Clarifying the IPO Speculation
Arthur Mensch, CEO of the Paris-based AI startup Mistral AI, has definitively addressed recent rumors surrounding a potential initial public offering (IPO). In an exclusive interview with Fortune at Nvidia’s GTC conference, Mensch clarified previous statements, putting to rest the speculation that an IPO was imminent. The initial buzz stemmed from a January interview with Bloomberg at Davos, where Mensch had emphasized Mistral’s commitment to remaining independent, stating that an IPO was “the plan.”
Mensch explained his earlier remarks to Fortune, stating, “I was asked about our future, and I said that we intend to remain an independent company, [so] the natural path is to get to an IPO at some point. Just to clarify, we’re not looking towards an IPO [right now].” This clarification underscores Mistral AI’s current focus on growth and solidifying its market position, rather than immediate plans for public listing.
Mistral AI burst onto the AI scene less than two years ago, founded by veterans of Google DeepMind and Meta’s AI research lab. The company quickly made headlines by securing $113 million in seed funding after only a few months in stealth mode, a record-breaking amount for a European seed round. Its first AI model, Mixtral 8x7B, released in March 2024, garnered significant attention and praise for its innovative design and strong performance.
The Challenge of Competing with AI Titans
Since the release of Mixtral 8x7B, industry analysts and observers have questioned whether the AI foundation model company has the resources and staying power to compete effectively against well-established and heavily funded rivals such as OpenAI and Anthropic. These competitors, along with tech giants like Meta, Google DeepMind, and Microsoft, possess significantly larger war chests. While Mistral has secured a respectable $1 billion in funding to date, including a recent $640 million Series B round that valued the company at $6 billion, this figure is dwarfed by the $18 billion raised by OpenAI (with Softbank potentially investing another $40 billion) and Anthropic’s $8 billion.
The AI landscape is characterized by a relentless need for capital. Staying at the forefront of general-purpose AI model development requires massive investments in computing power (primarily GPUs) and the recruitment and retention of top-tier AI talent. Furthermore, any performance advantage gained by a new model is often quickly commoditized, making it challenging for even the best-funded startups to maintain a long-term competitive edge. This raises questions about the long-term profitability of even the largest players, let alone smaller companies like Mistral.
Another significant challenge for Mistral is its lack of a Big Tech benefactor. Unlike OpenAI (backed by Microsoft) or Anthropic (backed by Google and Amazon), Mistral does not have a guaranteed, sustained access to the thousands of GPUs necessary for training and running large AI models. This reliance on external providers for crucial infrastructure could potentially limit Mistral’s ability to scale and innovate at the same pace as its competitors.
Mistral’s Open-Source Commitment: A Differentiating Factor
Mistral has strategically positioned itself as a champion of open-source AI, a key differentiator in a market increasingly dominated by proprietary models. Unlike the closed systems of OpenAI and Anthropic, where users interact with the AI model solely through an application programming interface (API), Mistral, along with other “open weight” companies, allows users to download the core “brain” of the AI model and the code for running it. This approach fosters transparency, collaboration, and customization, appealing to a segment of the market that values these principles.
However, a 2024 partnership with Microsoft raised concerns that Mistral was deviating from its open-source commitment. As part of the partnership, Mistral released several closed, proprietary models and made them available on Microsoft’s Azure cloud service. This move sparked criticism from some within the open-source community, who questioned Mistral’s dedication to its founding principles.
Mensch, however, has strongly reaffirmed Mistral’s unwavering commitment to open source. He emphasizes that the company is pursuing a dual strategy: generating revenue through premium proprietary models, an AI infrastructure platform called “Le Plateforme,” and pro subscriptions to “Le Chat,” the company’s AI assistant, while simultaneously continuing to release and support open-source models.
A new billboard on Highway 101 in Silicon Valley, a major thoroughfare in the heart of the tech industry, subtly highlights Mistral’s commitment, stating that most of its models are “actually” open. This is a direct jab at Meta, arguably the most prominent company offering free, open-weight models. Mensch points out that Meta’s models have more restrictive licensing terms than Mistral’s Apache 2.0 license, which grants users greater freedom to use, modify, and distribute the models.
It’s important to note, however, that the debate around “openness” in AI is complex. Almost none of the companies offering open-weight models, including Mistral, disclose the datasets used to train their models. This lack of transparency has drawn criticism from open-source purists, who argue that without knowing the training data, the models cannot be truly considered “open source.” The argument is that the training data can contain biases or copyrighted material, and without access to this information, users cannot fully understand or audit the model’s behavior.
Mensch addressed this criticism, stating, “It’s as open as you can be. We share the weights, we share the inference, we share a lot of findings around how we built it. There’s obviously some trade secrets that we keep, because how we bring our core value to work with customers.” This statement acknowledges the inherent tension between open-source principles and the need to protect proprietary information that gives Mistral a competitive edge.
The Enterprise Strategy: Focusing on Business Clients
Mistral’s strategy extends beyond the open-source community. The company has been actively pursuing enterprise clients, recognizing the significant revenue potential in providing AI solutions to businesses. Notable clients include Axa, Mars, and Cisco, representing the culmination of an 18-month effort to penetrate the competitive enterprise market.
These clients pay for access to “Le Plateforme,” Mistral’s AI infrastructure platform. This platform provides access to both Mistral’s open and closed-source models, along with a suite of AI tools and infrastructure designed to empower technical teams to build, customize, and deploy AI across their organizations. This offering caters to the growing demand from businesses to integrate AI into their operations, but who may lack the internal expertise or resources to develop their own AI solutions from scratch.
Mensch revealed to Fortune that Mistral’s revenue has experienced a remarkable 25-fold increase over the past year. While he declined to disclose the specific current sales figure or the baseline from which this growth originated, this rapid expansion indicates strong market traction and successful adoption of Mistral’s offerings.
Concurrent with its revenue surge, Mistral has significantly expanded its headcount and geographic presence. From a small team of a few dozen employees, primarily based in Paris a year ago, Mistral now boasts a workforce of 200, including 60 researchers. The company has offices in Paris, London, San Francisco, and a newly established outpost in Singapore, reflecting its global ambitions.
Mensch acknowledged that he has had to adapt to his role as CEO as the company has scaled. “For four or five months, I was still coding and doing science,” he admitted. “Now, I’m mostly focused on sales and product.” This transition reflects the changing demands of leadership as a startup matures and requires more focus on business development and strategic direction.
Geopolitical Tailwinds: The ‘Trump Bump’ and European Sovereignty
Beyond the technical capabilities of its models and its business strategy, Mistral is also benefiting from favorable geopolitical currents. European nations, particularly France, are increasingly emphasizing the need for “sovereign AI.” This concept refers to the desire to develop and control AI technologies within Europe, reducing reliance on AI systems developed in the U.S. or China. This sentiment stems from concerns about data privacy, national security, and economic competitiveness.
While this sentiment existed in 2023 when Mistral launched, it has intensified this year due to several factors. The Trump administration’s combative stance towards European tech regulation and escalating tensions between the U.S. and China have further fueled the desire for European technological independence. A recent article in The Economist suggested that “In the fast-growing world of artificial intelligence (AI), Mistral, a French startup, may be a beneficiary of the transatlantic tempest.”
Mistral has consistently enjoyed strong support from the French government. President Emmanuel Macron has frequently praised the Paris-based startup as a symbol of French innovation and evidence that the country can cultivate fast-growing tech companies capable of competing with those emerging from Silicon Valley. Like many European politicians, Macron views AI as a key driver of economic growth and productivity, and sees Mistral as a crucial player in this landscape. The fact that Cedric O, a confidante of Macron and former secretary of state for the digital economy, is now a “co-founder” of Mistral and advisor to the startup further strengthens its position and access to political support.
However, the advantage of being perceived as a “homegrown hero” may now extend beyond France to encompass Mistral’s commercial standing across Europe as a whole.
“European companies are looking to partner more closely with European technology,” Mensch observed. “They want an AI partner that can drive transformation, independent of geopolitical tensions. Our regional presence gives us an edge that others simply don’t have.” This statement highlights the growing preference among European businesses to work with European AI providers, driven by a desire for greater control, data security, and alignment with European values.
Mensch reported a “tremendous increase” in Mistral’s commercial traction in Europe over the past two months, although he claimed to be unsure whether it was directly linked to Trump’s inauguration or broader geopolitical trends. Mensch has been vocal about the importance of Europe remaining competitive in AI. At the Mobile World Congress in Barcelona earlier this month, he stated, “It feels like the conversation around AI is in the US and China, and Europe sometimes gets left out of that conversation.”
Mensch told Fortune that he was not certain that those in the U.S. fully appreciated the growing awareness among Europeans of the need to push back and assert themselves. “If Europe is mistreated, Europe is reacting,” he said, adding that “there’s definitely some pretty strong momentum around uniting, around technology, around automation, on AI.” This statement underscores the growing sense of European unity and determination to become a major player in the global AI landscape.
Ambition, Culture, and the Funding Conundrum
For the time being, Mensch stated that his focus remains on steering Mistral from a scrappy startup to a major AI player, in an era where experts question whether any AI model startup can keep pace with the industry leaders: OpenAI, Anthropic, Google, and Meta. While Mistral remains significantly smaller than these behemoths, Mensch highlighted the best practices that Mistral’s three original co-founders brought from their tenures at Google DeepMind and Meta – cultivating a mindset centered on rapid shipping and upholding rigorous scientific standards.
“We have created our own culture, which is low-ego and scrappy,” he declared. This emphasis on a lean and agile culture is crucial for a startup competing against larger, more established companies. While Mistral likely cannot match the exorbitant compensation and multi-million dollar equity stock grants offered by the foundation model giants, Mensch emphasized that the company’s open-source ethos is highly appealing to the AI research talent it seeks to attract.
“When you’re a scientist, you really want to contribute to the community, at the end of the day, you’re usually less interested in the business success of the company,” he explained. “So [open source] has been a great advantage, and that’s something that we will continue to promote.” This statement highlights the intrinsic motivation of many AI researchers, who are driven by a desire to advance the field and share their work with the broader community.
Mistral is also bolstering its research team to work on fundamental AI research that is not necessarily tied to product development. “There are so many things that need to be figured out, and new ways of thinking about architectures,” he noted. “If you’re only thinking about the product, you don’t have the time to think about these things.” This commitment to fundamental research is crucial for long-term innovation and staying ahead of the curve in a rapidly evolving field.
The precise funding mechanism for this research remains a question. Mistral has not publicly disclosed, for instance, receiving direct funding from the French government for its R&D activities, even though the French government has expressed its commitment to advancing AI within the country. It is possible that Mistral is receiving indirect support or benefits from government initiatives, but the details remain unclear.
Mistral has secured partnerships with European AI cloud platform Fluidstack, which is constructing what it claims will be Europe’s largest supercomputer. Mistral is slated to begin utilizing this AI computing cluster later this year. This partnership will provide Mistral with access to significant computing resources, addressing one of its key challenges in competing with larger players. It has also forged a collaboration with AI chip firm Cerebras, whose hardware enables Mistral’s AI assistant, Le Chat, to respond with unprecedented speed. This collaboration demonstrates Mistral’s commitment to leveraging cutting-edge hardware to enhance the performance of its AI models.
Mensch concluded by emphasizing that he and the entire team are keeping their sights firmly set on the ultimate objective: “We started very ambitiously, but we need to continue being very ambitious.” This statement encapsulates Mistral’s determination to remain a significant player in the rapidly evolving AI landscape, despite the challenges and intense competition.