Urgent.News

What's breaking now, across thousands of outlets.

AI

Asia’s top tech start-ups move beyond software towards physical AI

This includes deploying artificial intelligence across wind farms, warehouses and operating theatres.

Asia's leading technology startups are expanding their focus from software to physical applications of artificial intelligence (AI), according to the 2026 NextGen Tech 30 (NGT30). This list, compiled by Singapore's Granite Asia in partnership with the Singapore Exchange, features 30 growth-stage technology companies from across the region.

Over half of these companies have shifted their AI applications from purely digital to the physical world, utilizing AI across various industries such as wind farms, warehouses, construction sites, operating theatres, and even low-Earth orbit. Jenny Lee, a venture capitalist and senior managing partner at Granite Asia, noted that while software companies previously sought to change the world through software, the current cohort has become more grounded in practical applications, such as robots repairing wind turbines and AI models detecting cancer.

Chee Hong Tat, the Minister for National Development and Deputy Chairman of the Monetary Authority of Singapore, highlighted the depth and breadth of Asia's innovation base, praising the diverse range of industries these companies are impacting.

Written by urgent.news from Straits Times Business's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at straitstimes.com →

More in AI

VRAM for local LLMs: why memory bandwidth sets your tokens per second

How much VRAM for an LLM is the wrong first question. The better one is how fast that VRAM is, because a local model generating text reads its entire set of weights from memory for every single token.

  • Memory bandwidth, not VRAM size, determines LLM speed
  • Token generation requires reading all model weights from memory
  • RTX 3090's 936 GB/s bandwidth generates ~90 tokens/second

More from Wednesday 30 September →