Malaysia’s artificial intelligence ambitions are increasingly moving beyond simply adopting AI tools to building the infrastructure and technology needed to control how AI is developed and deployed.
For local Neo Cloud startup Grafilab, that shift presents an opportunity to build an AI infrastructure business around one of the region’s growing priorities: sovereign AI.
Grafilab describes itself as Malaysia’s first local Neo Cloud startup and one of the earliest Neo Cloud startups in Southeast Asia. It provides on-demand access to GPU computing and AI models through its marketplace, while also working with GPU owners to bring additional computing capacity onto its platform.
The company operates in a highly competitive market, however, where well-funded players from the US and China have the financial muscle to compete aggressively on pricing.
Rather than competing solely on scale, Grafilab has focused on what it believes can be a more defensible local advantage: domestically hosted infrastructure, data sovereignty and an understanding of the needs of Malaysia’s developer community.
“We have to look into local advantages of what we as a Malaysian start-up can offer. Infrastructure and data sovereignty came to our mind,” Grafilab Co-Founder and COO Rico Chee told TNGlobal in an interview.
The company initially acquired several GPU machines to validate demand among Malaysian developers, at a time when GPU prices were more manageable.
The response from the local technology community encouraged Grafilab to continue developing locally hosted AI infrastructure and services, ranging from computing power to AI models.
“Our close relationship with the developer community in Malaysia, which gives us a good understanding of what exactly they need for AI transformation, and our focus on creating a more secure solution for them on an infrastructure layer at globally competitive pricing are our biggest advantages,” Chee explained.
Building an AI marketplace
Grafilab’s business model is centred on providing affordable computing power and AI model services on an on-demand basis through its marketplace.
The company likens the model to Booking.com or Airbnb, positioning itself as an independent third-party marketplace connecting users with GPU-as-a-service (GPUaaS) and Models-as-a-Service (MaaS).
Its primary customers today are Malaysian small and medium-sized enterprises, including startups and developers.
Interestingly, some of Grafilab’s most consistent compute customers have come from outside Malaysia, particularly Vietnam and Indonesia.
The company attributes this partly to its pricing advantage from owning some of its machines and hosting them domestically.
Grafilab is also working on proof-of-concept projects with larger enterprises and has recently joined PIKOM, Malaysia’s national technology association, under its AI chapter to support member companies with their AI transformation.
The startup also has some cooperation with the government on national-level projects and is exploring further opportunities.
The mix of SME customers, developers and larger organisations reflects a broader challenge in the region: access to AI infrastructure remains expensive, while organisations in regulated industries increasingly need greater control over where their data is processed.
Sovereign AI could level the playing field
Data residency and sovereignty requirements are particularly relevant to industries such as financial services, healthcare, defence and critical infrastructure. For large corporations and government-linked companies with significant financial resources, buying GPUs and building their own AI infrastructure may be feasible.
For SMEs, however, the economics are considerably more difficult.
“Basically every sector that needs to comply with data residency requirements” could benefit from sovereign AI infrastructure, Grafilab said.
The company believes sovereign AI could allow smaller organisations to access advanced AI capabilities without having to build and maintain their own infrastructure.
The public sector could also benefit, particularly in areas where national security and sensitive information are involved.
Grafilab sees potential for AI to improve the efficiency of public services in Malaysia, while allowing government agencies to access AI capabilities within a more controlled infrastructure environment.
The opportunity extends beyond Malaysia, according to Chee.
Countries across Southeast Asia are increasingly exploring ways to develop greater control over their AI infrastructure, data and models. Grafilab believes this could create a regional market for Sovereign AI-as-a-Service.
“Depending on how each country defines it, we’re well-positioned to capture the data security and privacy aspect of it,” Chee noted.
For Malaysia, however, the company acknowledges that simply owning some locally hosted GPUs will not be enough to meet future demand.
It expects to work with larger infrastructure and technology partners in Malaysia and other Southeast Asian markets to scale its capacity.
From Malaysian startup to regional AI player
Grafilab’s ambitions extend beyond becoming a domestic infrastructure provider. Over the next three to five years, the company wants to build enough traction across Malaysia and Southeast Asia to work with major global AI companies and help take their ecosystems into international markets.
“When big global AI players decide to work with us to go global with their incredible AI ecosystem, that’s how we know we have truly made it as a global AI company beyond Malaysia,” it said.
The startup believes its positioning around sovereign AI could give it a role in helping Southeast Asian countries develop greater control over their AI infrastructure.
That ambition comes as governments across the region weigh the economic opportunities of AI against concerns over data security, dependence on foreign technology and the concentration of AI capabilities among a small number of global companies.
Malaysia must become an AI creator, not just a user
For Grafilab, Malaysia’s longer-term AI challenge is not simply about deploying AI faster. The country needs to develop technologies of its own, according to the company.
Citing Prime Minister Anwar Ibrahim’s call for Malaysia to become a “master” of AI technology rather than merely a user, Chee said the country should ensure that the economic value generated by AI does not disproportionately flow overseas.
“At some point, most countries would actively use AI, so merely using AI would not be sufficient,” Chee said. “Only the countries that hold the key to advanced AI technologies can win in that scenario.”
Grafilab hopes Malaysia can become one of those countries, he added.
But to achieve that, it believes Malaysia should be more selective about where it invests its resources.
Pick AI niches where Malaysia already has an edge
Rather than attempting to compete across every area of AI, Grafilab believes Malaysia should focus on sectors where it already has a strong industrial base.
Manufacturing is one such opportunity, Chee noted.
Malaysia is already a major manufacturing hub in the region, particularly in areas such as electronics and semiconductor-related industries. Developing AI technologies specifically for manufacturing could therefore provide more immediate economic benefits than attempting to replicate every aspect of the global AI stack.
“We need to allocate maximum resources on the AI niches that we already have an advantage on instead of trying to do everything to become a jack of all trades, but master of none,” Chee pointed out.
The company sees the development of AI technologies for manufacturing as a particularly promising avenue.
“We’re an advanced manufacturing powerhouse in the region, just like Germany is in the Europe region,” he said. “Creating native AI tech that helps us improve our manufacturing prowess will be making us very economically successful for the decades to come.”
More opportunities for startups
Grafilab also believes the government could play a bigger role in helping local AI startups commercialise their technology.
Rather than concentrating major government technology projects among large publicly listed companies, it wants to see more opportunities opened to startups with specialised technologies.
The company proposed multiple smaller grants for startups working on different AI challenges, ranging from climate change to education.
The approach, it suggested, could resemble a more serious, long-term hackathon: give several startups the opportunity to develop solutions, allow them to demonstrate their effectiveness and then provide a pathway for successful companies to receive investment and scale.
After proving their solutions domestically, these startups could then be connected to overseas markets.
For Grafilab, this would create a more systematic pipeline from experimentation to commercialisation and eventually international expansion.
A seed-stage company looking for strategic partners
Grafilab is currently at the seed stage and received $500,000 in incubator funding from Raijin Ventures, a Thai family-office venture capital firm, within the first few months of its operations, according to Chee.
The company is now open to raising additional capital, but he said it is particularly interested in strategic investors with strong AI industry expertise or networks that can help expand its business ecosystem.
That distinction could be important as the startup moves into its next phase.
Building sovereign AI infrastructure requires more than GPUs. It requires access to data centres, developers, enterprise customers, AI model providers and government projects — as well as partnerships that can take Malaysian technology into other markets.
For Grafilab, the opportunity is therefore not simply to become another cloud provider.
Its larger bet is that Southeast Asia’s AI future will require infrastructure that is not only affordable and scalable, but also locally controlled.
If that thesis plays out, the Malaysian startup hopes its domestic roots can become the foundation for a much larger regional role.
“Beyond Malaysia, we’re well-positioned to capture the data security and privacy aspect” of sovereign AI, Chee said.
The test now is whether it can turn that positioning into sufficient scale — and ultimately become the kind of regional AI company that global players want to work with.
Malaysia targets $5B in GDP gains from AI annually: Communications Minister