THE new US regulations on chips and the Nvidia saga may weigh heavy on the data centre (DC) boom, but in Malaysia, DC investments remain phenomenal.
However, will the DC investment boom help create an artificial intelligence (AI) economy, as government officials would like to believe?
The simple theory is this: the huge amount of computing power coming to our shores by virtue of the DCs will help create high-value jobs.
However, DCs don’t directly create many jobs.
Indirectly, though, they have created billions of ringgit in value from land sales, and construction and engineering work, as well as from electricity supply contracts with Tenaga Nasional Bhd
and renewable energy (RE) players.
These jobs, contrary to expectations, are mainly one-off and do not go into the essence of AI.
With just buildings, Malaysia risks becoming a mere host for foreign-owned AI infrastructure, while the high-value activities involving AI development, software and intellectual property (IP) remain overseas, says Assoc Prof Dr Zulkifly Abbas of Xiamen University Malaysia’s School of AI and Robotics.
In June this year, he warned that building DCs does not automatically create an AI economy; Malaysia must instead prioritise developing local talent, research capabilities, technological expertise and IP.
He stressed that Malaysian universities, researchers and software companies should have access to some of the computing power being deployed here by DCs.
Due to the massive DC investments, Malaysia now ranks among the world’s top nations for AI computing power. Yet, the spillover into the creation of related IP remains elusive.
To be fair, the government has been seeking to address the issue.
“We are considering the establishment of a National AI Compute Exchange (NACX) to connect Malaysian businesses, startups, researchers and universities with available computing resources,” Digital Minister Gobind Singh Deo tells StarBiz 7.
“By working with cloud and AI infrastructure providers, this initiative will aim to improve access, affordability and choice, enabling more Malaysians to develop and commercialise AI technologies.
“Our objective is to translate global technology investments into measurable domestic economic value – through stronger Malaysian companies, higher-value jobs, greater technological capabilities and wider access to the infrastructure needed to build Malaysia’s AI economy.”
The NACX is scheduled to be launched in the first quarter of 2027.
Then there’s Amazon Web Services (AWS), another builder of DCs in Malaysia, which launched “Semarak AI Malaysia”, to expand AI access and training for government officers, students, software companies and communities, including six months of free AWS AI tools for eligible small and medium enterprises (SMEs).
While this may push the AI adoption needle, it is not the same as AWS offering its compute resources at its Malaysian DC for free.
In fact, it seems more related to getting users accustomed to the company’s AI tools, a tried and tested marketing strategy in the technology world.
Hence, it is left to be seen how any of this computing power can be secured from the DCs in the country under the NACX, considering how expensive it is.
Malaysia has secured RM123.5bil in approved DC and cloud investments in the first nine months of 2026, with most of these facilities being operated by global technology firms that allocate their computing capacity to international clients at global market rates.
DCs are willing to fund their construction and purchase of equipment and chips based on the high demand from companies needing this computing power.
So will the Malaysian government try to force the DCs to allocate free or discounted computing resources? And will they bite?
It is also noteworthy that global tech giant Google is making hefty DC-related investments into Malaysia. Google has said that its new Malaysian facilities will sustain more than 9,000 local jobs annually once operational.
However, Google has not released an exact headcount split across specific roles for the annual operational jobs.
It is unlikely that this is a projection of 9,000 AI scientists, software engineers or even direct Google employees.
In fact, if you do an online job search for “AI scientist” roles there are hardly any openings, although a few AI cloud internships related to DCs in Malaysia do appear.
So where are the AI jobs?
In reply to the question on the breakdown of actual jobs created to date from DCs and how close are we to the government’s ambition of creating up to 300,000 AI-related jobs by 2030, Gobind explains that the two could be separate issues.
“The greater employment opportunities from AI extend well beyond DCs, as envisioned under the National AI Action Plan 2026–2030, which targets to generate 300,000 to 500,000 AI-related jobs.
“As AI transforms industries and business operations, it will create new occupations, increase demand for specialised skills and enable existing workers to become more productive”.
In other words, one does not need to have a DC cluster in order to create an AI economy.
Gobind insists that the DCs are still part of the ecosystem.
He says the plan is to maximise the economic benefits of DC investments through two approaches.
First, it aims to develop local industries supporting DCs, including energy infrastructure, cooling systems, engineering, equipment manufacturing and semiconductor-related activities.
A minimum 30% local-content requirement for approved DC projects is intended to ensure this happens.
Secondly, he says, the government wants to build an AI economy on top of DC infrastructure by providing Malaysian startups, researchers and businesses with affordable computing power to develop their own AI technologies, IP, products and services.
But again, the latter goal seems elusive.
Billion ringgit boom illusion
Policy analyst Tarmizi Anuwar warns that without tracking local value retention, mapping industry spillovers, and leaning into existing technical strengths, Malaysia risks remaining a passive landlord and tech consumer rather than an AI innovator.
“Malaysia has a spillover blindspot,” he tells StarBiz 7.
Tarmizi attributes this to “contract value versus retained value”.
Headline contract wins for local suppliers can often be misleading too, he says.
“A RM100mil contract awarded to a Malaysian vendor does not mean that hundred million ringgit stays in the economy if specialised hardware like graphics processing units (GPUs), cooling systems and server racks must be imported.”
He says Malaysia still lacks a clear understanding of which industries stand to benefit from DC investments and how these economic spillovers can be measured.
Tarmizi cited Powerwell as a good example of a Malaysian company benefitting from this demand. In September 2026, Powerwell Holdings, through its subsidiary Powerwell International, secured two purchase orders worth RM190.4mil to supply low-voltage switchgear and related equipment for a DC project in Johor.
“This demonstrates how local manufacturers can participate directly in the DC supply chain.”
He underscored the need for “per-sq-km” impact tracking around DC zones to measure how many surrounding businesses actually benefit, of which data Malaysia does not yet systematically collect.
“We can’t compete head-on in building foundational large language models or manufacturing every hardware component from scratch as it is prohibitively expensive and unnecessary,” Tarmizi adds.
An industry player points out that while the numbers are impressive, the hierarchy of who actually gets paid is less so.
He says: “The surest local gains sit in the construction and power-infrastructure layer: main contractors, high-voltage engineering, procurement, construction and commissioning firms, switchgear and cable makers, and the mechanical and cooling specialists.”
He says that a more durable share of the AI pie lies further along in electronics manufacturing services (EMS) and system integration, where servers get assembled, tested and fitted with liquid cooling.
“Without rules requiring a meaningful share of that work to be done onshore once the next wave of capacity comes online, much of it will simply stay overseas,” he added.
Last year, then Economy Minister Rafizi Ramli had suggested that more DCs in Malaysia use chips designed by Malaysian companies.
He also said that Malaysia should aim to locally produce its own GPU chips in the next five to 10 years, in line with the increasing demand from DCs.
Again, this may entail enforcing specific local content requirements on DC investors coming into the country.
The concern is that those requirements could put a damper on the business case of DC investments into Malaysia.
Meanwhile, RE offers another opening.
DCs are required to source at least 30% of their power from renewables, and solar can bridge the gap before gas plants arrive.
“Grid congestion, water limits and a thin talent pool will decide how much of the headline investment stays in Malaysia rather than leaking out as imported GPUs and cooling gear.
“Without tighter tracking of retained value and skills transfer, even the better contracts risk remaining the economic equivalent of a good catering job at someone else’s wedding.
“The AI IP upside still sits higher up the stack and will not materialise simply because the servers are humming in Johor,” the industry player notes.
Whether Malaysia can escape the DC infrastructure trap will hinge on execution and the right government policies.
Ideally, Malaysia must show that it can successfully translate record-breaking capital inflows into accessible computing power for Malaysia n companies.
Initiatives like the upcoming NACX and AWS’s localised training programmes offer a vital bridge, but bridging a chasm requires more than stepping stones.
