The Next Data Centre Bottleneck Isn't Computing Power. It's Power.

← Back to insights

Industry Insights

The race to build artificial intelligence infrastructure has largely been defined by computing power. Faster chips. Larger GPU clusters. More sophisticated cooling systems. Bigger data centres.

But another race is taking place, one that may ultimately determine how quickly AI infrastructure can be deployed.

The race for power.

A data centre can have the best location, the most advanced technology and customers ready to occupy it. Yet without sufficient, reliable electricity, it remains a development project rather than a productive asset.

The question is no longer simply how much computing capacity we can build.

It is how much computing capacity we can actually power.

AI is changing the electricity equation

The scale of the challenge is becoming increasingly clear.

According to the International Energy Agency (IEA), data centres consumed approximately 415 terawatt-hours (TWh) of electricity worldwide in 2024, representing around 1.5% of global electricity consumption. Its 2025 Energy and AI report projected that consumption would more than double to approximately 945 TWh by 2030.

The IEA's subsequent 2026 analysis reported that data centre electricity demand increased by 17% in 2025, with AI-focused facilities growing even faster.

These figures point to a structural change in the relationship between technology and energy.

Historically, electricity was often treated as one of several infrastructure requirements in data centre development. Increasingly, it is becoming a determining factor in investment decisions, site selection and project delivery timelines.

AI workloads are particularly demanding. High-density computing clusters require substantial electricity, advanced cooling and resilient electrical distribution systems.

And unlike many conventional industrial activities, data centres are expected to operate continuously, with extremely high standards of availability.

For developers and operators, this changes the equation.

Power is no longer simply a utility connection. It is a strategic development consideration.

The real constraint may be outside the data centre

Constructing a data centre and connecting it to the electricity network are two very different undertakings.

The building may be ready before the grid is.

A data centre can be developed within a few years, while the transmission infrastructure, substations and generation investments needed to support it may require considerably longer.

The IEA estimates that grid constraints could delay around 20% of planned global data centre capacity additions through 2030.

The challenge extends beyond the availability of electricity itself. It includes grid connection approvals, network reinforcement, transformers, switchgear and other critical electrical equipment.

These constraints are especially significant in locations where multiple large data centre developments are competing for access to the same electrical infrastructure.

This creates an important commercial distinction.

A developer may secure land and obtain planning approvals, but that does not automatically establish when sufficient power can be delivered to the site.

For customers, the distinction is even more consequential. A commitment to lease capacity is only as valuable as the confidence that the facility can be energised and commissioned when required.

In the AI economy, securing land gives you a location. Securing reliable power gives you a project.

Malaysia's opportunity comes with an energy challenge

Malaysia has emerged as an important destination for digital infrastructure investment, supported by its strategic location, industrial ecosystem, connectivity and growing demand for cloud and AI services.

The scale of the industry's electricity requirements is already visible.

According to information provided by Malaysia's Ministry of Energy Transition and Water Transformation (PETRA) in July 2026, TNB was supplying electricity to 36 operating data centres with approximately 4.5 gigawatts (GW) of planned supply capacity as of the first quarter of 2026. Another 23 data centre projects under construction represented approximately 3.8 GW of maximum demand capacity.

These figures demonstrate the scale of the pipeline, although planned supply capacity should not be confused with actual electricity consumption.

TNB's own reporting showed that actual data centre load utilisation stood at approximately 850 MW at the end of 2025.

This distinction matters. A growing pipeline of announced or contracted capacity does not mean that every megawatt is already being consumed.

Nevertheless, the direction is clear.

Malaysia must continue attracting high-value digital infrastructure investment while ensuring that electricity supply remains reliable, affordable and sustainable for the wider economy.

That means balancing the requirements of data centres with those of manufacturers, businesses, households and other emerging industries.

The objective should not simply be to approve more data centres.

It should be to develop the energy and digital infrastructure needed to support productive, sustainable economic growth.

The next generation of data centres needs a power strategy, not just a power connection

We believe successful AI infrastructure developments will increasingly distinguish themselves through four capabilities.

First, power certainty.

Developers need a realistic understanding of available capacity, grid connection requirements, infrastructure dependencies and energisation timelines.

A credible power delivery plan is fundamental to a credible project delivery plan.

Second, power resilience.

Availability is central to data centre operations. Electrical systems must be designed with appropriate redundancy, backup arrangements, protection systems and maintenance strategies to meet customers' operational requirements.

Reliability must be engineered into the facility, not assumed from the availability of a grid connection.

Third, power scalability.

AI computing requirements are evolving rapidly. Infrastructure must be designed to accommodate higher rack densities, changing cooling technologies and phased increases in electrical demand.

A facility that meets today's requirements but cannot adapt to tomorrow's workloads may face limitations well before the end of its intended operating life.

Fourth, energy sustainability and commercial competitiveness.

Energy is both an operating cost and an environmental consideration.

Renewable electricity procurement, energy efficiency, appropriate storage solutions and credible decarbonisation strategies will become increasingly relevant to customer decisions.

However, renewable energy availability and reliable round-the-clock electricity supply are not interchangeable. A credible strategy must consider intermittency, storage, firm supply and the physical capabilities of the electricity system.

There is no universal solution. Grid electricity, renewable procurement, storage and, where technically and commercially appropriate, alternative generation arrangements each have roles to play.

The winning approach will be one that balances reliability, cost, scalability and environmental performance.

Why digital infrastructure and energy can no longer be viewed separately

For much of the industry's history, data centre development and energy infrastructure planning were treated as closely related but largely separate disciplines.

That distinction is becoming harder to sustain.

Decisions about power availability influence where data centres can be developed. Decisions about electricity pricing influence their long-term competitiveness. Decisions about generation and grid infrastructure influence how quickly new capacity can enter the market.

For governments, this requires greater coordination between digital economy ambitions, energy policy, infrastructure investment and industrial development.

For investors, it means assessing the quality of a project's power strategy alongside its commercial structure, technical specifications and customer commitments.

For developers and operators, it means engaging energy stakeholders early rather than treating electricity as a problem to be resolved late in the development process.

At Perintis Akal, this relationship informs our strategic direction.

Our core focus is developing and operating AI-ready digital infrastructure, complemented by our pursuit of opportunities in energy infrastructure and solutions.

We see these sectors as increasingly interconnected. The ability to align digital infrastructure development with practical energy strategies will be an important consideration in delivering the next generation of data centres.

The next competitive advantage

The AI infrastructure industry will continue to evolve.

Computing technologies will improve. Chips will become more capable. Cooling systems will advance. Software and hardware efficiency will continue to develop.

But none of these advances eliminates the need for reliable electricity.

Indeed, improvements in computing efficiency may encourage wider AI adoption, increasing demand for digital services and infrastructure even as the energy required for individual tasks declines.

The next competitive advantage will therefore not be determined by computing capability alone.

It will also depend on the ability to secure power, develop infrastructure, deliver capacity on time and operate facilities reliably over the long term.

For Malaysia and the wider ASEAN region, this presents both an opportunity and a responsibility.

Countries that successfully align energy planning with digital infrastructure development will be better positioned to attract investment, support AI adoption and strengthen their competitiveness.

Those that fail to address the energy equation risk seeing promising digital investments constrained by physical infrastructure.

The future of AI may be written in code. But its ability to scale will depend on power.

Perintis Akal is a Malaysia-based infrastructure developer and operator focused on AI-ready data centres and digital infrastructure, with energy infrastructure and solutions as a complementary growth area.