Why Apple and ASML Back Mistral AI: Inside Its Tech, Funding and Controversies

Founding, Funding and Market Attention

Mistral AI was founded in Paris in April 2023 by three former DeepMind and Meta researchers—Arthur Mensch, Timothée Lacroix and Guillaume Lample. Within two months it secured €105 million in seed funding, a European record, followed by a €385 million Series A in December 2023 and a €600 million Series B in June 2024, lifting its valuation to €58 billion. Reuters (Sept 8 2024) reported a €1.7 billion Series C round, with valuation projected at €120 billion (US $140 billion). ASML led the round with €1.3 billion, becoming the largest shareholder and gaining a board seat.

Apple’s Acquisition Interest

Bloomberg journalist Mark Gurman disclosed that Apple is seriously considering acquiring Mistral AI, potentially its largest ever acquisition. Apple sees Mistral’s advanced natural‑language processing as a way to address Siri’s lag behind competitors such as Google Assistant and Amazon Alexa, and to enrich its productivity suite (Pages, Numbers, Keynote) and enterprise services (iCloud for Business).

Technical Highlights

The company’s website lists eight model families covering three task categories: simple batch tasks (e.g., classification, text generation), intermediate tasks (e.g., document summarization, email drafting), and complex high‑reasoning tasks (e.g., code generation, Retrieval‑Augmented Generation, agents).

Lightweight Language Models

The open‑source Mistral 7B model has 7 billion parameters but outperforms peers of similar size, delivering strong results on complex reasoning and coding benchmarks while requiring far less hardware, thus fitting low‑resource devices.

Mixture‑of‑Experts (MoE) Advancement

The improved Mixtral 8×7B MoE model became the most advanced MoE on Hugging Face at its release. In multiple authoritative benchmarks it surpassed closed‑source Llama 2 70B, and its inference speed was six times faster than traditional dense models. Under the Apache 2.0 license it outperformed GPT‑3.5 Turbo, Claude‑2.1, Gemini Pro and Llama 2 70B on human‑level tests (see https://hyper.ai/papers/2401.04088).

Multimodal Exploration

Pixtral Large combines image understanding with text generation. It can analyze medical images and produce diagnostic reports, suggest driving decisions from road‑scene photos, or generate descriptive captions for creative work, expanding AI applicability beyond pure language tasks (see https://hyper.ai/papers/2410.07073).

Open‑Source Strategy and Product Matrix

All major models (Mistral 7B, Mixtral, etc.) are released under Apache 2.0, encouraging community contributions and rapid iteration. The ecosystem includes developer‑focused tools such as La Plateforme and Codestral for model customization and code completion, and enterprise solutions like the conversational AI assistant Le Chat, which reportedly doubled business scale within 100 days of launch.

Controversies and Ethical Concerns

In August 2024 former employees alleged that Mistral’s latest model was distilled from DeepSeek’s work and misrepresented as original reinforcement‑learning research, with benchmark results allegedly exaggerated. Independent analyst Sam Paech (June 2025) identified high similarity between Mistral‑small‑3.2 and DeepSeek‑v3 outputs, suggesting possible copying. DeepMind researcher Susan Zhang publicly criticized the practice, while Hugging Face co‑founder Clement Delangue defended the legality of model distillation.

Strategic and Geopolitical Implications

Beyond technology, Mistral is viewed as Europe’s flagship AI model amid growing EU regulation of foreign AI services. Analysts speculate that investors see Mistral as a “European backup” should the EU restrict U.S. models, positioning it as a strategic asset in a potential trans‑Atlantic tech rivalry.

Conclusion

While Mistral AI’s valuation has surged and it attracts heavyweight investors, ongoing plagiarism accusations and the broader geopolitical context create uncertainty about its long‑term impact. Its technical innovations, open‑source ethos, and diversified product offerings make it a pivotal player in Europe’s AI landscape, but future success will depend on resolving ethical disputes and navigating policy shifts.