设施时，评估标准正在经历跃迁，且呈现出清晰的递进关系：：

  • 资源层：算力够不够、Token成本多高、推理效率如何——这仍是最基础的门槛，但已不再是决定性因素；

平台层：模型效果好不好，数据、训练、推理能否串成一条完整链路，能否提供端到端的智能服务、为业务真正提效——这成为采购的核心考量；

  • 信任层：业务结果能否同时通过合规与审计，安全地进入核心生产系统——这决定了采购能否最终落地。

三个层次层层递进：资源是门槛，平台是价值，信任是放行条。可以看到，市场对基础设施的锚定已从“资源”转向“结果”——过往以售卖算力为主的云基础设施，难以满足大模型时代的企业需求，市场亟需新的产品形态。随之而来的，是市场结构与竞争壁垒的重新书写。

需求侧呈现“算力—数据—智能—信任”四位一体，AI云市场正在形成

那么，企业需要的“新形态”究竟长什么样？随着模型应用与智能体部署加速，需求侧的答案正日益清晰——一个“算力—数据—智能—信任”四位一体的基础设施栈：：

•算力：不只是GPU卡数，而是由硬件、调度、推理效率等共同支撑的服务吞吐；

•数据：通过向量数据库、其他AI原生数据库等合理高效的数据库架构，将企业知识和数据资产输入给AI；

•智能：机器学习平台（训推平台）与MaaS，决定大模型能否被规模化地生产、调用、治理，进而为业务注入“智能”；

•信任：AI安全，决定AI能否进入各行业的高价值核心场景。

四者缺一不可。任何一环的缺口，都会让另外三环的价值无法兑现。正是在这样的需求牵引下，一个区别于传统IaaS的AI云市场，正在加速成形。

IDC正式定义AI云市场

IDC认为，AI云市场是：以云服务形态交付、支撑大模型全生命周期的完整基础设施与平台服务体系。它向下整合专为AI工作负载设计的算力与存储资源，中游提供数据处理、模型开发、训练、部署与运维的完整工具链，并以贯穿全链路的安全与治理能力。区别于以通用IT负载为中心、按资源计价的传统云市场，AI云市场的价值锚点是“智能的规模化交付”，是让企业数据与模型能力转化为可信业务结果的完整技术栈。

IDC洞察：2025年推理需求成为主导，推动中国AI云市场达684亿元。

2025年训推拐点加速到来，推理场景在支出中的占比显著上升；大模型应用场景急速拓展，广泛覆盖泛互联网（游戏、娱乐、教育）、金融、智能制造等行业，迅速渗透至角色扮演、市场营销、搜索、数据处理、数据分析与文档处理等场景。

需求的爆发直接体现在调用量上：2025年，中国企业级MaaS市场Token调用量达到1994万亿，同比增长约16倍；日均Token消耗量从年初的1.6万亿攀升至年末的9.6万亿，达到年初的6倍。

需求侧的爆发与供给侧的快速响应形成合力——技术提供商加速投入资源部署，行业用户加速落地大模型及应用——共同推动2025年中国AI云市场规模达到684亿元。

IDC洞察：AI云竞争从“资源”转向“结果”，模型、推理优化与AI安全成为关键

IDC认为，AI云的竞争维度正在从”谁能提供资源”，转向”谁能把资源转化为可交付、可信赖的结果”。根据IDC调研，企业选择AI云基础设施时，最看重的因素为：回答质量/模型效果、性能、安全合规——恰好对应着三重竞争要素：

  • 模型是入场券，但入场券的有效期正越来越短。每一次新模型的发布，都会直接转化为AI云的增长动能：更强的模型能力能够解锁此前不可行的应用场景——长上下文、多模态、Agent编排，并催生新的推理需求；企业为评测、试用和接入新模型涌入平台，带来Token调用量的跳涨与客户迁移。AI云厂商若能自研或首发、独家引入领先模型，还能借此吸引开发者生态，形成短期获客与品牌溢价。但这张“入场券”的时效正越来越短：开源模型正在快速抹平能力代差，模型领先的窗口期已被压缩到以季度计——上一个版本带来的客户与流量，往往在下一次开源更新中被追平。模型能决定客户“来不来”，却决定不了客户“留不留”。
  • 算力资源与推理优化是长期壁垒。推理已成主战场，资源优势固然有用，但竞争焦点正从“谁有更多卡”转向“谁能把每一张卡用得更稳、更快”——壁垒正从“资源壁垒”变为“工程壁垒”。
  • AI安全从成本项变为溢价来源。AI安全不是合规部门的工作，而是产品架构的工作：数据不出域、内容安全、Agent权限治理、审计与溯源，必须在架构层解决。

在上述要素上建立起综合优势的厂商——阿里巴巴、字节跳动、百度等厂商取得了优势与主要市场份额。

IDC给厂商及行业用户的建议

对厂商

  • 算力获取决定竞争上限，但投入规模与节奏需与收益对赌。​中短期内，算力资源紧缺仍是主旋律，资源获取能力依旧是决定竞争上限的关键变量，需要长期、持续地投入。但厂商也必须清醒权衡：算力是重资产，投入大、回报周期长，且受供给波动影响——当前的资源投入带有一定风险，规模与节奏都需量力而行。    
  • 算力效率优化，是与扩容并列甚至优先的路径。​智能体与大模型应用爆发，需求端快速增长，但硬件供给的增速受限，“用好存量”因此成为关键。通过调度、推理优化与平台化软件能力，厂商可以在硬件不变的情况下提高服务吞吐——相比硬件扩容更具性价比。资源到位之后，真正把资源转化为可交付服务的，正是这些工程能力。。
  • MaaS资源应向性价比端倾斜。大模型效果持续提升，已经越过了生产环境可用的阈值，预期未来会出现各类智能体应用的大爆发。在这种场景下，客户对智能上限的要求大部分已被满足，将会更看重服务的经济性、稳定性，后者是在生产环境大规模使用智能体的前提。
  • 把AI安全前置为产品能力，而不是合规成本。安全是进入政务、金融、能源等高价值场景的准入资格，也是溢价能力的来源。
  • 为智能体重构基础设施。智能体的负载特征（长上下文、高频工具调用、高并发、强成本敏感）与传统AI负载不同，需要重做调度、缓存、可观测性与权限体系。

对行业用户

  • 升级采购标准：以业务结果指标替代算力指标。用任务成功率、单位业务成本、长尾时延来约束招标，而不是只用卡数与算力峰值。
  • 用总拥有成本而非Token单价做决策。Token单价低不等于总成本低，必须按自身场景的消耗特征测算。
  • 把安全与数据边界写进采购要求。对政务、金融、能源等行业用户而言，这不是加分项，而是必要条件。

分析师观点

IDC中国高级分析师洪婉婷表示，2025年，推理时代已然到来。684亿元的AI云市场规模与16倍的Token调用量增长，宣告云计算叙事正式换轨：定价锚点从“资源”转向“业务结果”。在新赛场中，真正的长期壁垒藏在推理优化、平台工程与架构级AI安全之中。对厂商而言，算力投入须与收益对赌，工程效率优先于盲目扩容；对用户而言，则应以任务成功率与总拥有成本，替代卡数与Token单价来做决策。谁能把“智能”转化为可信、可规模化交付的业务结果，谁就能在下一阶段的AI云竞争乃至大模型商业竞争中，掌握主导权。

进一步交流

如希望就AI云市场定义厂商份额及采购评估框架进一步交流，欢迎与我们联系。请点击此处与我们联系。

Catherine Hong

Catherine Hong - Senior Market Analyst

Analyst Profile
Ms. Catherine Hong is served as the role of Senior Market Analyst in Emerging Technology sector for IDC China, and her main responsibility is conducting analysis and research for the China Blockchain market. She is also in support of regional…

随着生成式AI接入企业核心业务，传统WAF和API网关已经盖不住Prompt攻击、数据泄露这些新风险，但市场上还没有一套统一的评估标准——国际数据公司（IDC）于近日发布首份《中国大模型应用防火墙技术评估，2026》（Doc#CHC53617126，2026年9月）报告，第一次对中国大模型应用防火墙产品技术进行系统性技术评估，报告从大模型应用防火墙产品的内容安全检测与防护能力、访问控制与策略管理能力、数据安全防护能力、安全运营与可观测能力、技术创新力以及开放性与生态能力六大维度入手，对360数字安全集团、安恒信息、阿里巴巴、长亭科技、Fortinet、火山引擎、华为、君同未来、绿盟科技、领信数科、奇安信、启明星辰集团、瑞数信息、天融信、腾讯15家厂商进行了评估，并给出行业平均标准。

IDC基于本次大模型应用防火墙技术评估结果的行业平均标准发现，目前中国大模型应用防火墙产品技术仍处在发展初期，多数产品能力主要聚焦在输入输出防护、内容安全检测、敏感信息识别和防护等基础检测与防护能力上，对于复杂AI应用场景的安全治理能力仍存在诸多不足。

IDC认为，随着生成式AI和智能体进入企业核心业务，大模型应用防火墙正从早期的提示词检测和内容过滤产品，逐步演进为覆盖模型、智能体、数据和运行时的AI原生安全平台。未来，产品将围绕AI驱动防护、全模态安全、统一治理、自适应防御等方向持续演进，成为企业AI基础设施的重要组成部分，产品具体的技术发展趋势如下：

  • AI Native安全架构——从”规则防护”迈向”AI对抗AI”时代：未来，大模型应用防火墙将立足AI Native安全架构，采用”大模型+小模型+规则引擎”的联合决策模式，借助AI技术提升AI安全检测和防护能力。
  • 安全检测和防护能力将从文本扩展至全模态场景：大模型应用防火墙将逐步支持全模态安全检测和防护能力，对文本、图片、PDF、Office 文档、网页、音视频等输入输出内容进行统一安全检测，并结合OCR识别、多模态语义 理解和跨模态关联分析，实现对隐藏Prompt、图片注入、跨模态攻击等风险的精准识别。同时，安全策略也将实现跨模态统一治理，为企业多模态AI应用提供一致的安全保障能力。
  • 行业化、场景化解决方案将成为产品竞争的重要方向：为了应对行业和场景的特殊需求，大模型应用防火墙将逐步向行业化、场景化方向发展，结合行业知识、业务流程和监管要求构建差异化的安全能力。
  • 安全防护与业务可用性的平衡将成为核心竞争力：大模型应用防火墙将从”一刀切”的静态拦截逐步演进为基于风险等级的动态安全治理体系，根据用户身份、业务场景、上下文语义以及风险评分，动态选择阻断、脱敏、重写、告警或人工确认等不同处置策略。同时，通过上下文关联分析、持续学习和策略优化，不断降低误报率和漏报率，在保障安全的同时最大程度提升AI应用的可用性和用户体验。动态的安全治理一方面需要技术服务提供商提供底层技术支撑，另一方面更需要产品服务团队和用户共同开发、共同成长，使产品越用越懂业务、越用越懂安全防护的边界。
  • 向统一安全控制平面快速演进：大模型应用防火墙将逐步突破模型输入输出防护的定位，向覆盖模型、数据、智能体和运行时的统一AI安全控制平台演进，成为企业AI基础设施中的核心安全控制节点。AI安全控制平面（AI Security Control Plane）将统一承担模型接入、安全策略管理、身份认证、访问控制、运行时安全、风险监测、安全审计和可观测等能力，并将智能体运行时安全纳入统一治理体系，实现对Agent身份、工具调用、MCP协议、多智能体协作及运行行为的全生命周期保护。

IDC中国网络安全市场研究经理王一汀表示，当前，大模型应用防火墙技术和市场仍处在发展初期，对于多轮上下文、工具调用、多模型协同等复杂AI应用场景的安全治理能力仍有待提升。未来，随着企业AI应用的规模化落地，大模型应用防火墙将逐步从单一的安全防护组件演进为面向AI应用全生命周期的统一控制平台，融合模型接入管理、身份权限控制、数据安全、智能体安全、风险运营与自动化响应等能力，保障用户AI的全生命周期安全。IDC也将持续关注相关技术和市场的发展情况。

备注：IDC定义的大模型应用防火墙是一种专门针对大语言模型（LLM）应用场景的安全防护技术与产品形态。其作为AI应用的安全边界防护层，部署在用户、应用与大模型交互的输入/输出通道，对请求与响应进行实时检测、过滤和阻断，从而防止恶意提示词注入、越权访问、敏感数据泄露和对抗攻击。

进一步交流

如您希望进一步了解本次评估的详细方法、各厂商能力表现，或探讨大模型应用防火墙在企业AI安全治理中的落地路径，欢迎与IDC中国网络安全市场研究团队联系。我们期待与您就报告内容及行业实践展开更多交流。请点击此处与我们联系。

Sophia Wang, CISSP

Sophia Wang, CISSP - Research Manager

Analyst Profile
Sophia Wang is a Research Manager in IDC China. She is responsible for the analysis and research of China's cybersecurity market. Her primary focus is on China's cybersecurity appliance and services market and operational technology (OT) security market. Additionally, she…

中国医院资源计划系统（HRP）市场正式从增量普及期迈入存量深化运营新阶段，市场增长动力由全新系统建设转向存量升级、模块扩容与智能化改造。在DRG/DIP支付改革、本土化替代、紧密型县域医共体建设等多重政策共振之下，市场迎来增长机遇，2025年HRP市场规模达到12.6亿元人民币。

国际数据公司（IDC）于近日首次发布了《中国医院资源计划系统市场份额，2025》报告，本次IDC分析了2025年医院资源计划系统市场中领先供应商的市场份额及市场表现，梳理了医院资源计划系统市场发展动态和趋势，可供各级各类医疗机构在选择医院资源计划系统供应商时参考。

报告的核心发现是：HRP市场的增长引擎已经换挡。全新一体化项目的拉动作用减弱，存量升级、下沉增量与技术重构成为三大新动力。这一判断贯穿全文，具体体现在需求分层、技术演进与竞争格局三个层面的变化之中。

需求分层：存量深耕成主线，分层市场释放差异化机会

随着三级医院HRP基础覆盖率达到高位，市场需求正在发生结构性变化。过去依靠大型新建项目拉动市场增长的模式逐步弱化，千万级全新一体化HRP项目数量减少，医院采购重点开始转向既有系统升级、模块扩展、数据治理和运营能力提升，同时向下沉市场扩面。不同层级医疗机构的诉求出现明显分化：

  • 三级医院的采购重点逐步从系统建设转向精细化运营和高质量发展，围绕系统互联互通、内控协同、运营分析和管理决策产生升级需求，ODR运营数据中心、AI智能应用以及功能模块迭代成为重要建设方向。
  • 二级医院及县级医院正在成为市场新的增量来源。在紧密型县域医共体建设政策驱动下，HRP的首次建设需求集中爆发，但该类用户对于预算敏感度较高，更倾向于标准化、轻量化和高性价比的产品与服务模式。
  • 医疗集团和医共体的扩张，则带动集团级统一管理的需求。多院区财务管理、集中采购、财务合并和经营数据汇总等需求不断增加。

整体来看，三级医院的存量升级迭代是稳固市场基本盘、挖掘存量价值的核心；而二级和县级医院、医共体医疗集团、民营医疗机构的新增建设需求，为行业增长带来了空间。2025年，存量深挖与增量拓展双向并行，共同构成HRP市场增长的两大核心来源。

技术演进：HRP向一体化、智能化、数据驱动的管理平台演进

2025年成为HRP行业技术变革的关键拐点，AI大模型规模化商用、本土化改造持续推进，一体化架构迭代升级，ODR运营数据中心价值持续放大，多重技术要素共同推动HRP从传统的业务管理系统向更加智能化、数据驱动的管理平台演进。

随着大模型部署成本下降和医疗场景适配能力提升，头部厂商纷纷搭建医疗运营专属AI能力中台，打造医疗垂直智能体，深度嵌入成本核算、财务审核、医保对账、发票稽核等业务场景，并将智能体与HRP系统融合，由此推动HRP从记录和管理业务的系统，逐步升级为可以理解业务、分析业务并参与业务执行的智能化助手。

一体化架构逐步成为HRP产品演进的重要方向。相比传统多系统拼接和接口集成模式，原生一体化架构能够进一步减少系统之间的数据割裂，并通过事件驱动、HTAP等技术增强业务协同和实时分析能力。与此同时，ODR运营数据中心的重要性进一步提升，其角色也从单纯的数据汇总平台向覆盖数据治理、统一标准和运营分析的核心数据底座演进。

竞争格局：HRP市场向头部集中，ODR 运营大数据成为核心增长引擎

随着市场进入存量竞争阶段，HRP行业竞争格局也在发生变化。目前市场参与者分化为三大类型：综合型ERP厂商、医疗信息化厂商、HRP垂直类厂商，不同厂商在AI能力、本土化适配、一体化架构、ODR数据中台、服务网络、行业场景积累上各有优劣，共同抢占市场份额。

2025年中国资源计划系统市场份额前五的厂商合计占比49.1%。在医院IT预算整体趋紧的背景下，头部厂商依托规模化的客户基础、成熟产品积累和服务网络，继而通过存量客户持续升级形成相对稳定的业务来源，取得了较高份额。其中，望海康信深耕HRP市场二十余年，其产品和品牌受到市场广泛认可，位列市场首位；同时，熙软科技、用友网络在HRP市场积累了稳定的客户资源，从而分列二、三。

HRP市场竞争的边界也正在向运营大数据领域延伸。随着医院管理从财务核算向DRG/DIP成本管控、绩效管理、资源效率分析等全域运营管理升级，运营大数据逐渐成为厂商的重要的业务方向。长期深耕HRP的厂商在业财融合、成本核算、绩效模型和运营管理场景方面积累更深，因此在运营大数据市场具备较强的先发优势。从市场格局来看，2025年运营大数据解决方案市场前五名的厂商合计占比为55%。市场仍呈现较明显的垂直化特征，基于其在HRP的业务经验和技术积累，望海康信在这一市场也占据了领先地位，而熙软科技和卫宁健康的运营大数据业务在近年来也获取了一定的市场份额。

未来市场竞争从单一产品功能对比，转向“AI技术平台+行业Know how+服务网络+生态整合”的综合实力较量。厂商能否在输出可量化价值的AI落地场景、高效完成本地化全栈适配交付、夯实ODR数据中台底座的同时，具备生态整合、资源协同能力，综合解决医院面临的实际经营问题，将是厂商拉开竞争差距的核心。

IDC中国高级市场分析师林红表示，2025年国内医院资源计划HRP行业整体保持小幅稳步增长，医院端需求升维，从财务、物资管理系统的基础建设阶段，转向DRG/DIP成本管控、集团医共体统一运营、全院运营数据治理、AI智能等模块及功能的建设阶段，为行业中长期增长打开空间。在未来竞争中，市场竞争转向AI运营智能体、全栈本土化替换、集团化多院区协同，以及专业化技术和咨询服务的综合能力比拼。未来一阶段，市场资源将持续向具备完整产品矩阵、全链路数智服务能力的头部厂商集中。

进一步交流

如希望就HRP市场定义、厂商份额数据、AI智能体与ODR数据中台的技术路径，以及医院采购评估框架进一步交流，欢迎与我们联系。IDC将持续跟踪中国医院资源计划系统市场的发展动态，为医疗机构选型与厂商策略提供参考。请点击此处与我们联系。

Erin Lin

Erin Lin - Senior Market Analyst

Analyst Profile
Erin Lin serves as a senior market analyst of the IDC China Health Insights group. She is responsible for conducting research and analysis about the health industry for both domestic and regional markets. She is also involved in consulting and…

For years, enterprise technology leaders across industries have operated under a flawed assumption: fast decision-making and rigorous compliance exist on opposite sides of a zero-sum game.

But in today’s rapidly changing tech landscape where AI is accelerating everything we do, the math no longer adds up. According to IDC research, AI agents are projected to grow 78x between 2025 and 2029. That’s a clear sign of the velocity of today’s market, and it isn’t slowing down. Yet many organizations report seeing no return on their generative AI investments, and IDC expects 40% to miss their 2026 AI goals.

Scale and speed of delivery are outpacing organizational confidence. Yet, tech leaders are attempting to address two fundamentally different challenges with the same default instincts: staying in control of risk while growing and having confidence that investments will pay off.

The underlying challenge: Speed vs. control

As a dual CIO and CISO, I’m responsible for driving our organization’s innovation initiatives, while ensuring we’re not exposed to risk as we grow.

When I talk to fellow executives in tech-centric roles, they express a variation on the same tension: business partners want AI adoption now, but compliance teams need to feel confident that the security and compliance measures exist. Both instincts are correct, but closing that gap with instincts instead of evidence exposes the enterprise to risks:

  • Rising threats: Insider-driven compromise is up 2x year over year.

  • Costly lapses: 38% of breached organizations ultimately paid a ransom.

These lapses occur when the rate of technology adoption outpaces an IT and security team’s ability to verify who has access to what and why.

Protecting the bottom line is part of the job description

Execution friction extends directly into spending. In fact, research shows that 67% of organizations are overspending on their AI agent budgets by more than 10%. Meanwhile, more than half of us are negotiating vendor deals without proper pricing benchmarks for this rapidly developing category.

There’s no doubt that advanced AI capabilities can accelerate innovation and productivity. But, as leaders and executives, we also have a responsibility to protect the bottom line. Balancing the two means bringing rigorous evidence to both sides of the equation: balancing speed and innovation with defensible risk and cost modeling.

Verified data is crucial

Navigating these enterprise shifts requires unlearning old habits, as we stop treating speed and control as sequential steps and start grounding both in verified data.

Historically, architecture and investment decisions relied heavily on vendor-provided materials. While any good leader carefully vets those inputs, that dynamic has the same limitations as a student grading their own test: the data has blind spots.

By embedding an independent, external view directly into everyday workflows to validate vendors and calculate ROI in real time, leadership can eliminate guesswork across both execution speed and risk mitigation.

Putting verified data to work with IDC Quanta

IDC Quanta for Tech Leaders solves the equation. A new AI-powered technology intelligence platform built specifically for CIOs, CTOs, Heads of AI, and other leaders driving critical technology transformation, IDC Quanta for Tech Leaders combines decision-grade answers, technology vendor selection capabilities, and price benchmarking, delivered inside tools your teams already use like email, Claude, and more.

Every response, vendor shortlist and pricing analysis is sourced and verified, drawing on thousands of published technology leader research documents, $500B in tracked tech market spend, and 15B+ proprietary IDC data points tracked annually. Its intelligence backed by experts that not only allows your teams to move faster on architecture decisions, vendor evaluations, and deal terms; it’s a foundation that will hold up under scrutiny.

As “client zero,” I look at a product like this twice: once as a CIO asking whether it helps my team move faster to support innovation, and once as a CISO asking whether I would approve it. Four things stand out:

  • Evidence I can trace -Every response and recommendation in IDC Quanta is fully cited, so each data point leads back to the research behind it. That gives me numbers I can defend in front of the board.

  • Context that stays mine -IDC Quanta securely cross-references internal context, such as my IT roadmaps, vendor contracts, and data sets, with IDC’s independent global market research. That means my teams spend less time synthesizing information and more time delivering for our partners across the business.

  • Security I can sign off on -IDC Quanta runs in isolated private workspaces with AES-256 encryption, 90-day automatic deletion, and a strict zero-model-training policy. Those are the controls my security review looks for first.

  • Intelligence embedded in my workflows -IDC intelligence is available inside tools teams already use, including email, AI assistants like Anthropic’s Claude, and more, so information sourcing and validation happens at the moment of the decision.

The next evolution of enterprise intelligence

Every organization faces the mandate to do more, in less time. The leaders pulling ahead have made verification part of how they move, with evidence behind each vendor decision and budget request before it reaches the board.

If that is a conversation you want to have, I’m glad to talk it over. You can also see how IDC Quanta for Tech Leaders works at idc.com/quanta/tech-leader/ or schedule a personalized demonstration.

Dave McKenna - Chief Information Officer & CISO - IDC

As Chief Information Officer for IDC, Dave McKenna oversees IT operations, cybersecurity, and the company’s enterprise AI automation and modernization initiatives. Leveraging deep expertise in data architecture, risk management, and strategic technology roadmaps, Dave plays a central role in aligning IDC’s technology capabilities with its broader business objectives.

In IDC analysts’ conversations with technology leaders, reevaluating their vendor ecosystem, controlling volatile AI costs, and proving measurable AI ROI are the same three challenges that keep coming up.

Yet depending on the organization, the outcomes could hardly be further apart. In IDC’s 2026 MaturityScape Benchmark, only 3.1% of organizations worldwide have reached the most mature stage of AI adoption. Meanwhile, 61.3% are still in the two least mature stages. That is roughly 20 organizations near the starting line for every one organization at the front.

Now, picture two technology leaders at the same table, with those same three decisions in front of them. One belongs to the 3.1%, building what the business becomes. The other belongs to the 61.3%, still managing technology. What is the difference between them? The decisions are the same for both. The difference is in how each leader navigates them, and neither has a playbook to follow.

That’s what IDC Quanta for Tech Leaders is built to solve.

Introducing IDC Quanta for Tech Leaders

Today, IDC is introducing IDC Quanta for Tech Leaders, an AI-powered technology intelligence platform with capabilities and packaging built specifically for CIOs, CTOs, Heads of AI, and other leaders driving critical technology transformation. It gives you independent, cited evidence to make each decision defensible. A new way to work with IDC intelligence, it combines decision-grade answers, vendor selection capabilities, and price benchmarking in one place, delivered inside tools you already use.

IDC Quanta responses are sourced and verified against more than 5,000 published technology leader research documents. They also draw on $500B in tracked tech market spend, and 15B+ proprietary IDC data points.

You work with IDC Quanta the way you would with a colleague, in email or with an AI assistant, like Claude. Ask, “Which vendors belong on my shortlist for an AI governance platform, and how do their prices compare to what peers are saying?” Securely add your own data or documents, and the answer comes back in the context of your situation, with citations. Time once spent searching and reconciling sources can be redirected to the strategy or investment itself.

Expanded IDC Quanta capabilities include a canvas for data exploration, a sourcing workspace, vendor shortlists you can request directly by email, and deal review that brings personalized pricing analysis into the same workflow. IDC Quanta can also turn findings into send-ready RFP responses and negotiation guides.

IDC Quanta runs in a private, secure workspace. Queries, documents, and outputs stay protected, with 90-day automatic deletion and zero model training on your data.

IDC Quanta is built to help tech leaders:

  • Chart their Strategy, Faster by grounding architecture, vendor evaluations, and deal terms in independently verified market intelligence that’s sourced and cited.

  • Win their Mandate, Confidently by bringing the case for urgency, opportunity, and readiness to critical technology initiatives, backed by evidence.

  • Prove their Impact, Grounded in Data when they’re able to more effectively select the right technology at the right price, then keep proving that investment is paying off with real governance models.

  • Work on their Terms with IDC Quanta embedded directly in existing tools like email, Claude, and more.

IDC Quanta on AWS Marketplace: A new way to buy, starting today

IDC Quanta is also now available through AWS Marketplace, removing friction from budget and procurement processes to give you a faster path to results. You can apply existing AWS budget commitments against your IDC spend, so sourcing and technology-selection work can begin without a new line item in a research or sourcing budget.

Learn more about using your existing AWS Marketplace budget to access IDC Quanta.

Who benefits most?

IDC Quanta can be used to support leaders such as:

  • CIOs weighing whether a platform they’ve run for years still holds up against what’s now possible.

  • Heads of AI Transformation who need to prove a roadmap to a board that’s grown skeptical after a year of AI spend without matching results.

  • CTOs defending a multi-year architecture bet on internal conviction alone, the kind that doesn’t survive a board post-mortem.

  • CISOs who must defend a security budget in a review without cited, analyst-backed evidence.

  • Procurement teams negotiating a renewal with no independent peer-deal benchmark, and, until now, no way to fund the process without new budget.

  • IT Governance leaders trying to keep a framework current against regulations that won’t stop moving, with no benchmark the audit committee will accept.

See more ways IDC Quanta can support your work – IDC Quanta Use Cases

Start building what your business becomes

Closing the distance between the leading cohort and everyone else happens one decision at a time. Each holds up better when it rests on independent, cited evidence.

The evidence already exists in IDC’s data: more than 60 years of market research and vendor intelligence, backed by technology spend data. IDC Quanta puts it inside the tools you already use, with citations attached, so you walk into the board meeting or the vendor renewal with the answer in hand, and the business can see you as the reason it can move with agility.

It’s time to start building what your business becomes.

See IDC Quanta for Tech Leaders

Jamie Fiorda - Senior Vice President of Product Marketing - IDC

Jamie Fiorda is Senior Vice President of Product Marketing at IDC, where he leads go-to-market strategy for IDC's research, data, and AI-enabled intelligence offerings.

The questions I asked about AI at 14 still hold up today: can a machine ask “Why? and How?”, and does it benefit society? Forty-two years ago, I wrote about AI in the non-fiction, math models section for Quest magazine, a high school literary compilation that tapped into creative talent from students at a Colorado high school. I was a bit of a nerd at the time, in 9th grade, a sysop of a BBS, running my Apple //e overnight to keep my board up and running. I finally upgraded the slow 1200 baud modem, which trickled text across the screen line by line, to 9600 baud the following year, but it still beeped horrendously when people called into my dad’s landline to connect (this was before mobile was a thing). AI was continuously hitting a wall at the time. Bayesian logic was primitive and computers weren’t as powerful as they are today. I had a part-time job as a teaching assistant at a local community college and helped the students (and professor) understand Beginner’s All-Purpose Symbolic Instruction Code (BASIC), something I was well versed in, given my 3 to 5 hours per day on the machine. I was a fanatic with 6502 assembler, and later moved up to 8086 machine language. To be clear, machine language in those days had a different meaning from today’s ML. It was based on hexadecimal code using memory locations and stacks to transfer data from point to point in a large memory packet (large being 64 kilobytes).

My AI essay was prescient and outlined some of the fundamental aspects of AI at the time. Below are some excerpts from the essay and how a nerdy 14-year-old thought about AI in 1984. Word processors in those days weren’t usually on computers; they were electric typewriters with memory, so the essay was typed up in standard Courier font, which I try to replicate below. I’ve left out some parts of the text, and my grammar and writing capability were limited at best, as I was more of a math geek.

Artificial Intelligence, Fact or Fiction?

“

Intelligence is a broad and diverse subject. Mankind has been trying to simulate this for many decades. The study of Artificial Intelligence (AI) is being conducted at some of the best universities and schools in the world. Artificial Intelligence must first be defined before someone may actual go ahead and start trying to create it via computer, analog devices, chemically, or any other means. AI can be define as “fake knowledge”, yet that has already been created through computers which can perform functions such as compute the probable outcome of a baseball game or can predict the weather fairly accurately. AI can also be defined as deductive reasoning, which is the ability to make a reasonable decision out of a set number of variables. Computers have been able to simulate this and actually perform this as well, so this is not what researchers are looking for. A good definition for AI is the ability to think and to specifically ask the questions “Why? or How?”. This form of AI has been the main concern of modern researchers…

…The main type of AI research are ones using computers and digital electronics. Computers use switches to perform their functions. The computers simply check if a certain circuit has power or not. Doing this they can be programmed to [do] almost anything. A limitation includes that of not being able to do something without command of an user. This is the actual goal AI researchers are trying to accomplish. Once this goal is achieved, a computer’s power will be limitless. One thing scientists have tried to do is to create a program which can create other programs, as if it was learning as it was growing from child to man. It could develop itself slowly by mimicking human behavior as small children do, and by getting information from its environment…Computers are able to comprehend spoken language and to recognize photographic patterns…With guidance however, it is likely that computers will eventually be able to ask the questions “Why? and How?”. Digital electronic switches have been used extensively for attempting to get results of predictions…

AI has been worked on extensively. It is steadily progressing, although a little slowly. It will not be finished in our lifetime, but in the future it may be accomplished…Intelligence is much more than simple knowledge. It is memory, intuition, consciousness, and many other unexplainable factors combined. AI needs some thought on why we humans do need it. A few such reasons are: teaching ability, quicker thinking and research (to help answer questions unknown to us) and most importantly to decrease the strain of unnecessary mental tasks so that humans can challenge greater conquests without wasting time. AI research will prove to be a challenging subject for humans in the future. AI still needs much work, not only in digital electronics, but also in psychology, cephalogy, and other medical professions before the ultimate goal, a complete understanding of AI, can be accomplished.

— Ananda Chakravarty, ca 1984

So this brings us to two questions: “So what does this history lesson matter today?” and “Can you glaze your childhood anymore?” AI has expanded well beyond the traditional capabilities the essay touched on, which included non-digital AI, ‘fake knowledge’, predictive algorithms, deductive reasoning and simulation. The essay opens with thinking and the ability to ask “Why? and How?” Its next section covers autonomous action, a computer performing actions without a user, and suggests this will be an AI superpower, in line with agentic AI today. Then comes the ability for AI to create other programs, the basic capability of recursive self-improvement, followed by mimicry, the way a child learns, which is what all AI tools do today in training (reinforcement learning). That section also notes what computers could already do in the 80s: basic NLP, pattern recognition and predictions. The last section turns to why AI matters and what it means for society. Its main reason is freeing humans to achieve greater conquests. The essay also expects the work to take a long time and defines intelligence as memory, intuition, consciousness and “many other unexplainable factors” combined.

In the 80s, the concepts for AI were straightforward: help improve society. Today a new culture has arrived, with its own jargon and ways of thinking, and much of the discussion turns on regulation, guardrails, execution checks and ethical boundaries, little of which the essay covered.

AI capabilities keep growing. Stanford University’s AI Index, as covered in IEEE Spectrum, outlines how quickly AI has matched or exceeded human performance on a growing number of benchmarks (see Figure 1).

Figure 1. Select AI Index technical performance benchmarks compared to human performance, 2012 to 2025. Source: Stanford University AI Index, 2026, as covered in IEEE Spectrum.

Today’s AI tools can read, understand visual diagrams, compete mathematically, offer logical reasoning, classify images, multitask, autonomously execute complex tasks and understand scientific principles, and on a number of benchmarks they match or exceed humans at comprehending writing. AI has moved faster than the essay anticipated, and the pace has caused confusion about what it means for society. Parts of the debate now turn to security, warfighting, hacking and theories of human extinction. Science fiction writer Isaac Asimov would find the concerns familiar. The fundamental questions have not changed, and even a 14-year-old would ask them. Should AI aim to benefit society rather than generate profit at scale for individual companies and organizations? That question sits at the center of today’s competitive AI race. Across 42 years, the constants have been the ethical and societal impacts of AI. The essay’s goal also still holds: an autonomous AI that asks “Why? and How?”, and humans free to “challenge greater conquests without wasting time.”

Learn more about AI in retail by reaching out to the IDC Retail Insights team or IDC Quanta to find the underlying strategies, details and statistics that enable AI today and will shape what comes next.

Ananda Chakravarty

Ananda Chakravarty - Research VP, Retail Merchandising and Marketing Analytics Strategies

Analyst Profile
Ananda Chakravarty is Vice President for IDC Retail Insights, responsible for the Retail Merchandising and Marketing Analytics Strategies practice. Mr. Chakravarty’s core research covers in-store and digital retail merchandising, digital tools, artificial intelligence, intelligent store operations, retail marketing, and retail…

I called the death of the SaaS UI “wonderful” two weeks ago in my prior blog, and I stand by it. But a death this size leaves something behind: real estate. Somebody inherits the land the UI vacates, and that fight is already underway.

Salesforce didn’t just retire its own interface with AIforce. It handed the keys to Claude, Slack, and a dedicated coworker agent, surfaces that sit outside the traditional Salesforce screen. That is the tell. There’s plenty more to come, and I predict Microsoft Office isn’t far behind. The moment data, workflows, and business logic stop living inside a vendor’s own screen, they have to live somewhere, and whoever owns that somewhere will become a much more valuable piece of enterprise software, one that is likely still undervalued today.

The list of landlords will grow

Microsoft Teams and Office. Slack. Claude. ChatGPT. These are just four of many vendors that already sit inside the daily workflow of hundreds of millions of knowledge workers, and that positioning is not an accident. That positioning will increasingly become the entire strategy. Every SaaS vendor that goes headless is, functionally speaking, applying to be a tenant inside someone else’s building. The landlord collects the attention, the habit, and eventually the data exhaust (metadata such as which workflows get invoked, how often, in what sequence, by which roles, and at what time of day).

Meta just decided it wants to be a landlord too. Meta Enterprise Platform, launched September 28, bundles Muse, Meta Business Agent, Muse API, and Muse Code. Meta also extended Muse for Small Business, adding direct connections to QuickBooks, Shopify, Slack, and Stripe. Meta is aiming to become the surface where those vendors’ workflows show up, and it’s targeting a customer base in which, by Meta’s count, more than a million businesses already run Business Agent on WhatsApp and Messenger. That is its fastest path to scale.

This is not a new war: It is an old one with a new battlefield

We have fought this exact fight before, just with a different asset underneath it. Browsers fought over who owned the gateway to the internet. Mobile operating systems fought over who owned the gateway to apps. Now the fight is over who owns the gateway to agentic work itself, and IDC’s research on this question is clear: the next platform war will not be won by whoever builds the best agent. It will be won by whoever controls where humans and agents actually meet.

What this means for the rest of the market

Every enterprise software vendor just got a second, harder decision to make. The first decision, go headless, is basically settled. Vendors who try to fight this direction will face a difficult road. The second decision, which surface to show up inside and on what terms, is wide open, and it will matter far more than the first one.

App vendors shouldn’t waste time. Negotiate distribution terms with these UI surfaces now, while there are still several competing for their workflows. That leverage will shrink over time as UI surface “winners” emerge and competition consolidates. Vendors should also build for multiple surfaces simultaneously. Diversifying across surfaces, since no one yet knows which landlord will own the best real estate, is a sound strategy.

The SaaS UI is dying easily because nobody feels truly attached to it. The fight over what replaces it will not be as easy because for the first time in thirty years, enterprise software vendors do not control their own front doors.

Eric Newmark

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

Analyst Profile
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…