Is Malaysia an AI nation or just another AI landlord?

Kua Kia Soong

Malaysia is racing to become a regional hub for data centres.

According to Malaysia’s official National AI Action Plan 2026-2030 and Bernama’s report on RM385.7bn in data centre-related investment approvals, vast tracts of land are being allocated, and electricity and water supplies are being committed to these gigantic facilities.

The government tells us that this is part of Malaysia’s transformation into an artificial intelligence (AI) nation.

But what exactly is Malaysia getting out of this data centre boom? Are we developing our own AI capabilities, or are we simply providing the land, electricity, water and infrastructure for Nimby (not in my backyard) foreign technology companies to conduct their businesses here?

And where is the comprehensive national study that demonstrates how these investments will benefit the people of Malaysia in the long run?

Where is the national AI strategy and its implementation report?

To be fair, Malaysia is not without an AI policy. The government introduced the National AI Roadmap 2021-25 and has now published its National AI Action Plan 2026-2030, also known as AI Nation 2030.

The national AI office has been institutionalised as AI Malaysia Berhad under the Ministry of Digital.

The new action plan sets ambitious targets: 300,000- 500,000 AI-related jobs, an additional 0.8 to 1.2 percentage points of annual gross domestic product (GDP) growth, and a place among the world’s top 10 countries in AI readiness by 2030.

But where is the detailed public accounting of what we have achieved under the previous roadmap? How much have we invested in AI research and development? How many Malaysian AI researchers and engineers have we trained?

How many Malaysian companies have developed commercially viable AI products? How many research projects have progressed from universities into successful businesses?

What proportion of our AI investments is going into developing Malaysian intellectual property rather than buying foreign technology?

The new plan identifies 28 national initiatives, comprising 14 sectoral impact engines and 14 foundational enablers.

But the public deserves more than a list of initiatives and ambitious projections. We need clear budgets, responsible agencies, implementation timetables, measurable outcomes and annual progress reports. And the government must publish these details and submit its progress to parliamentary scrutiny.

RM386bn in data-centre investments: What are we actually buying?

According to the Ministry of Investment, Trade and Industry, Malaysia’s data-centre-related investment approvals reached RM386bn between 2021 and 2026.

In the first half of 2026 alone, data centre and cloud computing projects accounted for RM96bn in approved investments, nearly 44% of all approved investments during that period.

These are impressive figures. But approved investment is not necessarily money already spent, and the value of a foreign company’s investment is not the same as the value retained by Malaysia.

We must ask some elementary economic questions:

How much of this investment goes towards Malaysian contractors, suppliers, engineers and service providers?

How much of the equipment, servers, semiconductors, cooling systems and software is imported?

How many permanent jobs are actually created for Malaysians, as distinct from temporary construction employment?

What are the average salaries and skill levels of these jobs?

How many Malaysian graduates will be employed in research, engineering and management positions rather than routine maintenance and security?

The Associated Press has reported that many data centres typically employ only 30 to 50 permanent workers, although larger facilities may employ more. This raises a legitimate question about the relationship between the enormous capital investment and the number of sustainable jobs created.

We should not confuse the value of a building filled with imported servers with the development of a sophisticated domestic technology industry. A data centre can be worth billions of ringgit without necessarily generating billions of ringgit in value for the Malaysian economy.

Who benefits from our land, electricity and water?

This is perhaps the most urgent question, particularly for Johor, which has become one of the region’s fastest-growing data centre locations.

Data centres are extraordinarily resource-intensive. They require large areas of land, reliable electricity around the clock, and substantial quantities of water, particularly for cooling.

News agency Associated Press has highlighted projections showing electricity demand from Malaysia’s data centres could exceed five gigawatts by 2035. It has also raised concerns about the implications for Malaysia’s electricity supply, which has historically depended heavily on fossil fuels.

We must therefore ask:

Will electricity infrastructure have to be expanded at public expense to accommodate these foreign companies?

Who ultimately pays for new transmission lines, substations and power-generation capacity?

Will households and local businesses face higher electricity costs or greater pressure on supply?

How much freshwater will these facilities consume, and what happens when communities face water shortages?

Are operators required to use recycled water, renewable energy and the most efficient cooling technologies?

Are the environmental costs and carbon emissions properly accounted for before projects are approved?

Johor has already tightened its data centre approval requirements because of concerns about water and energy consumption. This is a welcome recognition that investment cannot be pursued regardless of its environmental consequences.

The government must also disclose the tax incentives, electricity arrangements, water tariffs and other concessions offered to these investors. Ordinary people in Malaysia have a right to know whether the public is subsidising private companies while bearing the environmental and infrastructural costs. We should not allow the pursuit of foreign investment to take precedence over the needs of ordinary Malaysians, local industries and future generations.

Are we developing our own AI models or merely hosting foreign ones?

This is the most important strategic question. Data centres are infrastructure. They provide the computing capacity to store information, operate cloud services and train or run AI systems.

But owning or hosting a data centre does not mean owning the technology that operates within it. If foreign companies bring in their own chips, servers, software, AI models and technical expertise, and use Malaysian infrastructure to serve customers elsewhere, what exactly are Malaysians learning and developing?

Are we acquiring the ability to design advanced chips, develop AI algorithms, train large language models, build software platforms and own the intellectual property?

Or are we simply renting out our resources while the most valuable technology remains in the hands of foreign companies?

There is no reason why Malaysia cannot develop its own AI capabilities. We already have universities, research institutions, computer scientists, engineers and a multilingual population that provides a particularly rich environment for AI development.

There have also been promising beginnings. The MaLLaM project has developed Malay-language large language models of different sizes. The government announced in September 2025 that it was working on a national large language model, with the Malay language as its principal focus. These initiatives deserve serious support.

But we must distinguish between developing our own models and hosting foreign models on Malaysian servers. The deployment of China’s DeepSeek model on locally hosted infrastructure, for example, can improve local data control and access to computing resources.

But hosting DeepSeek does not make DeepSeek a Malaysian invention. Data sovereignty is important, but it is not the same as technological sovereignty.

Malaysia should be capable of developing AI systems that understand Malay, Mandarin, Tamil and the many languages and dialects spoken by our communities. Such systems should understand Malaysian history, culture, laws, institutions and social realities.

We should also develop AI applications for areas where Malaysia has particular needs and strengths: agriculture, healthcare, education, manufacturing, public administration, climate adaptation and the management of natural resources.

We do not necessarily have to compete with American and Chinese companies to produce the world’s largest AI model. That would be an extraordinarily expensive undertaking.

We can instead develop smaller, specialised and multilingual models, adapt open-source technologies, train our own researchers and build intellectual property around applications that are relevant to Malaysia and Southeast Asia. Singapore’s Sea-Lion initiative, which develops language models for Southeast Asian languages, provides a useful regional example.

The question is whether Malaysia has the determination to invest in comparable capabilities rather than merely providing the infrastructure for others.

We must not merely be a landlord of the AI revolution

Malaysia has a genuine opportunity to benefit from the global AI revolution.

But we must not mistake the construction of data centres for the development of an AI industry.

The history of economic development teaches us that countries do not become technologically advanced merely by hosting foreign factories, assembling imported components or providing cheap resources to multinational companies.

They advance by acquiring knowledge, developing local expertise, investing in research, and creating products and intellectual property of their own.

Dr Kua Kia Soong is a former MP and former adviser to human rights group Suaram.

Support Aliran's work with an online donation. Scan this QR code using your mobile phone e-wallet or banking app: