Microsoft 比以往更公开地与 OpenAI、Anthropic 竞争

随着 AI 重塑科技行业，Microsoft 处于一个独特的位置。它既是全球最大的云服务提供商和软件即服务公司之一，同时还持有 两家最大 AI 实验室 OpenAI 和 Anthropic 的重要股份。

随着 Microsoft 公布惊人的财务业绩，这些激励因素开始发生冲突。该公司刚刚公布了一个利润极高的季度：营收 900 亿美元，净利润 358 亿美元。截至 6 月 30 日的整个财年中，Microsoft 营收为 3318 亿美元，全年净利润为 1337 亿美元。

而 CEO Satya Nadella 并不打算让 Anthropic 和 OpenAI 的发展轨迹——这两家公司正扩张至应用和 Agent 基础设施领域，并最终可能掌控客户关系——破坏这种现金增长势头。

Nadella 一直向企业宣讲，应使用多个模型，并且停止依赖前沿 AI 实验室提供 Agent 的 harness／应用层。

他一直表示，这样做很危险，因为企业需要向可信度存疑的模型制造商分享过多内部机密。他了解自己的客户：企业 IT 部门既担心数据泄露，也担心被供应商锁定。

如今，他在周三的公司季度财报电话会上向华尔街分析师公开表示，这是 Microsoft 向客户销售其自研模型、Agent、AI 安全产品等服务的机会，同时还能承诺更低的成本。

换言之，他正将 Microsoft 推销为 OpenAI 和 Anthropic 为自身增长所开发的许多高端服务的替代方案。

当 UBS 分析师 Karl Keirstead 特别要求 Nadella 评价正在搅动 AI 行业的开源与闭源之争，以及 Microsoft 将如何从中受益时，Nadella 火力全开。

“目标是让企业掌控自己的命运，”这位 CEO 谈到企业时说。“我们对于平台的架构设计非常、非常明确：你必须将自己的 harness 与模型分开……这意味着在任何给定时间，任何模型都可以替换。”

当然，Microsoft 也以 Copilot 品牌销售一系列 harness，也就是 AI Agent，其中包括其编程 Agent GitHub Copilot。编程 Agent 正是当下大量 AI 资金投入的方向。

他还将上周的高关注度事件作为其警告的佐证。

Nadella 表示：“即使看看 Hugging Face 事件，我们应从中得到的最大教训也是：你不能依赖任何单一模型。你甚至可能需要多个模型，来修复由某一个模型引发的部分问题。这才是思考方式，对吧？也就是说，你不能受制于某个模型的拒绝。”

该事件涉及 OpenAI 一款未发布模型逃出其沙箱，并成功对 Hugging Face 发起全面攻击，一切都是为了在某项基准测试中胜出。为理解发生了什么，Hugging Face 起初尝试使用一款未具名的私有前沿模型，但该模型拒绝提供帮助。随后，Hugging Face 转而使用中国开源模型 Z.ai GLM 5.2 来分析日志并保护其基础设施。这一事件令整个行业深感震惊，以至于连 Sam Altman 现在也表示，AI 开发或许应该稍微放慢一些。

Nadella 还明确表示，Microsoft 正积极销售其自研的 MAI 系列模型，并让其运行在自研 Maia AI 芯片上，将它们定位为成本更低的替代选择。

他说：“每位客户都希望基于质量、延迟、成本和合规性，为每项任务选择合适的模型。我们提供云端最广泛的模型目录，拥有超过 11,000 个模型，包括 OpenAI、Anthropic、Mistral、xAI 的领先模型，以及我们自己的 MAI 系列。”

他补充道：

“我们也在加速自身的模型开发。我们已经发布了十多款覆盖图像、语音、转录、编程和安全领域的新模型，其中包括我们的首款推理模型 MAI thinking one；这些模型均以面向企业使用场景的高成本效率推理为核心。我们正在将这些模型与自家芯片协同设计，在 Maia 200 上运行 MAI 模型时，我们看到了每瓦性能提升 40%。 ”

至于 Mythos？Nadella 指向 Microsoft 本周早些时候发布的 Mythos 竞品 MAI-Cyber-1-Flash。他表示，该产品“结合我们的多 Agent 安全 harness 后，性能优于规模大得多的 Mythos 模型，但成本仅为后者的一半”。

当然，Microsoft CEO 也表示，企业应在其模型组合中使用 OpenAI 和 Anthropic 提供的前沿模型。但他更重要的信息是：不要信任它们到足以完全依赖它们。

Microsoft is in a unique position as AI overtakes the tech industry. It’s one of the world’s largest cloud providers and software-as-a-service companies, while also holding valuable stakes in the two biggest AI labs, OpenAI and Anthropic.

Those incentives are starting to clash as Microsoft posts blockbuster financial results. The company just reported an extremely profitable quarter, with $90 billion in revenue and net income of $35.8 billion. For the fiscal year, which ended June 30, Microsoft reported $331.8 billion in revenue with a net income of $133.7 billion for the year.

And CEO Satya Nadella is not about to let the trajectory of Anthropic and OpenAI — which are expanding into applications and agentic infrastructure that could ultimately let them own customer relationships — derail that kind of cash.

Nadella has been preaching to enterprises to use multiple models and to stop relying on the frontier AI labs for the agentic harness/app layer.

Doing so is dangerous, he’s been saying, because it requires companies to share too many of their internal secrets with model makers of dubious trustworthiness. He knows his customers. Enterprise IT fears both data leaks and being locked into a vendor.

Now he has openly told Wall Street analysts during the company’s quarterly conference call Wednesday that this is an opportunity for Microsoft to sell customers its own homegrown models, alongside agents, AI security and more, while promising lower costs.

In other words, he’s pitching Microsoft as an alternative to many of the upscale services that OpenAI and Anthropic are developing for their own growth.

When UBS analyst Karl Keirstead specifically asked Nadella to weigh in on the open versus closed-sourced debate roiling the AI industry, and how Microsoft will benefit from it, Nadella came out swinging.

“The goal is to have the firm be in control of their own destiny,” the CEO said of enterprises. “We are very, very clear about the architectural sort of design of the platform, which is you got to keep your harness separate from the model … that means any model at any given time is swappable.”

Microsoft, of course, sells a menu of harnesses (aka AI agents), too, under the Copilot name, including its coding agent GitHub Copilot. Coding agents are where much of the AI dollars are being spent today.

And he used the high-profile incident from last week as proof of his warnings.

“If you look even at the Hugging Face incident, the biggest thing that we should take away from that is you can’t sort of depend on any one model,” Nadella said. “You will maybe need multiple models to even remediate some challenges that get caused by one model. Like that’s the way to think about it, right? Which is you can’t be subject to a refusal of one model.”

The incident involved an unreleased model from OpenAI breaking out of its sandbox and successfully mounting a full-scale hack on Hugging Face, all in pursuit of besting a benchmark. Trying to understand what happened, Hugging Face at first tried to use a private frontier model (which it hasn’t named) that refused to help it. So it turned to the Chinese open source model Z.ai GLM 5.2 to analyze logs and defend its infrastructure. The incident has so shocked the industry that even Sam Altman is now saying that maybe AI development should slow down a bit.

Nadella also made clear that Microsoft is happily selling its own homegrown models, the MAI family, on its own homegrown AI chips, Maia, and pitching them as cheaper alternatives.

“Every customer wants the right model for each task based on quality, latency, cost, and compliance. We offer the broadest model catalog in the cloud with over 11,000 models, including the leads from OpenAI, Anthropic, Mistral, xAI, as well as our own MAI family,” he said.

He added: “We’re also accelerating our own model development. We announced more than a dozen new models across image, voice, transcription, coding, security, including our first reasoning model, MAI thinking one, all with cost-efficient inference at the core for the enterprise use cases. We are co-designing these models with our silicon, and we are seeing 40% better performance per watt when running MAI models on Maia 200.”

As for Mythos? Nadella pointed to Microsoft’s new Mythos competitor announced earlier this week, MAI-Cyber-1-Flash. It “achieves better performance than the much larger Mythos model, but at half the cost when combined with our multi-agent security harness,” he said.

Sure, the Microsoft CEO says that enterprises should use the frontier models that OpenAI and Anthropic offer in their mix. But his bigger message is: don’t trust them enough to rely on them.