Cohere has opened a Defence AI Accelerator in Singapore, a dedicated hub designed to help defence organisations and other sovereign customers move artificial intelligence projects out of pilot testing and into live, mission-ready operation. Announced on November 3, the accelerator targets one of the hardest problems in government AI adoption: systems that work in a demo but stall before they can be trusted in highly secure environments.
The hub will offer hands-on collaboration with Cohere engineers, technical expertise and engineering advice covering the full journey from proof-of-concept to production systems. Customers will be able to experiment with Cohere's models and its agentic AI platform, North, in real-world settings, then scale the approaches that prove themselves into operational infrastructure.
Built for the most sensitive workloads
The accelerator is aimed squarely at organisations that cannot simply hand their data to a public cloud service. Cohere's platform supports on-premises, air-gapped and hybrid deployments, letting customers keep full control over their data and infrastructure while still running modern AI systems. The company says its North platform has already been deployed securely with several regulated organisations in Singapore, giving the new hub local reference points rather than just a sales pitch.
That positioning matters because defence and public-sector buyers judge AI vendors on governance as much as capability. The accelerator promises access to enterprise-grade models built for high-priority workloads and large-scale deployments, with security, privacy and transparency treated as design requirements rather than add-ons.
Sovereign AI is now a regional priority
The launch lands as sovereign AI investment across Asia-Pacific reaches new highs. Industry researcher IDC expects 80 percent of enterprises in the region to prioritise AI sovereignty for mission-critical workloads, and sovereign AI has become the region's second-highest government investment priority. Governments want the productivity gains of AI without depending on foreign infrastructure or losing control of sensitive data, and vendors that can prove local, governable deployment have an opening.
Cohere is betting that opening is large and durable. The Singapore accelerator forms part of its plan to quadruple its regional presence by the end of 2027 through new investment, local partnerships and additional offices across Asia-Pacific. The hub is also meant to deepen collaboration with Singapore's public-sector institutions and strengthen support for the public sector there.
From experimentation to operations
The distinction Cohere draws between experimentation and operational deployment is the heart of the announcement. Many government AI projects never leave the pilot stage because security reviews, integration work and reliability requirements prove harder than building a prototype. An accelerator that pairs models with engineers, deployment experience and a platform already proven in regulated settings is designed to close exactly that gap.
For Singapore, the hub adds another piece to its ambitions as a trusted AI centre in Asia. For Cohere, founded in 2019 as a provider of sovereign AI for regulated industries and governments, it is a concrete step in turning regional expansion plans into on-the-ground capacity — and a test of whether hands-on, security-first support can convert sovereign AI interest into systems that actually run the mission.
