Artificial intelligence has been the defining investment theme of the past few years. Companies at the centre of the AI revolution have delivered extraordinary returns, with Nvidia emerging as the poster child of the boom. As demand for graphics processing units (GPUs) exploded, Nvidia became one of the world’s most valuable companies, while semiconductor manufacturers, cloud providers and hyperscale technology firms also enjoyed massive gains.
The enthusiasm has been driven by an unprecedented wave of investment into AI infrastructure. Technology giants including Microsoft, Amazon, Alphabet and Meta continue to spend hundreds of billions of dollars building AI data centres, purchasing advanced chips and expanding cloud computing capacity. AI-related companies have consistently outperformed broader markets, reinforcing investor confidence that the technology remains in its early stages of growth.
However, according to DBS, the next phase of the AI boom will not be determined solely by who builds the fastest chips or the smartest AI models. Instead, the biggest constraint may be something far less glamorous: electricity.
The bank believes that investors should continue to stay bullish on technology, but increasingly diversify into energy infrastructure as a critical enabler of the AI economy.
AI’s biggest challenge is no longer computing power
Speaking at DBS’ Chief Investment Office media briefing, Chief Investment Officer Hou Wey Fook described energy as the “chokepoint” for AI’s next stage of growth.
While AI models continue to become more powerful, every improvement comes with significantly higher electricity consumption. Training large language models, operating AI-powered search engines and running inference workloads around the clock all require enormous amounts of energy.
Hou pointed to comments by Nvidia Chief Executive Jensen Huang, who has said that the next era of computing could require up to 1,000 times more power than current systems.
That presents a significant challenge.
While semiconductor technology continues to improve rapidly, electricity generation, transmission networks and power infrastructure cannot be expanded overnight. Without sufficient power capacity, many planned AI data centres could face delays regardless of how many advanced chips are available.
In other words, AI may soon become constrained not by computing hardware, but by the availability of reliable electricity.
Why energy infrastructure matters
DBS expects approximately US$1 trillion of annual AI-related capital expenditure over the coming years.
Much of this spending will naturally go towards semiconductors, servers and networking equipment. However, every AI data centre also requires enormous supporting infrastructure.
This includes:
- Electricity generation
- Transmission grids
- High-voltage transformers
- Battery storage systems
- Cooling infrastructure
- Backup power solutions
- Nuclear and renewable energy projects
As AI adoption accelerates, demand across the entire energy value chain is expected to increase.
The bank also believes that governments are once again prioritising energy security, creating additional long-term investment opportunities beyond AI alone.
DBS recommends a layered investment strategy
Rather than focusing exclusively on technology stocks, DBS recommends investors build exposure across different parts of the energy sector.
Investment strategist Goh Jun Yong described this as a “layered” approach.
The first layer consists of major integrated oil companies.
These companies continue to generate strong cash flows while benefiting from sustained global energy demand. Although renewable energy continues to expand, traditional hydrocarbons are still expected to play an important role in meeting near-term electricity needs.
The second layer involves companies responsible for generating and transmitting electricity.
These include utilities, power producers and businesses involved in expanding electricity grids. As governments modernise ageing infrastructure and connect new renewable projects to the grid, transmission companies are expected to benefit from increased investment.
The final layer focuses on alternative energy and specialised infrastructure manufacturers.
This includes renewable energy developers, battery storage providers, uranium producers and companies involved in nuclear power. DBS believes these areas could experience significant growth as countries seek cleaner and more reliable sources of electricity to support rising AI demand.
Nuclear power and uranium may benefit
One of the more notable themes highlighted by DBS is nuclear energy.
Unlike intermittent renewable sources such as solar and wind, nuclear power provides stable baseload electricity around the clock. This makes it particularly attractive for energy-intensive AI data centres that require uninterrupted power supplies.
DBS also suggested uranium exposure could form part of investors’ alternative energy allocation.
Growing interest in extending the lifespan of existing nuclear reactors and building new facilities has improved the outlook for uranium demand in several major economies.
While nuclear projects typically require long development timelines, they could become increasingly important as AI electricity consumption continues rising.
Singapore companies could benefit
Although many of today’s AI winners are listed in the United States or Taiwan, DBS believes opportunities also exist closer to home.
Senior investment strategist Joanne Goh highlighted Singapore-listed utilities, renewable energy companies, clean energy businesses, REITs and firms involved in power infrastructure.
She noted that Singapore companies have yet to fully benefit from AI-driven infrastructure spending across the value chain.
However, current valuations appear attractive, particularly if infrastructure orders begin to recover.
As AI data centre construction expands across Asia, businesses supplying electrical equipment, engineering services and power infrastructure could see stronger demand over the coming years.
Technology remains a core investment theme
Despite recommending greater exposure to energy, DBS is not turning bearish on technology.
In fact, the bank continues to see AI as one of the strongest long-term structural investment themes.
According to DBS, AI-related companies contributed around 80% of total equity market returns, underlining how dominant the sector has become.
The bank also noted that US corporate earnings have remained robust, led by companies directly benefiting from AI adoption. Meanwhile, valuation multiples have moderated somewhat as earnings have grown, helping alleviate concerns that the sector has become excessively expensive.
Within Asia, Taiwan and South Korea continue to outperform thanks to their leading positions in semiconductor manufacturing.
Demand for advanced chips remains exceptionally strong as cloud providers continue expanding AI infrastructure worldwide.
The next generation of chips will focus on energy efficiency
One interesting observation from DBS is that future semiconductor innovation may increasingly prioritise power efficiency rather than simply delivering higher computing performance.
As electricity becomes scarcer and more expensive, AI developers will seek processors capable of performing more computations while consuming less power.
This could create opportunities for semiconductor companies specialising in low-power designs, advanced manufacturing processes and energy-efficient architectures.
Power-efficient computing may become one of the defining competitive advantages in the next stage of AI development.
Indonesia remains a value opportunity
Beyond AI, DBS also remains constructive on Indonesia despite recent market weakness.
The Jakarta Composite Index has declined sharply this year amid investor uncertainty surrounding market classification reviews, oil prices and inflation.
However, DBS believes investor confidence could recover once these near-term uncertainties subside.
The bank favours Indonesian banks, telecommunications companies, utilities and commodity producers, particularly those supplying materials used in electricity generation and infrastructure development.
Given current valuations, DBS believes foreign investment could return if market sentiment improves.
The bottom line
The AI investment story is evolving.
For the past few years, investors have largely focused on companies designing chips, developing AI software or building cloud platforms. Those businesses remain attractive, and DBS continues to recommend maintaining long-term exposure to technology.
However, the bank believes the next wave of AI investment opportunities will extend far beyond Silicon Valley.
Every AI model requires electricity. Every AI data centre requires reliable power, transmission networks and supporting infrastructure. As AI adoption accelerates globally, the companies supplying that energy may become just as important as the companies building the technology itself.
For investors, that means the AI theme is no longer simply about owning semiconductor stocks such as Nvidia. It is increasingly about owning the infrastructure that makes the AI revolution possible.
DBS’ message is clear: remain structurally bullish on technology, but complement those holdings with energy infrastructure investments that stand to benefit from the enormous power demands of the AI economy.