Google to Invest $15 Billion in Malaysia Data Centres in 2027–28 in Its Biggest Bet Yet on Southeast Asian AI Infrastructure

Google will spend fifteen billion dollars in Malaysia over the next two years, building new data centre campuses, adding clean power to the national grid, funding shared water infrastructure and training hundreds of thousands of Malaysians in artificial intelligence. The plan, announced in Kuala Lumpur on Wednesday, November 4, 2026, is the largest single digital infrastructure commitment the company has ever made in Southeast Asia, and it arrives at a moment when Malaysia is trying to turn itself from a consumer of digital technology into a producer of it. The scale of the pledge dwarfs the two billion dollars Google committed in May 2024 for its first Malaysian data centre and cloud region, and it signals that the country has become one of the central battlegrounds in the global race to build the computing capacity behind artificial intelligence.

The announcement matters for reasons that go well beyond one company’s capital spending. Data centres are the factories of the AI era, and where they are built decides where the jobs, the power demand, the water use and the technical skills will concentrate for decades. Malaysia has spent the past two years welcoming that investment, particularly in the southern state of Johor, while at the same time beginning to ask harder questions about how much electricity and water the industry consumes and who pays for the grid capacity it reserves. Google’s answer, at least on paper, is to pay for its own power, add more than a gigawatt of clean generation to the grid, build water infrastructure that the public will also use, and train a workforce large enough to staff an industry that barely existed in the country a decade ago. Whether those promises hold, and what they cost the country in return, is the story behind the headline figure of Google $15 billion Malaysia data centres 2027 2028.

1. What Google Announced in Kuala Lumpur

1.1 The $15 billion figure in context

The headline number is fifteen billion US dollars, or roughly sixty-one billion Malaysian ringgit, to be spent across 2027 and 2028. It is a plan for future spending, not money already spent, and Google has not published a detailed breakdown of how the sum divides between data centre construction, clean energy, water projects and training. That distinction matters, because a two-year investment pledge and a two-year construction programme are not the same thing, and the facilities themselves will take years to build, fill with servers and reach full operation. What the figure does establish is the order of magnitude: at seven and a half times the size of Google’s first Malaysian commitment in 2024, it is the company’s biggest bet on a single Southeast Asian market, and one of the largest private digital infrastructure pledges ever made in Malaysia.

The timing of the spending is concentrated in the two calendar years named in the announcement, with construction activity expected to peak across 2027 and 2028. Google says the money will expand its existing campus at Elmina Business Park in Selangor and fund new campuses in Selangor and Negeri Sembilan, while also paying for the power, water and grid-connection infrastructure those campuses need. The company has framed the investment as foundational computing capacity, the physical layer on which cloud services, AI tools and digital services for Malaysian businesses and consumers will run. For a country whose government has made digital transformation a centrepiece of its economic strategy, the pledge lands as both a vote of confidence and a test of whether the infrastructure around it — power, water, planning permission and skilled labour — can keep pace.

1.2 Who announced it and where

The commitment was announced by Ruth Porat, president and chief investment officer of Alphabet and Google, at an event in Kuala Lumpur attended by Prime Minister Anwar Ibrahim. The event, held at Universiti Malaya under the banner Mantap Malaysia Bersama AI, brought together government ministers, university leaders and industry figures, and the Prime Minister’s presence underlined how the government wants the investment to be read: not as a routine corporate expansion, but as a cornerstone of Malaysia’s push to become a producer of digital technology rather than merely a consumer of it. Porat said the plan was intended to provide the foundational computing capacity to deliver the upside of innovation across the country, while adding local power and water resources as the company’s operations scale.

The choice of venue was deliberate. A university campus, rather than a hotel ballroom or a construction site, put the skills and training element of the announcement on the same stage as the concrete and the capital. Anwar, for his part, presented the clean energy element as coming without public subsidies, a point designed to answer critics who worry that data centres consume public resources while their benefits flow mainly to foreign companies. More than eighty Malaysian entities are involved in developing and operating the infrastructure, according to Google, a figure that spreads the work across local construction firms, utilities, suppliers and service providers. The announcement was set out in a Google Malaysia blog post on November 4, and the Prime Minister’s Office published the text of the Prime Minister’s speech the same day.

2. Where the Campuses Will Go

2.1 Selangor: Elmina, Puncak Alam and Serendah

The centre of gravity for Google’s Malaysian expansion remains Selangor, the state that wraps around Kuala Lumpur and has become the country’s main corridor for high-technology industry. Google’s first Malaysian data centre campus, at Elmina Business Park in Selangor, is now operational, having been funded by the two-billion-dollar commitment announced in May 2024. That campus also hosts the Google Cloud region for Malaysia, which is still to launch for enterprise customers, and the new investment will expand the Elmina site while adding capacity elsewhere in the state. Local technology reporting names Eco Business Park in Puncak Alam and the UMW High Value Manufacturing Park in Serendah as the locations for the new Selangor campuses, both in industrial zones that the state government has been promoting for advanced manufacturing and logistics.

Selangor’s appeal is straightforward: it has the land, the grid connections, the workforce and the proximity to the capital that data centre operators prize, and it has been building the roads, power lines and water mains that large industrial users need. The computing capacity built in Selangor will support the upcoming Malaysian cloud region, giving local businesses access to cloud and AI services hosted inside the country rather than routed through Singapore or further afield. That matters for companies and government agencies that want their data stored domestically, and for the latency-sensitive AI workloads that work best when the servers sit close to the users. What Google has not disclosed is the electrical capacity of the new campuses in megawatts, which makes it hard to judge the scale of the build against the power grid it will draw on, or to compare it precisely with rival operators’ projects in the same state.

2.2 Negeri Sembilan: Springhill and Port Dickson

The second half of the expansion moves south into Negeri Sembilan, where a new campus is planned at Springhill Industrial Park in Port Dickson, a coastal town better known for its beaches and its port than for server halls. The choice of Port Dickson reflects a wider pattern in Malaysia’s data centre boom: as land and power grow scarcer and more expensive around Kuala Lumpur and in Johor, operators are looking further afield, and state governments outside the established hubs are competing hard to attract them. Negeri Sembilan’s gain is the most concrete sign yet that the boom is spreading beyond its original centres, and the state government has been closely involved in the water infrastructure that will serve the new campus.

The Springhill campus is also where Google’s water commitments become most visible, with a shared reservoir and treatment plant planned along the Linggi River to serve both the campus and the public. That dual-purpose design is central to how Google is presenting the project: infrastructure that the company needs, but that the state keeps. The campus itself will be built in an industrial park designed for heavy industry, which means the planning, zoning and utility work is already further advanced than it would be on a greenfield site. Even so, the final operational dates for the new campuses have not been publicly established, and the sequence in which Selangor and Negeri Sembilan come online will determine how quickly the promised jobs and economic activity actually arrive.

3. The Power Question: More Than a Gigawatt of Clean Energy

3.1 The Tenaga Nasional partnership

Electricity is the single largest running cost of a data centre and the single largest constraint on where one can be built, and Google’s announcement treats power as a central part of the deal rather than an afterthought. The company says it will help add more than one gigawatt of new clean-energy generating capacity to Malaysia’s national grid, working with the national utility Tenaga Nasional Berhad under the Corporate Renewable Energy Supply Scheme. A gigawatt is roughly the output of a large conventional power station, and adding that much clean generation would be a meaningful contribution to a grid that is still heavily dependent on coal and gas. Google has also committed to paying for all the electricity its campuses consume, and for the direct infrastructure costs of connecting them to the grid.

The clean energy will be developed through power purchase agreements with a consortium led by Shizen Energy, Solarvest and TotalEnergies, which will build the solar and other renewable capacity that feeds the grid. That structure matters because it means Google is not simply buying certificates or offsets; the agreements are intended to fund new generation that would not otherwise be built. Anwar said the additional clean power would come without public subsidies, a claim that, if it holds, answers one of the most persistent criticisms of data centre investment: that taxpayers end up underwriting the power infrastructure for private profit. The precise timeline for the one-gigawatt target has not been published, and neither has the mix of technologies that will deliver it, so the pace at which the clean capacity arrives remains an open question.

3.2 Paying for reserved capacity

The power commitment lands at a sensitive moment in Malaysian energy policy. Over the past month, ministers have told parliament that the era of cheap, plentiful power for data centres is ending, and that operators will pay for grid capacity they reserve but leave idle. Analysis of utility data suggests operators currently use only about a fifth of the grid capacity they have reserved, which means the grid is built out — at public expense — for demand that never arrives, while other users wait for connections. The government’s plan to bill operators for the capacity they reserve, whether or not they use it, is designed to end that free option, and it changes the economics of every data centre project in the country, including Google’s.

Google’s pledge to pay for its own electricity and connection costs fits neatly into that new policy direction, and it may be no coincidence that the announcement emphasised the point. But the commitment also raises a question the announcement does not answer: how much capacity will Google’s new campuses actually reserve, and how much of it will they use? Without a published megawatt figure for the campuses, neither the government nor the public can judge whether the company’s power use will be efficient or wasteful. The Johor experience looms over the discussion: the southern state has absorbed most of Malaysia’s data centre boom, and its experience with surging power demand, rising land prices and strained water supplies is the cautionary tale that the new rules are meant to address.

4. Water, Efficiency and the Tropical Data Centre

4.1 The $160 million Linggi River project

Water is the less visible half of the data centre equation, and in a tropical country it is the half that most directly affects the people who live near the servers. Data centres use large volumes of water for cooling, and in a hot, humid climate that demand is constant. Google says it will invest one hundred and sixty million dollars in shared water infrastructure in Negeri Sembilan, and has appointed Malaysian construction firm Gamuda to build an off-river storage reservoir and a water treatment plant along the Linggi River, in coordination with the state government and the state water company. Once finished, the plant is expected to produce sixty-five million litres of treated water a day, and the facilities will be handed to the state government when complete.

The design is deliberately dual-purpose: the reservoir and treatment plant will serve Google’s planned Springhill campus, but they will also serve the public water supply, which means the investment leaves behind infrastructure that the state owns and operates. That is a materially different model from a private borehole or a dedicated pipeline that serves only the data centre, and it is the part of the announcement most likely to matter to the residents of Port Dickson and the surrounding districts. The Linggi River is a working river that already supplies towns and agriculture, so the engineering, the environmental approvals and the allocation of water between the campus and the public will all be watched closely. Gamuda’s appointment gives the project a contractor with deep experience in Malaysian water infrastructure, but the construction timetable and the date the plant begins delivering water have not been published.

4.2 A PUE target of 1.12 to 1.14

Alongside the water investment, Google has set an efficiency target for its Malaysian campuses that it presents as a meaningful advance on the industry norm. Power Usage Effectiveness, or PUE, measures how much of the electricity entering a data centre is used by the computing equipment itself rather than by cooling, lighting and other overheads. A PUE of 1.0 would mean every watt goes to computing; the industry average is around 1.54, which means more than a third of the power is consumed by everything around the servers. Google says its Malaysian campuses are designed to operate at a PUE of 1.12 to 1.14, a figure that, if achieved in practice, would put them among the most efficient large data centres operating in a tropical climate.

The target is ambitious precisely because Malaysia’s climate works against it. Cooling is hardest where the air is hot and humid year-round, and many operators in Southeast Asia accept higher PUE figures as the price of the location. Google’s claim rests on its cooling design, its use of outside air where conditions allow, and the efficiency of its server and power-distribution hardware, all of which have been refined across its global fleet. The number is a design target rather than a measured result, and it will only be proven once the campuses are running at full load in Malaysian conditions. Independent verification of the figure, once the facilities are operational, will be the real test, and it is a test the company has invited by publishing the number at all.

5. Jobs, Wages and the GDP Claim

5.1 The construction-phase numbers

The economic case for the investment rests on a set of projections that Google itself has supplied, and they need to be read with that authorship in mind. The company projects that the construction phase, running across 2027 and 2028, will contribute an average of one point one billion dollars a year to Malaysia’s gross domestic product, and support more than thirty-seven thousand local jobs a year, including roughly six thousand construction roles, with average annual labour income of three hundred and seventy-eight million dollars. Those are large numbers, and they are the figures most likely to be quoted by ministers and most likely to be remembered by the public. They are also, by Google’s own framing, projections, not independently audited estimates, and they describe a two-year construction window rather than a permanent state of affairs.

The construction workforce is the most immediate and visible part of the jobs story. Six thousand construction roles across the two years means work for builders, electricians, engineers, project managers and the suppliers who feed a large industrial build, and the wider figure of thirty-seven thousand jobs a year includes the indirect employment that construction spending supports in transport, catering, materials and services. With more than eighty Malaysian entities involved in developing and operating the infrastructure, the spending is intended to spread beyond a single contractor or a single state. The question that follows any construction boom, however, is what remains when the cranes leave, and that is where the longer-term jobs picture becomes the more important measure.

5.2 The long-term jobs picture

Once the campuses are running, the employment profile changes completely. Google cites just over nine thousand long-term jobs once the facilities are operational, at wages it says will average forty-five percent above Malaysia’s median. Nine thousand permanent roles is a substantial number, but it is a quarter of the construction-phase figure, and it reflects the reality of modern data centres: they are capital-intensive and highly automated, and they employ far fewer people per dollar invested than a factory or an office park. The wage premium matters more than the headcount. Jobs paying nearly half as much again as the national median — in operations, engineering, security, facilities management and the cloud services that run on the infrastructure — are the kind of employment Malaysia’s economic planners want to create, and they are the jobs that anchor a skilled workforce in the country.

The GDP figure deserves the same careful reading. An average contribution of one point one billion dollars a year during construction is Google’s own estimate of the value added by the building programme, and it will fall away once construction ends unless the operating phase generates its own, smaller but steadier, stream of economic activity. The company has not published a long-term GDP figure for the operating phase, and it has not said how much of the spending will go to Malaysian firms versus imported equipment and foreign contractors. Those gaps do not make the projections wrong, but they do mean the headline numbers should be treated as the company’s case for its investment rather than as a settled forecast, and judged against the results as the campuses come online.

6. The Skills Bet: 800,000 Malaysians by 2031

6.1 What the training covers

The most far-reaching element of the announcement may be the one with the smallest price tag attached to it. Google says it will upskill approximately eight hundred thousand Malaysians between 2027 and 2031, training people in artificial intelligence and digital skills through programmes that run alongside the construction of the campuses. The scale of the ambition — nearly a million people over five years — is striking, and it reflects a recognition that the binding constraint on Malaysia’s digital ambitions may not be servers or power, but people who know how to use them. The training is aimed at a broad population rather than only at the engineers who will staff the data centres, and it is presented as part of the same package as the infrastructure spending.

Details of the curriculum, the delivery partners and the certification that trainees will receive have not been set out in full, and the difference between a short online course and a substantive qualification is large. Google has run AI training programmes in other markets, and the Malaysian programme is expected to draw on that experience, working with universities, vocational institutions and employers. The choice of Universiti Malaya as the launch venue points to the higher-education sector as a central partner, and the five-year window to 2031 gives the programme time to build beyond an initial cohort. Whether eight hundred thousand people complete meaningful training, or whether the figure counts every registration and every short module, will determine how the promise is judged.

6.2 Why skills matter as much as servers

The skills commitment matters because it addresses the most persistent criticism of data centre investment: that the buildings arrive, the power is consumed, and the high-value work stays elsewhere. A data centre on its own creates relatively few jobs and transfers relatively little knowledge. A data centre paired with a workforce that can build AI applications, run cloud services and start digital businesses is a different proposition, and it is the proposition Malaysia’s government has been pursuing through its Malaysia MADANI framework and the Malaysia Digital 2030 Action Plan unveiled in June 2026. Those frameworks set out the ambition of making Malaysia a producer of digital technology, and a trained workforce is the precondition for that shift.

The bet also reflects a competitive reality. Singapore, Indonesia, Thailand and Vietnam are all courting the same investment, and the countries that win the next wave of data centre spending will be the ones that can offer not just land and power but engineers, technicians and AI-literate graduates. Training eight hundred thousand people is Google’s way of helping to create the workforce its own campuses and its customers will need, and it aligns the company’s interests with the government’s. The risk is that the training is measured by enrolments rather than by employment outcomes, and that the people trained find no local jobs that use their new skills. The test, as with the jobs figures, will be in the results reported in 2028 and beyond, not in the announcement.

7. Why Malaysia, and Why Now

7.1 The data centre boom and Johor’s shadow

Malaysia’s data centre boom did not begin with Google’s announcement, and understanding the announcement means understanding the market it joins. Over the past three years the country has become one of the fastest-growing data centre markets in Asia, driven by its proximity to Singapore, its relatively abundant land, its competitive power prices and a government that has actively courted the industry. Johor, the southern state that borders Singapore, has absorbed most of that growth, with operators drawn by the chance to serve Singaporean demand from across the causeway at a fraction of Singaporean costs. Singapore’s own moratorium on new data centres, imposed while it reviewed the industry’s power and land use, pushed a wave of investment south, and Malaysia was the principal beneficiary.

That boom has brought investment, construction jobs and a new industry to the country, but it has also brought the strains that now shape the policy debate. Power demand in Johor has surged, land prices have risen, and water supplies have come under pressure, while residents and businesses have begun to ask what the state gets in return for hosting infrastructure that mainly serves customers elsewhere. The government’s new rules on paying for reserved grid capacity are a direct response to those strains, and Google’s announcement — with its emphasis on paying for power, adding clean generation and building shared water infrastructure — reads as an attempt to show that the next wave of investment can be built on different terms. Notably, Johor was not named among the locations for Google’s new campuses, a detail that may reflect the state’s growing constraints as much as it reflects the opportunities in Selangor and Negeri Sembilan.

7.2 The scrutiny Google faces elsewhere

Google’s expansion in Malaysia is also unfolding against a backdrop of scrutiny of the company’s data centre programme in other countries. In October, a Finnish authority ordered work halted on two Google data centre sites until mandatory environmental impact assessments are completed, a reminder that the environmental cost of large-scale computing is now a live regulatory issue in Europe as well as in Asia. The Finnish case turned on process — whether the proper assessments had been done before construction began — but it illustrates the direction of travel: data centres are no longer treated as ordinary industrial buildings, and their approvals are increasingly contested on climate, water and land-use grounds.

That scrutiny is one reason the Malaysian announcement leans so heavily on clean energy, water and training. The company is presenting a model of data centre development that anticipates the objections rather than reacting to them, and it is doing so in a market where the government is receptive to the argument. The contrast with the Finnish experience also underlines how much depends on the regulatory environment: in Malaysia, the government is an active partner in the expansion, while in Finland the authorities have shown they are willing to stop work. Both approaches are shaping the same global build-out, and the Malaysian model — government partnership, shared infrastructure, published efficiency targets — will be watched by other countries weighing how to host the industry.

8. What Remains Unconfirmed and What to Watch Next

8.1 The numbers Google has not published

For all its headline figures, the announcement leaves several important numbers unpublished, and those gaps limit how precisely the investment can be judged. Google has not disclosed the electrical capacity of the new campuses in megawatts, which means the scale of their power demand, and the adequacy of the one gigawatt of new clean generation, cannot be independently assessed. It has not published a breakdown of the fifteen billion dollars across campuses, power, water and training, so it is not possible to say how much of the money is construction spending and how much is longer-term investment. It has not given a launch date for the Malaysian Google Cloud region, nor a timeline for delivering the one gigawatt of clean energy, nor the operational dates for the new campuses in Selangor and Negeri Sembilan.

The jobs and GDP figures carry their own caveats. They are Google’s projections, not independent estimates, and they describe the construction phase rather than the steady state. The long-term figure of just over nine thousand jobs, and the wage premium of forty-five percent above the median, are likewise company figures that have not been independently verified. The training figure of eight hundred thousand people by 2031 counts a programme whose curriculum, delivery partners and certification have not been fully described. None of this makes the announcement unreliable, but it does mean the numbers should be reported as Google’s case for its investment, with the company’s authorship stated plainly, and that the real verdict will come from the results reported as the campuses are built and opened.

8.2 What to watch next

The next twelve months will show whether the announcement becomes a construction programme or remains a pledge. The first markers to watch are the planning and environmental approvals for the new campuses in Selangor and Negeri Sembilan, and the start of construction at Puncak Alam, Serendah and Springhill. The power purchase agreements with the Shizen Energy, Solarvest and TotalEnergies consortium will show how quickly the clean energy capacity is actually contracted, and Tenaga Nasional’s grid-connection work will show whether the infrastructure keeps pace with the build. Gamuda’s work on the Linggi River reservoir and treatment plant will be the most visible early test of the water commitment, and the date the plant begins delivering treated water will matter more than the date it is announced.

Beyond the construction milestones, the numbers to watch are the ones Google has projected: the jobs created during the build, the wages paid, the GDP contribution reported, and the enrolments and outcomes of the training programme. The Malaysian Cloud region’s launch date will show whether the computing capacity is reaching customers, and the published PUE figures, once the campuses are running, will show whether the efficiency target was met in practice. The government’s new rules on reserved grid capacity will also be tested by this project, and how Google’s power use is billed under them will set a precedent for every operator that follows. The announcement is the beginning of the story, and the construction sites, the power contracts and the training classrooms are where it will be decided. Until then, the fifteen billion dollars remains a plan, and the plan’s value will be measured in what is built, what is powered, what is delivered in water, and who is trained to use it.

Comments

comments