内容来源：https://aibusiness.com/generative-ai/how-mistral-s-new-funding-bridge-sovereign-ai

内容总结：

法国AI初创公司Mistral日前宣布完成由三星电子领投的D轮融资，募资35亿美元，公司估值已突破240亿美元。这家成立仅三年的巴黎企业凭借此轮融资进一步巩固了其作为欧洲领先AI供应商的地位，并持续推进其主权AI与开放权重战略。

除投资外，三星电子还与Mistral达成合作伙伴关系，将利用其模型和服务开发专用半导体。分析人士指出，三星的投资表明Mistral正朝着为特定行业构建专用模型的方向发展，尤其是半导体设计与制造领域。

在融资规模和市场估值上，Mistral与美国及中国的竞争对手相比仍有差距。Anthropic融资1300亿美元，估值达9650亿美元；OpenAI融资1900亿美元，估值8520亿美元。中国初创企业月之暗面估值500亿美元，DeepSeek估值745亿美元，增速同样迅猛。

然而，许多欧洲企业愈发重视“主权AI”——即利用本国或本地区自身的基础设施、数据、模型和人才来构建、运行和管理AI系统。Mistral自创立之初便与主权AI理念结盟，最初专注于开发小型开放权重模型，如今已发展为同时提供AI模型、基础设施和平台服务的综合供应商。

研究机构Gartner分析师阿伦·钱德拉塞卡兰表示，Mistral赢得的新业务主要得益于其AI主权定位和欧洲模型供应商的身份。他补充说，主权AI虽是Mistral的核心优势，但该公司也在借助新一轮融资拓展东南亚等新市场。

据悉，这笔资金将用于持续投资基础设施。此前Mistral已投入14亿美元在瑞典建设大型AI数据中心，并获取8.3亿美元债务融资用于巴黎数据中心建设。Mistral表示，除基础设施外，还将利用这笔资金打造面向企业的AI平台。

Futurum集团分析师尼克·帕蒂恩斯指出，Mistral与三星的合作表明其战略方向，即为特定行业场景构建模型。不过他也提醒，尽管Mistral持续向开源和主权AI方向转型，但如何从这一战略中实现营收仍是其面临的主要挑战。大多数企业客户并不关心公司估值，更看重其交付能力和稳定性。

中文翻译：

由谷歌云赞助
选择您的首批生成式AI应用场景
要开始使用生成式AI，首先要聚焦于能够改善人类与信息交互体验的领域。
起初以开放权重初创公司起家的Mistral，鉴于其所处的欧洲市场，已转向主权AI战略。
法国AI供应商Mistral周二宣布，在由三星电子领投的D轮融资中筹集了35亿美元后，公司估值现已超过240亿美元。此轮融资到位之际，这家欧洲供应商正继续推进其主权AI和开放权重战略。
除了投资Mistral之外，韩国电子巨头三星还与这家AI供应商达成合作，将使用其模型和服务来开发专用半导体。
尽管Mistral AI成立仅三年，这家总部位于巴黎的初创公司已赢得了欧洲领先AI供应商的声誉，即便欧洲在AI竞赛中明显落后于美国和中国。其美国竞争对手同样也是初创公司，但在融资和销售方面已将Mistral远远甩在身后——Anthropic已融资1300亿美元，估值达9650亿美元；OpenAI融资1900亿美元，估值达8520亿美元。就连更晚成立的中国初创公司也增长更快，月之暗面目前估值500亿美元，DeepSeek估值745亿美元。
然而，许多欧洲企业重视主权AI——这一日益流行的理念主张国家、地区或组织利用自己的基础设施、数据、模型和人才来构建、运行和管理AI。
Mistral从成立之初就与主权AI结盟，但最初专注于构建小型开放权重模型。如今它已演进为在提供AI模型之外，优先发展主权AI和基础设施。作为欧洲顶尖AI公司，它对那些在意数据去向以及AI运营地点和方式的企业颇具吸引力。
“Mistral目前赢得的新业务主要归功于其AI主权性以及他们是欧洲模型供应商这一事实，”高德纳分析师阿伦·钱德拉塞卡兰表示。“主权性是他们在充分利用的优势。”
他还补充说，虽然聚焦主权AI对Mistral行之有效，但这家供应商也在寻求利用新资金拓展其他领域，并将东南亚视为机遇。
这笔资金将使该供应商能够继续投资于基础设施。此前，Mistral已斥资14亿美元在瑞典建设了一座大型AI数据中心，还获得了8.3亿美元的债务融资用于资助巴黎的一处数据中心。除了基础设施之外，Mistral表示将利用这笔资金打造面向企业的AI平台。
“他们意识到，要想取得成功，就必须成为能够将基础设施、模型和平台工具打包在一起的服务提供商，”钱德拉塞卡兰说。
除三星外，Scaleup欧洲基金和波士顿的PSG股权投资公司也是本轮融资的投资者。三星这样的大型跨国企业为Mistral背书这一事实也值得关注，因为它表明了Mistral的前进方向，Futurum集团分析师尼克·佩申斯表示。
“关键在于为行业特定应用场景构建模型，”佩申斯说，并指出三星对该公司的投资与其为半导体设计和制造专门构建模型有关。
三星不仅是投资者，也是这家法国初创公司的合作伙伴。通过新的合作关系，三星将把Mistral的模型和服务整合到其半导体业务中。该公司还将使用Mistral模型来设计定制AI系统，以协助专用半导体的设计和制造。
然而，尽管Mistral继续将其战略转向开源和AI主权，其主要挑战在于如何从该战略中获得收入，佩申斯表示。
他还补充说，大多数企业买家并不关心公司的估值，而更关注它能否兑现承诺并保持稳定。

英文来源：

Sponsored by Google Cloud
Choosing Your First Generative AI Use Cases
To get started with generative AI, first focus on areas that can improve human experiences with information.
After starting as an open-weight startup, Mistral has shifted toward sovereign AI, given the European market it operates in.
French AI vendor Mistral said on Tuesday that it is now worth more than $24 billion after raising $3.5 billion in a Series D round led by Samsung Electronics. The funding comes as the European vendor continues to expand its sovereign AI and open-weight strategies.
In addition to investing in Mistral, South Korean electronics giant Samsung is partnering with the AI vendor to use its models and services to develop specialized semiconductors.
While Mistral AI is only three years old, the Paris-based startup has earned a reputation as Europe’s leading AI vendor, even as Europe is significantly behind the U.S. and China in the AI race. Its U.S. rivals, also startups, have dwarfed Mistral in fundraising and sales, with Anthropic raising $130 billion at a $965 billion valuation and OpenAI raising $190 billion at an $852 billion valuation. Even more recent Chinese startups have grown faster, with Moonshot AI now valued at $50 billion and DeepSeek at $74.5 billion.
However, many European enterprises value sovereign AI, the increasingly popular approach that enables a country, region, or organization to build, run, and govern AI using its own infrastructure, data, models, and talent.
Mistral, from its inception, allied itself with sovereign AI but focused first on building small open-weight models. It has now evolved to prioritize sovereign AI and infrastructure in addition to AI models. And, as the top AI company in Europe, it appeals to enterprises there that are conscious of where their data is going and where and how their AI operations are taking place.
“The new business that [Mistral is] winning is primarily because of their AI sovereignty and the fact that they are a European model provider,” said Arun Chandrasekaran, an analyst at Gartner. “The sovereignty is something that they are taking advantage of.”
He added that while the focus on sovereign AI works for Mistral, the vendor is also looking to expand to other areas with its new funding and sees Southeast Asia as an opportunity.
The funding will enable the vendor to continue investing in infrastructure. Previously, Mistral spent $1.4 billion to build a major AI data center in Sweden. It also secured $830 million in debt financing to fund a data center in Paris. Beyond infrastructure, Mistral said it will use the money to become an AI platform for enterprises.
“They realize that for them to be successful, they have to be a platform provider that can package infrastructure, models and platform tools together,” Chandrasekaran said.
Aside from Samsung, the Scaleup Europe Fund and Boston-based PSG Equity were investors in the funding round. The fact that Mistral is backed by a giant multinational like Samsung is also noteworthy, as it indicates the direction Mistral is headed, said Nick Patience, an analyst at Futurum Group.
“It’s about building models for industry-specific use cases,” Patience said, noting that Samsung’s investment in the company is tied to building models specifically for semiconductor design and manufacturing.
Samsung is not only an investor but also a partner with the French startup. With the new partnership, Samsung will integrate Mistral’s models and services across its semiconductor operations. The company will also use Mistral models to design custom AI systems that help with the design and manufacturing of specialized semiconductors.
However, while Mistral continues to shift its strategy toward open source and AI sovereignty, its main challenge is generating revenue from that strategy, Patience said.
He added that most enterprise buyers don’t care about a company's valuation but are more concerned about whether it can deliver and remain stable.