AEM Holdings has given Singapore investors something increasingly rare: a local semiconductor stock with direct exposure to the artificial-intelligence infrastructure boom.
The company does not design the AI chips powering data centres. Instead, it sits one step further down the supply chain, providing semiconductor test and handling solutions used to validate increasingly complex chips.
That distinction matters.
As AI processors become more powerful, expensive and thermally demanding, testing becomes more important — and potentially more valuable.
AEM’s recent numbers suggest that this thesis is translating into actual earnings.
But there is a catch.
The market discovered the AEM AI story long before investors saw the latest half-year numbers.
AEM’s share price had already surged more than 500% during the first half of 2026, making it one of Singapore’s strongest-performing stocks.
So the investment question has changed.
It is no longer simply:
Is AEM benefiting from AI?
The answer to that is increasingly obvious.
The more important question is:
Are AEM’s future earnings growing fast enough to justify what investors are already paying for the stock?
That is where the story gets much more interesting.
AEM’s AI exposure is translating into real operating momentum
AEM entered 2026 with a business that was already showing signs of a significant earnings recovery.
For the first quarter of FY2026, revenue increased 35.8% year on year to S$116.9 million.
But the more important number was profitability.
Net profit jumped 329.4% to S$14.3 million, while profit before tax increased 370% to S$17.8 million.
The net profit margin expanded to 12.3%, from just 3.9% a year earlier.
That tells investors something important.
This is not simply a case of AEM selling more equipment.
The company is beginning to generate considerably more profit from each dollar of revenue.
That operating leverage is potentially more important than the headline revenue growth.
AEM’s Test Cell Solutions business was the main engine.
Revenue from the segment increased 72% year on year to S$88.1 million in the first quarter, accounting for roughly three-quarters of group revenue.
The growth was driven by production deployments for a major fabless AI/HPC customer.
And this is precisely where the AI investment thesis becomes tangible.
AEM is not merely telling investors that it has an “AI strategy”.
It is supplying equipment into the production ramp of AI/HPC chips.
The AI testing opportunity is bigger than simply “more chips”
There is a deeper reason why AEM is attracting investor attention.
The AI semiconductor industry is moving towards increasingly complex chip architectures, including chiplets and high-performance packages.
These devices generate greater thermal and power-management challenges.
That creates additional testing requirements.
AEM has positioned its technology around high-parallel testing and advanced thermal management — capabilities that become increasingly relevant as AI chips become more complex.
The company’s FY2025 annual report highlighted rising test intensity and increasing thermal and power-density challenges as important industry trends.
This creates an interesting investment dynamic.
If AI chip complexity rises faster than chip volumes, semiconductor test companies can potentially benefit from higher test content per chip, rather than relying solely on unit growth.
That could make AEM a leveraged play on the AI hardware cycle.
Management has already raised its FY2026 revenue guidance
The strongest evidence that the recovery is not merely a one-quarter phenomenon is management’s guidance.
After its strong first-quarter performance, AEM raised FY2026 revenue guidance by approximately 20% to between S$550 million and S$600 million.
That compares with FY2025 revenue of S$399.3 million.
At the midpoint, the new guidance implies revenue of approximately S$575 million — roughly 44% above FY2025.
That is a substantial acceleration.
And it explains why investors have been willing to assign a dramatically higher valuation to AEM.
The market is not valuing AEM based on its old earnings profile.
It is attempting to value what AEM could earn if the AI/HPC production ramp continues.
But here is the uncomfortable part: the share price moved first
This is where investors need to exercise valuation discipline.
AEM’s share price had already surged by more than 500% in the first half of 2026.
That means today’s investor is not buying the same AEM that existed at the beginning of the year.
The stock has already undergone a massive re-rating.
This distinction is crucial.
A company can report excellent earnings and still deliver poor future investment returns if investors have already paid too much for those earnings.
Consider a simplified example.
Suppose a company’s earnings double.
If its share price also quadruples because investors become dramatically more optimistic, shareholders have not necessarily benefited from the earnings growth on a proportional basis.
The problem is multiple expansion.
AEM investors therefore need to watch two things simultaneously:
earnings growth and valuation compression.
If earnings continue to surge, the current valuation can eventually become more reasonable.
But if earnings growth merely meets expectations, the stock could still struggle if investors begin demanding a lower valuation multiple.
The market has already priced in a large part of the turnaround
There are several reasons why investors should not treat AEM simply as another cheap semiconductor recovery stock.
For FY2025, AEM generated S$399.3 million of revenue and S$17.1 million of net profit.
That translates into a net profit margin of only about 4.3%.
The improvement in 2026 is therefore coming from a combination of:
- stronger AI/HPC demand;
- higher Test Cell Solutions revenue;
- improved operating leverage;
- better utilisation;
- and margin expansion.
This creates substantial upside if the trend continues.
But it also creates substantial downside if the market has extrapolated today’s margins too aggressively into the future.
The critical question is therefore not whether AEM can grow.
It is how sustainable the current earnings trajectory is.
AEM is becoming less dependent on a single AI customer — but concentration remains a risk
Another important development is customer diversification.
AEM said its second AI/HPC customer contributed more than 25% of Test Cell Solutions revenue in FY2025 and was expected to grow further in FY2026, potentially becoming the group’s largest customer.
That is encouraging because it reduces the historical perception that AEM’s fortunes are overwhelmingly tied to one customer.
But it also highlights another risk.
AEM remains exposed to a relatively small number of major semiconductor customers.
If an important customer delays a production ramp, changes its testing strategy or reduces capital expenditure, AEM’s quarterly numbers could be affected disproportionately.
For investors, this means that customer concentration remains one of the most important variables to monitor.
The ASE partnership could be more important than it initially appears
There is also a strategic development that deserves more attention.
In March 2026, AEM announced a strategic partnership with ASE Technology, one of the world’s largest outsourced semiconductor assembly and testing companies.
ASE agreed to subscribe for approximately S$12 million of new AEM shares.
The partnership combines AEM’s test technology with ASE’s manufacturing scale and market access, with both companies targeting AI and HPC testing applications.
This could potentially expand AEM’s addressable market beyond its existing customer relationships.
More importantly, it gives investors another way to think about AEM.
The company is not simply trying to sell more equipment into the existing semiconductor cycle.
It is attempting to become a more strategically important participant in the emerging AI/HPC testing ecosystem.
That could matter considerably over the next few years.
So, is the easy money already made?
For investors who bought AEM before the 2026 rally, the answer may be very different from the answer for someone considering the stock today.
The early investors were effectively betting on a semiconductor recovery before the earnings had fully recovered.
Today’s investors are buying after the market has already recognised the AI opportunity.
That makes the risk-reward profile different.
The bull case is straightforward.
If AI/HPC demand continues accelerating, AEM could enjoy:
- sustained revenue growth;
- further operating leverage;
- higher margins;
- stronger free cash flow;
- additional customer diversification; and
- potentially higher earnings estimates from analysts.
If those factors materialise, today’s seemingly expensive valuation could eventually look more reasonable.
But there is a bear case.
If AI semiconductor spending slows, if production ramps are delayed, if customer concentration becomes a problem or if AEM’s margins fail to expand as quickly as investors expect, the stock could face a significant valuation reset.
The danger is therefore not necessarily that AEM’s business deteriorates.
The danger is that the business performs well — just not well enough to justify the expectations already embedded in the share price.
What AEM investors should watch next
For investors considering AEM after its enormous rally, five indicators may matter more than the headline profit number.
1. Test Cell Solutions growth
This remains the clearest indicator of whether the AI/HPC thesis is working.
If TCS revenue continues growing rapidly, the market has more reason to believe the AI opportunity is structural rather than temporary.
2. Net profit margins
Revenue growth is only half the story.
Investors should watch whether AEM can maintain the significant improvement in profitability seen in Q1 2026.
If margins continue expanding, earnings could grow much faster than revenue.
3. Customer concentration
AEM’s AI/HPC customers are major sources of growth, but investors should track how diversified the revenue base becomes.
More customers would reduce the risk associated with any single production programme.
4. Cash generation
Rapid revenue growth is attractive.
Rapid revenue growth accompanied by strong operating cash flow is much more attractive.
AEM’s FY2025 operating cash flow improved dramatically to S$136 million, compared with negative S$17.5 million in FY2024.
Investors should watch whether this improvement continues as the business scales.
5. Valuation versus earnings growth
This may ultimately be the most important variable.
AEM does not need to become a bad company for its shares to fall.
It simply needs to become less exceptional than the market currently expects.
That is the distinction investors should remember.
The investment takeaway
AEM Holdings is arguably one of Singapore’s clearest listed beneficiaries of the AI hardware boom.
The company sits in an attractive part of the semiconductor value chain, where increasingly complex AI chips require increasingly sophisticated testing.
Its 2026 operating performance is providing genuine evidence that the AI/HPC opportunity is translating into revenue and earnings.
And management has already raised FY2026 revenue guidance substantially.
But the stock is no longer an undiscovered AI play.
The market has already repriced AEM dramatically.
That changes the investment question.
AEM may still have a compelling long-term business story. But after a more than 500% first-half rally, investors need to stop asking whether AEM is an AI winner and start asking whether its future earnings can outrun the expectations already embedded in its valuation.
For existing shareholders, the question is whether to continue holding through what could be a much larger AI/HPC earnings cycle.
For new investors, the question is more difficult:
Are you buying the beginning of AEM’s earnings super-cycle — or buying after the market has already priced in much of it?
That is ultimately the more important AEM investment debate in the second half of 2026.