.png){kind=logo}
Asia’s technology startup ecosystem is entering a new phase as companies take artificial intelligence beyond software and deploy it across physical environments. From autonomous drones and humanoid robots to healthcare and industrial systems, startups are building AI that can interact with the real world.
More than half of the 30 companies selected for Granite Asia’s 2026 NextGen Tech 30 (NGT30) are working in areas that reflect this shift. The programme, launched with the Singapore Exchange and supported by public and institutional stakeholders, highlights growth-stage technology companies with ambitions to scale internationally.
The applications span several industries. Startups are developing robots for wind-turbine maintenance, systems for underwater vessel repairs and AI models designed to support cancer screening. The trend signals a move from AI that primarily processes digital information to systems that can perceive and act in physical environments.
Singapore emerged as an important base for this new generation of companies. Its share of NGT30 companies increased from 18% in 2024 to 42% in 2026. Japan also accounts for one-third of the 2026 cohort, with startups working across legal AI, cancer diagnostics, healthcare software and warehouse robotics.
The expansion into physical AI is closely connected to Asia’s manufacturing and supply-chain capabilities. Granite Asia previously argued that the region’s hardware, engineering and industrial ecosystems could support faster development and deployment of robotics and embodied AI.
The transition is already visible beyond the NGT30 cohort. Grab, for example, described its strategy as a move from ‘bits to atoms.’ Experts suggest that autonomous vehicles, delivery robots and other specialised robotic systems form part of its embodied-AI ambitions. Looking ahead, the next test will be turning promising physical-AI demonstrations into repeatable commercial deployments.
Also Read: Trump-Tech CEO’s Meeting: AI Renamed ‘Super Intelligence’, Safety Pact & More
