In a closing keynote at The Asian Banker Summit 2026, Malaysia's Minister for Digital Gobind Singh Deo framed the country's digital trajectory around a single test: whether the investment flowing in translates into domestic capability — stronger SMEs, high-value jobs, and financial services that reach more people.

The investment base is substantial. Malaysia has attracted MYR 144.4 billion (approximately $30 billion) in data centre and cloud investment between 2021 and June 2025, with the broader digital investment pipeline exceeding $59 billion as of April 2025. Microsoft, Google, AWS, Oracle and Nvidia are among the technology players currently expanding operations in the country. The digital economy's contribution to GDP stood at roughly 23% in 2023. Prime Minister Anwar Ibrahim has set 30% as the 2030 target, articulated through the AI Nation 2030 vision announced in August 2024.

But Deo's argument was less about the targets than about the conditions required to reach them. An AI-native economy, he said, is one where intelligence is embedded into how decisions are made across every sector. Getting there requires governments and industries to change how they work — not adopt new tools within existing structures, but reorganise around new ones. "Humans today are not just users of technology," he said. "They're becoming part of the infrastructure that enables businesses to grow."

Building the base: four pillars, one significant constraint

Malaysia's AI readiness strategy rests on four pillars: connectivity, compute infrastructure, cybersecurity and talent. The connectivity foundation is largely in place, with 4G coverage at 95–96% of locations nationally and 5G approaching 90%.

Compute infrastructure has been the primary active investment, with data centres drawing most of the capital. The Cybersecurity Act 2024, tabled within the ministry's first three months of existence, establishes a regulatory framework around National Critical Information Infrastructure and sets layered security requirements for operators. "Investors want to know that if they come along and partake in this journey, the systems they are working with are safe," Deo said.

A significant operational constraint runs underneath all four pillars: the state of data inside government itself. The Data Sharing Act 2025 was designed to enable inter-agency data flows, allowing ministries to pool information across departmental lines. Implementation revealed the prior problem: a substantial portion of government-held data has not been digitised. "When you ask for data," Deo said, "someone turns up at your office with a stack of books." The ministry has since launched a digitisation audit policy requiring each ministry to inventory its data holdings before any AI integration can proceed.

Malaysia’s National Digital Economy Blueprint (MyDIGITAL) is now in its third phase. The first two laid regulatory and infrastructure groundwork. Phase three, beginning this year, is where sectoral transformation has to follow.

What an AI-native economy requires from every sector

The National AI Office, launched in December 2024, is the mechanism through which the Ministry of Digital consults individual sectors, such as financial services, health and agriculture, before drafting AI policy. Its mandate is to map sector-specific constraints in Malaysia, assess what AI has changed in equivalent economies globally, and produce frameworks institutions can act on.

Deo pointed to India's Aadhaar programme as the reference point. Government-built identity infrastructure changed what financial institutions could do, but only for those willing to change their processes to use it. "Are you prepared to take that step? The radical step, that total shift in how you operate?"

The question was directed at financial services, but Deo framed it as the central challenge for every sector. Malaysia's domestic market of 34 million is not the primary argument for the investment. The regional one is. Deo cited projections showing ASEAN economies that don't accelerate digital adoption stand to forgo approximately $1 trillion in potential gains between 2025 and 2030. With roughly 690 million people, ASEAN is where the returns on Malaysia's infrastructure investment compound, if the country builds the ecosystem capable of reaching them.

Government as proof of concept

The government is not asking industries to change while keeping its own operating model intact. The Government Innovation Initiative (GII), set for launch on 25 May, is the most concrete demonstration of this. It replaces traditional RFP processes with open problem statements: ministries publish what they can't solve, share relevant data sets, and invite industry to propose solutions. Respondents can also identify adjacent problems solvable with the same data.

The traffic management example Deo gave is instructive. A standard procurement on a congestion problem between two points produces one solution. Under the GII model, opening the problem to industry with full data access allows partners to address safety, accident response and infrastructure planning alongside the original brief, because they can see what the underlying data actually contains. Six or seven outcomes through one exercise, in Deo's framing, rather than six or seven separate procurements.

The model asks something genuinely difficult of government: to publish what it doesn't know how to solve, share its data, and accept that the best answer may come from outside existing supplier relationships. "Of course, you get pushback because people are very comfortable doing things the traditional way," Deo acknowledged. "If the banks can do it, come on, we can do it."

The accountability question underneath all of it

None of the operating model Deo described works at scale without a resolved answer to a prior question: who is liable when an AI system causes harm? Without that answer, the ecosystem has a ceiling.

A lawyer by training, Deo drew a parallel with the historical development of corporate legal identity: attaching accountability to a collective entity rather than a single person was a significant conceptual shift, one that took time to work through law and regulation. Governing AI, he argued, is at an analogous early stage. "We need to start thinking of how you can create governance around technology, so that technology understands it has a responsibility to make decisions that do not harm people. And in the event they did, there were consequences."

Deo said he had raised the question of how to assign legal accountability for AI-generated harm directly with the Prime Minister as a priority for AI Nation 2030. How Malaysia answers the question, and how quickly, will shape the pace at which the rest of the AI-native economy can actually be built.