Citi has initiated coverage of CSE Global with a Buy rating and S$1.70 target price. But the target price is not the most interesting part. The bigger story is whether this Singapore-listed engineering company is quietly becoming a picks-and-shovels beneficiary of the AI data-centre boom.

When investors think about AI infrastructure, the obvious names are semiconductor companies, cloud providers and data-centre operators.

But there is another, less glamorous part of the AI investment chain.

Electricity.

Every new AI data centre needs enormous amounts of power. That power has to be distributed, switched, controlled and delivered reliably before a single GPU can run.

This is where CSE Global is becoming increasingly interesting.

Citi has initiated coverage of CSE Global with a Buy / S$1.70 target price, compared with S$1.18 at Wednesday’s close. That represents approximately 44% upside to the analyst’s target.

But investors should look beyond the target.

The more interesting question is whether CSE is transitioning from a conventional engineering and systems-integration company into a critical power infrastructure provider for the global data-centre build-out.

If that transition succeeds, the market could eventually value CSE very differently.

CSE is riding the physical side of the AI boom

The AI investment boom ultimately requires physical infrastructure.

CSE’s electrification business provides infrastructure such as power distribution centres, switchgear and other critical electrical systems used in data-centre environments.

This is important because the company’s growth is increasingly being driven by electrification.

CSE’s first-half 2026 revenue increased 27.4% year on year to S$561.5 million, while its electrification segment generated S$283.0 million of revenue — more than half of group revenue.

Citi expects electrification to account for approximately 53% of CSE’s revenue by FY2026.

That is a significant change in the company’s earnings profile.

The investment thesis is therefore no longer simply about CSE winning more engineering projects.

It is about whether the company can become a meaningful supplier of the electrical infrastructure required to expand AI computing capacity.

The Amazon connection is what makes the story particularly interesting

The most eye-catching part of Citi’s thesis is Amazon.

Citi estimates that CSE has a potential US$1.5 billion data-centre electrification pipeline from Amazon.

This figure needs to be handled carefully.

It is a potential project pipeline cited by Citi, not S$1.5 billion of contracted revenue that CSE has already booked.

That distinction is crucial.

Nevertheless, the potential scale illustrates why CSE’s US expansion matters.

Amazon is one of the world’s largest technology companies and a major investor in data-centre capacity.

CSE does not need to build Amazon’s data centres or manufacture its AI chips.

It can potentially benefit from the infrastructure required to power them.

That is the classic picks-and-shovels investment argument.

Amazon could potentially be more than just a customer

There is an even more unusual aspect to the relationship.

Citi notes that Amazon could potentially exercise warrants equivalent to up to 8% of CSE’s expanded share base through FY2030.

If that structure ultimately develops as expected, CSE could potentially have Amazon in two strategically important roles:

a major customer and a strategic investor.

That would be unusual for a company of CSE’s size.

But investors should not overinterpret it.

A potential warrant position is not the same as Amazon already owning 8% of CSE.

And a potential project pipeline is not the same as guaranteed future revenue.

The significance is strategic rather than immediately financial.

It suggests that CSE’s relationship with a major hyperscaler could become deeper as its US data-centre business expands.

The problem: CSE’s earnings have not caught up yet

This is where the story becomes considerably more interesting.

CSE’s revenue is growing rapidly.

Its profits are not.

For 1H2026, revenue rose 27.4% to S$561.5 million, but net profit attributable to owners fell 19.3% to S$13.2 million.

Gross profit increased only 8.8%, while gross margin compressed from 27.9% to 23.8%.

That tells investors something important.

CSE is currently paying for growth.

The company has been investing in its US capabilities, while the business mix has also shifted towards data-centre projects.

Those projects can be attractive long term, but they can also carry lower margins during the ramp-up phase.

CSE’s new Champion facility in the US has also incurred start-up costs.

The result is a familiar pattern in expansion stories:

revenue first, operating leverage later.

But “later” is the critical word.

Investors are being asked to believe that today’s investment will produce substantially better profitability tomorrow.

Citi is betting on margin expansion

This is the key part of Citi’s thesis.

Citi forecasts approximately 15% two-year revenue CAGR and 22% two-year EPS CAGR.

The fact that EPS is expected to grow faster than revenue is important.

It implies that CSE’s profitability should improve materially as the business scales.

Citi expects EBITDA margins to rise to approximately 9.8% in 2027 and 10.5% in 2028.

That is the real investment thesis.

It is not simply:

AI data centres will create more demand for CSE.

It is:

AI data-centre demand will allow CSE’s expanded US infrastructure to reach sufficient scale for operating leverage to emerge.

That is a much more demanding proposition.

The bull case

The bullish scenario is relatively straightforward.

AI-related data-centre construction continues at a strong pace.

Hyperscalers continue investing heavily in capacity.

Demand for electrical infrastructure remains strong.

CSE’s US facility reaches efficient utilisation.

Amazon-related opportunities convert into actual orders.

CSE wins additional business from other hyperscalers.

Revenue rises rapidly.

Fixed costs are spread over a larger revenue base.

Margins recover.

EPS grows faster than revenue.

And investors begin to value CSE as a higher-growth infrastructure company rather than a traditional engineering-services business.

If that happens, Citi’s S$1.70 target may not look particularly aggressive.

The bear case

But there is an equally important downside scenario.

AI data-centre capital expenditure could slow.

Hyperscalers could delay projects.

US data-centre construction could face regulatory or community opposition.

CSE’s facilities could take longer to reach efficient utilisation.

Project execution could generate additional cost overruns.

Margins might fail to recover as quickly as expected.

And the market could conclude that CSE deserves a conventional engineering-services valuation rather than an AI-infrastructure premium.

That would leave investors with a company that has invested heavily for future growth without generating the expected level of earnings leverage.

The biggest risk is execution, not whether AI is real

This distinction is critical.

Investors do not need to decide whether artificial intelligence is a temporary fad.

The physical infrastructure requirements are real.

The more difficult question is whether CSE can capture enough of that spending profitably.

The company’s 1H26 results illustrate the problem.

Revenue growth was strong, but profit declined.

That means the company’s expansion is currently producing more activity than earnings.

The next phase of the investment thesis therefore depends on operating leverage.

If the US business scales successfully, the current earnings weakness could prove temporary.

If it does not, the market may question why investors should pay a premium for the AI connection.

Why CSE could be an interesting AI infrastructure stock

One attraction of the story is that CSE does not need to win the AI technology race.

It does not need to develop the best large language model.

It does not need to manufacture the dominant GPU.

It does not need to compete directly with the hyperscalers.

Its opportunity exists one layer underneath.

AI computing requires:

power → distribution → switching → electrical equipment → data-centre infrastructure.

CSE is positioned within that physical infrastructure chain.

That gives investors an indirect way to participate in AI capital expenditure without betting directly on which AI model, chip architecture or cloud platform ultimately wins.

But investors should not call CSE an “AI company”

There is a danger in attaching the AI label to every company supplying data centres.

CSE remains a diversified engineering and systems-integration business.

Its earnings depend on multiple industries and geographies.

The more accurate description is:

CSE Global is becoming an increasingly interesting beneficiary of the physical infrastructure required by the AI data-centre build-out.

That distinction matters.

It keeps the investment thesis grounded in what the company actually does.

The valuation question

Citi’s S$1.70 target implies approximately 44% upside from the S$1.18 reference price.

But investors should not interpret the target price as evidence that the stock is objectively worth S$1.70.

It is an analyst valuation based on assumptions about future earnings, growth and margins.

The important question is what expectations are already embedded in the current share price.

If CSE successfully delivers the earnings growth Citi expects, the stock could have room to re-rate.

If the US expansion disappoints, however, the downside could be amplified because investors may simultaneously cut their earnings forecasts and valuation multiple.

That is why Citi’s “Buy / High Risk” designation is important.

The opportunity is attractive precisely because expectations about future earnings are changing.

But that also creates greater execution risk.

What investors should watch next

The next few reporting periods will be more important than another analyst target-price revision.

1. Actual Amazon orders

The most important question is whether the potential Amazon pipeline converts into contracted projects and recognised revenue.

That would move the thesis from analyst expectation to company financial performance.

2. US facility utilisation

CSE’s US expansion needs to generate sufficient throughput to absorb its fixed costs.

Improving utilisation would provide evidence that the operating-leverage thesis is working.

3. Electrification margins

Revenue growth is not enough.

Investors need to see margins recovering as the data-centre business scales.

4. Order intake and order book

CSE’s 1H26 order intake increased 28.3% year on year to S$470.2 million, while its order book stood at S$620.4 million at June 30.

Those figures provide an important indicator of future revenue visibility.

The key question is whether future orders continue to grow while profitability improves.

5. Additional hyperscaler relationships

Amazon is the obvious name in the current thesis.

But CSE has indicated ambitions to build relationships with other hyperscalers.

If the company succeeds, the investment case becomes less dependent on a single customer and potentially much larger.

The overlooked part of the AI boom

This may ultimately be the most interesting aspect of CSE Global.

The AI boom is creating an enormous amount of demand for physical infrastructure.

Investors tend to focus on the companies producing the technology.

But infrastructure booms often create opportunities further down the supply chain.

The winners are not necessarily the companies with the most exciting technology.

They can also be the companies supplying the electricity, equipment and infrastructure needed to deploy it.

That is the role CSE is trying to occupy.

And if its US expansion succeeds, today’s engineering company could gradually develop a very different earnings profile.

Bottom line: CSE is interesting — but the numbers still have to catch up

CSE Global has a potentially compelling combination:

AI data-centre growth + electrification + US expansion + hyperscaler exposure + operating leverage.

Citi’s forecast of 15% revenue CAGR and 22% EPS CAGR over two years provides a clear illustration of what the bull case could look like.

The potential Amazon pipeline adds another layer of upside.

But the company’s latest results show why investors should not simply buy the AI narrative.

Revenue increased strongly in 1H26.

Profit did not.

CSE is still in the investment phase of its US expansion, and the market needs to see evidence that the spending will translate into higher margins and earnings.

That makes CSE a particularly interesting high-risk AI infrastructure play, rather than a straightforward AI stock.

The key question for investors is not:

“Will AI data centres need more power?”

They almost certainly will.

The question is:

“Can CSE Global convert that demand into high-margin, scalable revenue?”

If it can, CSE could become one of Singapore’s more overlooked beneficiaries of the global AI infrastructure boom.

If it cannot, the AI label will not be enough to justify the valuation.

Investment view: WATCH / SPECULATIVE ACCUMULATE

For investors comfortable with small- and mid-cap execution risk, CSE deserves close attention. The next confirmation should come from US utilisation, order conversion, margin recovery and earnings growth — not from the AI narrative alone.