Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseAPI / Product · ActionableTNW→ original

Mistral Opens Singapore Hub for Asia, Plans to Grow to 200 Staff by 2027

French AI lab Mistral has opened a regional headquarters for the Asia-Pacific region in Singapore, which will also serve as a research center. The company plans to more than double its regional team to 200 people by 2027, to build models for public services, robotics and factories. The bet rests on "sovereign AI": open-weight models that companies and governments run on their own servers, keeping their data in-house. For Asian public-sector and industrial buyers, that is an alternative to American cloud AI.

AI-processed from TNW; edited by Hamidun News
Mistral Opens Singapore Hub for Asia, Plans to Grow to 200 Staff by 2027
Source: TNW. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceTNW ↗
Editor-in-Chief
Key metric #1HTX, Mistral and Microsoft began building models for law enforcement in May 2025
Series C in September 2025€1.7 billion at a valuation of about €11.7 billion
Key metric #3In July, a model with 8 billion parameters was released: a robot navigates a building using a single camera

French AI lab Mistral has opened a regional headquarters for the Asia-Pacific region in Singapore, and at the same time made it a research center. There, the company will build models tailored to local industries: public services, robotics and high-tech manufacturing. The company plans to more than double its regional headcount to 200 people by 2027. The idea behind this is "sovereign AI": a government or company does not rent someone else's model as a black box, but runs it on its own servers and decides for itself where the data sits and how the system works. For developers and founders, this is a signal that such an approach is spreading beyond Europe.

What Mistral Is Building in Singapore

The company intends to build its own advanced models for three areas: public services, robotics and manufacturing. The Singapore office is meant to produce not a general-purpose chatbot, but models tuned to the tasks of specific industries in the region.

The team is growing in parallel. Mistral wants to more than double its headcount in the region to 200 people by 2027. New hires will go into research, applied AI, engineering and robotics.

Co-founder and CEO Arthur Mensch framed the idea: Singapore should become a hub from which governments and companies in the region can keep ownership of their infrastructure and "intelligence." Put simply, the customer gets a model it can adapt to its own needs, rather than only using someone else's service.

Why the Open-Weights Bet Could Work Here

The core of Mistral's offer is open-weight models. They can be deployed inside an organization's own infrastructure, and the organization decides for itself where data is stored and how the system behaves. In Europe, this argument resonated amid concerns about dependence on American tech companies. Now Mistral is testing it in Asia.

The company already has a foothold in Singapore. In May 2025, HTX, the agency that develops technology for public safety services, signed an agreement with Mistral and Microsoft on models for such tasks. That was followed by a strategic partnership with HTX, as well as collaboration with Singtel, NCS and ST Engineering.

According to Mistral, projects with HTX and DSO National Laboratories, Singapore's defense research organization, were among the first in which models were fine-tuned for a single client. That experience helped shape a broader product lineup. HTX's Chief AI Officer, Ang Chi Wei, noted that the partnership tied development to the real work of government agencies.

Robotics and "Physical AI"

Robotics is a separate bet. In July, Mistral released an 8-billion-parameter model that helps robots orient themselves in unfamiliar buildings from the image of a single ordinary RGB camera. It was trained entirely in simulation, without real-world test sites.

The new center will continue this line. Research will cover physical AI and engineering simulations for factory environments, that is, tasks where the system interacts with the physical world rather than processing text or generating content.

Funding and Context

Mistral was founded in Paris in 2023 and has become one of the most heavily funded European AI companies. In September 2025, a €1.7 billion Series C round was led by lithography equipment maker ASML, valuing the company at roughly €11.7 billion. Singapore's Economic Development Board welcomed the expansion, saying it would strengthen the country's position in advanced AI and give local specialists the chance to work on serious research.

What This Means for Practitioners

First, companies in the region gain another model provider whose models can run on their own servers and be fine-tuned for industry tasks. This matters where data cannot be sent to external clouds: in the public sector, security and manufacturing.

Second, there are limits. These are stated plans so far: the company has not named specific new models for the region, release timelines or test results. The real value will show in the first products and in how fine-tuned models perform on customers' tasks compared with American solutions.

Third, for founders this is an example of strategy: instead of competing for the title of "smartest model," Mistral is betting on control, local partnerships and industry-specific tuning. For the company, Singapore is at once a market, a research site and a test of whether the idea of sovereign AI works beyond Europe.

Story

Key Questions Answered

What Mistral Is Building in Singapore?

The company intends to build its own advanced models for three areas: public services, robotics and manufacturing. The Singapore office is meant to produce not a general-purpose chatbot, but models tuned to the tasks of specific industries in the region. The team is growing in parallel. Mistral wants to more than double its headcount in the region to 200 people by 2027. New hires will go into research, applied AI, engineering and robotics.

Why the Open-Weights Bet Could Work Here?

The core of Mistral's offer is open-weight models. They can be deployed inside an organization's own infrastructure, and the organization decides for itself where data is stored and how the system behaves. In Europe, this argument resonated amid concerns about dependence on American tech companies. Now Mistral is testing it in Asia. The company already has a foothold in Singapore.

What are the key details on Robotics and "Physical AI"?

Robotics is a separate bet. In July, Mistral released an 8-billion-parameter model that helps robots orient themselves in unfamiliar buildings from the image of a single ordinary RGB camera. It was trained entirely in simulation, without real-world test sites. The new center will continue this line. Research will cover physical AI and engineering simulations for factory environments, that is, tasks where the system interacts with the physical world rather than processing text or generating content.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Unified AI Workspace · Nomad

Test Frontier LLMs & Autonomous Agents in One Workspace

Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.

Try Nomad AI →Launch Nomad Agent
What do you think?
Loading comments…