McKinsey: Malaysia must turn AI investment momentum into real economic value -

Well positioned to strengthen its regional role if it builds on current momentum
In reality, AI transformation is a business and people focused transformation, not a tech one

Malaysia is emerging as a key node in Southeast Asia’s growing artificial intelligence (AI) ecosystem with one of the highest concentrations of data centres per square foot globally, particularly in Johor. Still, as global technology companies continue to pour investments into the region, a pressing question is coming into focus.

How can this momentum be translated into real economic value, especially against the reality that billion-dollar data centres are run by less than 100 people?

That question anchored a recent media roundtable hosted by McKinsey & Company, where execs shared insights from a new report titled “AI in Southeast Asia: An Era of Opportunity.”

Vivek Lath, Partner at McKinsey, and Vidhya Ganesan, Managing Partner for McKinsey Malaysia, explored how companies across the region are moving beyond AI experimentation and beginning to deploy the technology at scale.

“Our survey suggests about two million SMEs across Southeast Asia are already using AI,” Vidhya said. “However, only one in ten report measurable value reaching the bottom line.”

While hyperscaler investments form the backbone of Malaysia’s AI infrastructure, the broader impact will depend on how businesses, particularly smaller local firms, adopt and deploy the technology.

Can Malaysia turn hyperscaler investments into lasting AI value?

One question raised during the discussion was whether hyperscaler investments from companies such as Microsoft, AWS, and Google will translate into meaningful gains for local firms.

Vivek (pic) said Malaysia has strong potential at both corporate and national levels.

“At a company level, Malaysian firms can accelerate their AI journey and emerge as global leaders,” he said.

He added that Malaysia is well positioned to strengthen its regional role if it builds on current momentum.

On hyperscaler investment, he said benefits are already visible but depend on ecosystem maturity.

“Local companies do get a fair share during construction and beyond,” he said. However, he cautioned that Malaysia must move beyond being an infrastructure base and focus on higher-value activities.

Vidhya highlighted both the trade-offs and opportunities of large-scale data centre investments, particularly around land, water, and energy use.

“The disadvantages are clear,” she said. “But the upside can be significant if we play it well.”

She noted that hyperscaler presence creates direct engagement with global leadership teams and opens doors for partnerships such as lighthouse projects, and co-developed use cases.

She also pointed to a potential “talent halo effect,” where Malaysia benefits from exposure, knowledge transfer, and ecosystem development, provided the right policies and talent strategies are in place.

Pointing to the strong data centre investments into Malaysia, Vidhya said, “There is no shortage of investment from hyperscalers such as Microsoft, AWS, and Google. The question is how that investment translates into GDP growth and good jobs. That means moving up the value chain, rather than simply being land providers for infrastructure.”

She drew parallels with Malaysia’s semiconductor sector, where the country has long been strong in assembly and testing but has struggled to move into higher-value segments such as design and intellectual property.

“The challenge is how we move up the value chain,” she said. “We need innovation, intellectual property, and higher-value capabilities.”

She also stressed the importance of a more strategic approach to foreign direct investment.

“We need to be more thoughtful about what we ask for in return,” she said. “That includes developing local vendors, building talent, and localising parts of the supply chain.”

Beyond large corporations, Malaysia’s AI story is also being shaped by SMEs and tech startups. AI is lowering barriers to entry, allowing smaller firms to compete more effectively with larger players.

Where large companies once had an advantage through scale and manpower, SMEs can now use AI to personalise marketing, improve efficiency, and optimise operations.

Local solution providers, system integrators, and startups are therefore well placed to bridge the gap between technology and business needs, helping organisations move from pilots to scaled deployment.

AI adoption rises in Southeast Asia, but value capture lags behind

According to Vivek, Southeast Asia is not lagging behind in the AI race. In fact, the region is quickly becoming one of the world’s most active AI markets.

“One interesting question people often ask is whether Southeast Asia is behind when it comes to AI, Vivek said. “From what we are seeing, that is not the case. The region is becoming the world’s AI arena.”

He added that the region’s unique advantage lies in its access to both US and Chinese hyperscalers. “That gives companies more flexibility when designing their technology stacks,” he said.

Across the region, organisations are shifting from experimentation to execution. Generative AI has pushed adoption into the boardroom, with CXOs increasingly focused on transformation at scale.

However, scaling remains difficult. McKinsey’s research shows that while more than 70% of companies are adopting AI, most are struggling to generate meaningful financial returns.

“In terms of EBITDA impact, many companies still see less than a 10% improvement,” Vivek said. “They are investing more, but still asking where the value is.”

The report surveyed 330 AI-using companies across Indonesia, Malaysia, the Philippines, Singapore, Thailand, and Vietnam. Respondents included firms from financial services, consumer goods, healthcare, technology, and telecommunications, ranging from small companies with under US$100 million in annual revenue to large enterprises earning more than US$250 million.

It found that most organisations are allocating between 11% and 40% of their technology budgets to AI, yet returns remain limited. Around 60% reported less than a 5% improvement in EBIT, while nearly 20% saw no measurable impact.

Vivek also highlighted that AI tools, while powerful, still face reliability challenges. This creates hesitation in deploying them in critical workflows.

“A couple of weeks ago, I read about a lawyer using generative AI to generate case references, but the citations were fabricated,” he said. “It raises questions about trust in high-stakes environments.”

At the same time, cost remains a concern. While model prices are falling, integration, infrastructure, and operational costs remain significant.

“Companies need to look at end-to-end costs,” he said.

Despite these challenges, Vivek noted that companies deploying AI effectively are already seeing results.

“Some organisations are seeing revenue growth,” he said, citing DBS and Grab from Singapore, and Petronas from Malaysia. “The value is there, but it requires the right approach.”

The report carried a number of interviews with organisations adopting AI, including with Dr Rajamani Sambasivam, chief data scientist, Petronas who said the company, which has been on its AI journey since 2017, views AI as a business tool to be used in a value-based approach, rather than as a technology tool.

“The business strategy is our AI strategy. We didn’t want to make a separate AI strategy or a digital strategy,” said Rajamani. This approach has led to over 85% of the value being delivered through digital solutions coming from AI and data science.

While Petronas has seen a good adoption of AI, sustaining these solutions over time can be a challenge. “The value for the organization doesn’t come by doing new things all the time but by consistently using the tools that have been deployed, day in and day out,” said Rajamani

To address this, Petronas has implemented a citizen analytics program that focuses on real-life problem-solving rather than theoretical training. It has upskilled over 26,000 employees and more than 5,000 have been trained to build machine learning models.

Why AI transformations fail to move beyond pilots

A key takeaway from the discussion is that AI transformation is not primarily a technology challenge.

“People often think AI transformation is a technology shift,” Vivek said. “In reality, it is a business and people transformation.”

Many organisations still fall into the trap of running multiple pilots without a clear path to scale. “One company joked they had more pilots running than an airline,” he said. “That rarely delivers value.”

McKinsey’s framework for success includes business-led prioritisation, workflow redesign, talent investment, and strong adoption at scale.

“If AI is used without reimagining processes, it simply digitises existing inefficiencies,” Vivek said. He added that companies need a “rewired” approach, where AI is embedded into business transformation, supported by changes in operating models, talent, and data capabilities.

Talent remains a critical gap. While demand for engineers and data scientists continues to grow, Vidhya highlighted another essential role.

“We also need what we call translators,” she said. “People who can bridge the technology and business worlds.”

These roles are key to identifying where AI can deliver real impact, particularly in large and complex organisations. As AI evolves rapidly, both speakers also addressed the growing hype around the technology.

Vivek described the current moment as a “once-in-a-generation opportunity,” comparable to the rise of the internet and mobile technology. “At the same time, transformation is already happening with today’s tools,” he said.

Vidhya offered a more grounded view, particularly on agentic AI. “The truth usually lies somewhere in the middle,” she said. “Some processes can be fully automated, some require human oversight, and others still need humans in the lead.”