Global AI, Malaysia Angle / AI news for Malaysia
NVIDIA AI Day names Malaysia's ITMAX and YTL as Sea adopts Vera Rubin
NVIDIA's Singapore showcase put a Kuala Lumpur traffic system and a Malaysian model builder on its regional list. The performance numbers come from the companies, and key details on scale are missing.

In brief
- NVIDIA's AI Day Singapore post of 23 September 2026 says Malaysia's YTL AI Labs is fine-tuning Nemotron models and ITMAX is using Cosmos for city traffic operations.[1]
- ITMAX says its SamurAI system is in a phased first deployment in Kuala Lumpur with DBKL, and reports a gain of more than 70 per cent in alert verification accuracy.[2]
- NVIDIA says Sea, the owner of Shopee, is the first enterprise in ASEAN to adopt its Vera Rubin platform. It gives no scale, timing or location.[1]
What NVIDIA announced in Singapore
NVIDIA used its AI Day in Singapore, held on 22 and 23 September 2026 at the Raffles City Convention Centre, to list companies across Southeast Asia building on its models and chips. Two Malaysian names made the list: YTL AI Labs and ITMAX System Berhad, a Malaysian digital infrastructure company.[1][2]
NVIDIA also said Sea Limited, the Singapore-founded owner of Shopee, is the first enterprise in the ASEAN region to adopt its Vera Rubin platform. For Malaysian readers, the ITMAX entry may matter more, because it describes AI already working on Kuala Lumpur roads.[1][2]

Two Malaysian names on NVIDIA's list
NVIDIA's post says YTL AI Labs is fine-tuning Nemotron models for enterprise and citizen services. Nemotron is NVIDIA's family of open models, which companies can adapt with their own data.[1]
The same post says ITMAX uses NVIDIA's Cosmos world models with its video search and summarisation blueprint to improve city traffic operations. For YTL, it names no product, customer, launch date or language. That describes work in progress, not a product.[1]

What ITMAX says is running in Kuala Lumpur
ITMAX's own release, dated 23 September 2026, adds detail. It says its SamurAI computer vision platform and City OS operations platform have an initial deployment under way in Kuala Lumpur with Kuala Lumpur City Hall (DBKL). It starts with a first set of corridors before widening across the city.[2]
The company says it fine-tuned NVIDIA's Cosmos 3 model on Malaysia-specific road and incident footage. It reports more than a 70 per cent improvement in alert verification accuracy in the initial deployment, and says an agent built on Nemotron has cut the time needed to find a specific incident by more than 90 per cent.[2]
These are ITMAX's own figures. The release does not say how accuracy was measured, over what period or against which baseline, and it does not name the corridors covered.[2]
Sea's Vera Rubin move and what it leaves out
Sea runs Shopee, the Monee financial services arm and the Garena games business. NVIDIA says Sea will use Vera Rubin to develop and deploy AI models and agents at greater scale, building on AI that already helps Shopee sellers write product listings, improve discovery and make business decisions.[1]
The announcement does not say how many systems Sea is buying, when they arrive, where they will be hosted or what they cost. It does not say whether Malaysian Shopee sellers will see any new feature. First in ASEAN is NVIDIA's description, and we could not check it independently.[1]
Singapore's warning about pilots
Singapore's Senior Minister of State Tan Kiat How used his address at the event to question the gap between AI capacity and results. He said he has seen many companies build impressive AI pilots, and that a successful pilot alone is not a successful transformation.[3]
He described three shifts: start from a real problem, redesign the workflow instead of adding a smarter model to an unchanged one, and link AI spending to measurable outcomes such as productivity, revenue, quality and safety. He cited Sea's use of AI on Shopee as an example.[3]
Why Malaysia should care
Two Malaysian companies appear in NVIDIA's regional line-up, and one describes live work with DBKL on Kuala Lumpur roads. Its performance numbers come only from ITMAX and have not been independently checked.
Malaysian SME owners
Shopee's seller tools are set to get more computing power behind them, but no Malaysian feature is announced.[1]
Practical move: Watch Shopee Seller Centre notices, not NVIDIA releases, for anything you can use.
Councils and agencies
A Malaysian vendor reports a phased AI traffic deployment with DBKL.[2]
Practical move: Ask for measured baselines before adopting the same approach elsewhere.
What Malaysians can do now
- If you sell on Shopee, watch for new listing and discovery tools and judge them on your own conversion numbers.
- If you work with road or city data, ask how ITMAX measured its 70 per cent accuracy gain before relying on it.
- If you build AI products, try fine-tuning an open model on Malaysian data, as YTL AI Labs and ITMAX say they are doing.
What we still do not know
What we still do not know
- Which Kuala Lumpur corridors SamurAI covers, and how its accuracy figure was measured.
- What YTL AI Labs will release from its Nemotron work, and when.
- How much Vera Rubin capacity Sea is buying, and where it will run.
Sources
- 1.At AI Day Singapore, NVIDIA and Partners Showcase AI Advancements Across Southeast Asia NVIDIA, 23 September 2026
- 2.ITMAX Advances AI for Malaysia's Smart Cities with NVIDIA Cosmos ITMAX System Berhad, 23 September 2026
- 3.Address by SMS Tan Kiat How at NVIDIA AI Day Singapore Ministry of Digital Development and Information, Singapore, 23 September 2026
- 4.File:5 Science Park Drive (Shopee), Singapore (2025) - img 05.jpg Wikimedia Commons, 3 March 2025
- 5.File:Raffles City, Singapur, 2023-08-17, DD 18.jpg Wikimedia Commons, 17 August 2023
- 6.File:Jalan Tun Razak, Kuala Lumpur 20230828 151447.jpg Wikimedia Commons, 28 August 2023


