Singapore and Taiwan-based CloudMile is doubling down on the next hot trend of AI tech, especially after being part of the NVIDIA Inception Program and attending the GTC 2026 keynote in San Jose, where Jensen Huang laid out that whole shift toward AI Factories.
With the consensus moving away from traditional software models and into something much more autonomous, compute power and infrastructure basically become the foundation of how AI-driven businesses operate.
From a technical standpoint, CloudMile’s Tech Lead David Liang broke it down in a pretty interesting way, pointing to how the Vera Rubin architecture is being designed specifically with AI Agents in mind. By combining ultra-low-latency inference capabilities with open ecosystems like Nemoclaw, the expectation is that token usage is going to explode at scale. The way he framed it, we’re entering a phase where execution speed finally catches up with how humans actually think, which is a big deal for real-time AI applications.
Then stepping back to the bigger picture, CEO Spencer Liu is clearly positioning the company beyond just deploying models. He’s framing this as a transition from SaaS to Agent as a Service, or AaaS, where AI Agents essentially function like long-term digital workers. Instead of just running software, businesses would be operating with what he describes as lifetime digital employees that convert compute into actual business outcomes.
And tying it all together at NVIDIA GTC 2026, CloudMile is presenting itself as more than just a service provider in APAC – it’s aiming to act as a kind of AI foundry. The AEGIS AI Agent platform is positioned to handle the full lifecycle, from deployment to governance, while keeping things grounded in security and tailored infrastructure needs. So overall, what you’re seeing here is CloudMile trying to bridge that gap between cutting-edge AI infrastructure and something enterprises can actually deploy and scale in the real world.
