• 10 months ago
AI-driven data centres are reshaping Malaysia’s electricity demand more concentrated, more intensive, and less predictable. This discussion breaks down what grid upgrades, real-time data capabilities, and regulatory reforms are needed to keep reliability high while scaling renewables and cross-border power trade.

Category

🗞
News
Transcript
00:00Malaysia's power grid is entering a stress test it was never originally designed for.
00:04The search of AI-driven data centers is not just adding demand,
00:07it is concentrating it, intensifying it and making it far more dynamic than traditional consumption patterns.
00:13At the same time, the system itself is evolving from one-way power delivery to a two-way data-driven,
00:19renewable-heavy network shaped by rooftop solar, storage and cross-border electricity trade.
00:24The real question is no longer whether demand will grow but whether the grid can become intelligent,
00:28flexible and predictive fast enough.
00:31Because if modernization keeps space, AI data centers can become strategic grid assets.
00:37If it doesn't, they risk becoming systemic pressure points on reliability, price stability and energy security.
00:44To discuss this further and deepen our understanding, we take a closer look at what must change technically,
00:49structurally and form a policy standpoint to ensure the grid is future ready.
00:53And together with us today, Mr. Jason Ong, General Manager for Asia Gentrack Limited.
00:59Okay Jason, to start off, maybe we want to start by saying that Malaysia is seeing a rapid influx of AI-driven data centers.
01:05How is this changing where, when and how intensively electricity demand shows up on the grid?
01:11And why does this challenge the way our grid was originally designed?
01:15Yeah, salamat pagi.
01:18So first of all, the AI and data center boom is happening globally, not just in Malaysia.
01:25So countries like the US, India are growing by almost 70 percent,
01:30while in this region, Thailand and Vietnam are increasing their number of data centers by more than 50 percent.
01:36And Malaysia itself is adding 30 new data centers on top of the existing 125 that are operating today,
01:44mostly across Kuala Lumpur and Johor.
01:47Now, each data center requires up to 100 megawatts of electricity,
01:52and they run 24 hours a day, 365 days a year.
01:57And 100 megawatts of electricity is the equivalent of how much it takes to power 100,000 residential homes
02:03or up to 50 shopping malls.
02:06So in summary, with Malaysia adding another 30 data centers within the next 10 years,
02:12that is the equivalent of an additional 3 million residential homes
02:17or up to 150 more Aon shopping malls.
02:21And in the meantime, more homes will continue to be built,
02:24shopping malls will continue to be developed.
02:26So you can imagine the amount of electricity that the power grid needs
02:30in order to deliver to keep up with the electricity demand.
02:34And the Energy Ministry, Petra, they have projected that data centers alone
02:40could require more than 20 gigawatts of electricity by 2040.
02:45So Malaysia's power system is looking at one of the fastest demand growth across Southeast Asia
02:50and facing one of the biggest challenges in history.
02:54Yeah, and traditionally, power grids were built for one-way delivery
02:58from centralized generation to consumers.
03:00With more renewables, rooftop solar and regional power trade,
03:04what structural changes must Malaysia's grid undergo
03:06to operate as a two-way intelligence system then?
03:11Yeah, so the grid is becoming much more complex than before.
03:16Traditionally, with the one-way infrastructure,
03:18you have a group of power plants.
03:20They generate and feed electricity into the grid.
03:23It goes through the transmission lines
03:25and eventually into the homes and the factories and the buildings.
03:28But this is not sufficient if you are talking about coping
03:32with the current rising demand for electricity
03:35due to the rising number of data centers.
03:39So we could build more power plants to solve this problem,
03:43but that would just bring us further away from our net zero goals.
03:47So with the introduction of solar energy powering the buildings
03:52and feeding back to the grid,
03:54as well as the adoption of new renewable energy sources such as hydrogen
04:00and also with regional cross-border interconnections between countries,
04:05the power grid has to become more modernized
04:07in order to cope with the stress of managing the flow of electricity
04:12across these new sources of power supply.
04:14And the modernization of the grid requires real-time data.
04:21It's all about data.
04:23Now, ironically, while the underlying cause of the increase
04:26in the number of data centers is the massive growth of data
04:29and data-driven use cases,
04:33the key enabler to balance the demand and supply of energy
04:38is the availability of real-time data.
04:41Because with real-time data, we can make faster and smarter decisions
04:47to manage the volatility and the spikes in energy demand
04:51because real-time data enables what we call predictive balancing.
04:56Predictive balancing allows us to dispatch energy that is stored,
05:00perhaps in battery energy storage systems
05:03that have been powered by solar energy.
05:05Predictive balancing allows us to adjust energy imports.
05:10And more importantly, it provides us with the ability
05:14to promote what we call demand flexibility.
05:18Now, if we're trying to achieve our net zero goals
05:21only by tackling the supply side of the equation,
05:26it is like a dog chasing its tail.
05:28It's endless and it's a futile effort.
05:31So, we need to do more to tackle the demand side of the equation
05:36and ever more important now with the increase in the data centers.
05:42So, for instance, we know that the large commercial buildings
05:46and industrial factories,
05:47they take up more than 70% of the demand for energy.
05:52And so, for a start, the Naga National or DNB,
05:55they provide time of use charging for businesses,
05:58making it cheaper for electricity usage during the off-peak hours
06:03from 10 p.m. to 2 p.m. on weekdays
06:06and full day on weekends and holidays.
06:08So, this means businesses can shift their power consumption
06:13towards the off-peak periods by adjusting the operations accordingly.
06:17And this is to help reduce the peak load of electricity
06:20and it also helps them save on energy costs.
06:24But I still believe that more can be done
06:28by introducing perhaps more pricing layers
06:31and more tariff structures.
06:35And now, AI heavy data centers are often seen as major demand risk.
06:39What would it take in terms of technically and commercially
06:42to turn them into flexible grid assets
06:44that can respond to real-time conditions
06:46instead of just drawing constant power?
06:48Yeah. So, with more real-time data,
06:54we actually have more accurate forecasting
06:57of power supply and demand
06:59and we can have more robust orchestration of the power grid.
07:05It also means that the key stakeholders
07:08across Malaysia's energy sector,
07:11such as the Energy Ministry, Petra,
07:15the Energy Commission, ST,
07:16and TNB,
07:18they can now have the ability
07:21for better planning and decision-making
07:23because now there is less room for error
07:27with these AI heavy data centers running 24 by 7.
07:34Traditionally, the operators,
07:37they tend to be more reactive.
07:41The problems are only detected
07:43when the energy reserves are thin.
07:46During, you know, surges in electricity demand,
07:51it could be outages caused by unexpected conditions
07:54such as weather factors or operational issues.
07:58So, with the availability of real-time data
08:02and better forecasting,
08:05the operators can now take a more proactive approach
08:08by introducing measures
08:10to address the risk of these problems
08:12before they take place.
08:14For example,
08:16there are technology solutions out there,
08:18such as Gentrack,
08:20that can help the stakeholders
08:21and the operators leverage
08:23on more effective tariffs
08:26and grid orchestration.
08:29They can potentially help shift the load
08:32away from the non-critical consumption
08:35of electricity during supply shortages
08:37or during the demand spikes.
08:40And this is in order to keep the lights on
08:43for the critical consumers,
08:46such as these AI-driven data centers,
08:49public services, hospitals,
08:52vulnerable consumers,
08:53those who are on life support systems.
08:56And then there is the introduction
08:58of battery energy storage systems
09:01that are powered by solar energy.
09:03And these can be a good source of power
09:05during the peak periods
09:07because energy can then be drawn from them
09:10to power the buildings and the houses
09:13and can be sold back to the power grid
09:16when it is not utilized.
09:18So just to give some examples,
09:21in Australia,
09:23energy retailers,
09:24they have used the Gentrack technology
09:27to enable them to bundle
09:29their solar panels
09:30and battery energy storages
09:32as part of their energy plans.
09:35So think about it,
09:36like the telcos
09:37and how they bundle the mobile phones
09:39with the telco plans.
09:41It's the same thing,
09:42same use case.
09:43So another example in Australia
09:47is how energy retailers
09:49have used Gentrack
09:50to create what we call
09:52virtual power plants.
09:54So virtual power plants
09:55are basically a network
09:57or aggregation
09:59of smaller scale
10:00distributed energy resources
10:03such as home solar panels,
10:06battery storage,
10:08or even electric vehicles.
10:10You know that
10:11a fully charged electric vehicle
10:13can actually power a home
10:15for 24 hours.
10:17So by the use of technology
10:19such as Gentrack,
10:20they can help
10:21all these distributed energy resources
10:25operate as one single
10:27flexible power plant.
10:29One example is
10:31Ember Energy in Australia.
10:33They are one of these providers
10:35and operators
10:36of a virtual power plant
10:38through what they call
10:39their SmartShift program.
10:41And one last example
10:43is in Singapore.
10:44One of our customers,
10:46Sunoco Energy,
10:47they offer
10:48peer-to-peer energy
10:50trading services
10:51called SolarShare.
10:53It allows their consumers,
10:54their customers,
10:55to buy and sell
10:56electricity
10:57with other businesses
10:59who also have
11:00solar assets.
11:02So lots of examples,
11:03lots of use cases.
11:04we know,
11:05we know that
11:23we're going to
11:27start

Recommended