Singapore Home Team uses NVIDIA’s latest AI chips to supercharge public safety robots
Singapore's Home Team Science and Technology Agency (HTX) is deploying Nvidia's new GB300 AI chips. This makes HTX the first organization in Singapore to use the GB300, which became commercially available in early 2026. The chips will power HTX's NGINE AI infrastructure, improving its ability to train multimodal and embodied AI models for public safety robots. HTX also plans to establish the Home Team Humanoid Robotics Centre (H2RC) with a $100 million commitment to develop robots for public safety enforcement.
Singapore's Home Team is moving beyond basic AI applications like facial recognition. The deployment of Nvidia's GB300 chips and the $100 million commitment to the H2RC show a clear shift towards embodied AI and robotics. This focus on physical AI systems for "dull, dirty, and dangerous" tasks reflects Singapore's unique constraints of land, labor, and an aging workforce. It is a pragmatic response to real operational challenges.
Singapore is positioning itself as a leader in sovereign AI infrastructure for public safety. HTX's emphasis on keeping code and data within its control addresses a key concern for governments adopting advanced AI. While the immediate impact is on Singapore's Home Team, this could set a precedent for other Asian nations facing similar demographic and security pressures. The region watches for how Singapore scales these deployments.
The critical test for HTX is translating this advanced chip power into operational robots that genuinely reduce man-hours and enhance safety. The reported saving of close to a million man-hours annually from existing AI use is a strong start. However, the complexity of embodied AI in dynamic, real-world scenarios presents a greater challenge. Success here will validate Singapore's investment and drive further adoption across Asia.
Share this article
Related reading
6 stories
The most dangerous AI claims aren’t wrong — they’re overstated

I Built an AI Safety Layer on a 4GB Phone

Don't let the AI change its own pass condition: pick the protected files, stop on any change

Your SKILL.md is portable. Is its behavior? Introducing Skillporter

GPT-6.2 Sol vs. Claude Opus 5.5: Not as accurate, but still might be the better buy

