He said the site is supposed to draw power from both solar energy as well as 600 megawatts of backup supply from the national grid.
“We already have 300 megawatts being built and commissioned on the site,” he told reporters after meeting Prime Minister Anwar Ibrahim and Nvidia CEO Jensen Huang on the sidelines of the Apec Economic Leaders’ Meeting here yesterday.
“It’s growing at a very rapid pace. We reckon in the next two years, this will be fully taken up.”
YTL Power yesterday announced the completion of its AI data centre, powered by Nvidia’s latest liquid-cooled NVL72 Grace Blackwell (GB 200) GPUs.
It is located on a 664ha site at the data centre campus in Johor and will support large-scale, high-performance AI and machine learning workloads.
Yeoh said the Nvidia-powered AI data centre, developed in partnership with the US tech giant, was commissioned this month and is already operational, with customers, including global hyperscalers (large cloud service providers), beginning to use its facilities.
“We are now talking about further expansion but the speed of further expansion will also be very reliant on the adaptation of the Malaysian government to AI utilisation,” he added.
YTL Power has so far invested RM10 billion on its AI drive – half in data centre infrastructure and half in AI solutions, including Malaysia’s first large language model, Ilmu.
Yeoh also described Malaysia’s progress in AI development as “remarkable”, noting that Johor has become the region’s emerging AI hub just 22 months after Huang’s first visit.
“The prime minister was very pleased that in a short 22 months, so much progress has been made through AI in Malaysia.
“So moving forward, the prime minister also wanted to have ideas from the Nvidia CEO to have a committee to give him ideas on how to further accelerate this AI programme in Malaysia,” he said.
He added that Malaysia’s AI ecosystem was already showing strong momentum, with around 200 local startups working with YTL Power to connect to their APIs (Application Programming Interfaces) and build their own AI-driven solutions.