In His Own Words: AEM CEO's Take on Unprecedented Growth Opportunities In AI Boom

The following is the CEO of AEM's opening remarks at the company's 1Q2026 earnings call. He describes how the AI boom is a transformative catalyst for AEM, driving unprecedented growth, expanding its customer base, and perfectly aligning with its specialized technological strengths. 
 

Third, and this one I'm particularly proud of our team for achieving. We've been awarded the 2026 Intel EPIC Award. This is the highest Intel recognition for a supplier due to excellence given for their performance, innovation, and collaboration within their global supply chain. This is not a marketing award. This is a technical and operational endorsement from one of the most demanding customers in the semiconductor industry. It reflects the depth of the relationship that has been built over many years.

Together, the guidance upgrade, the ASE partnership, and the Intel Epic Award, this quarter signals something important. AEM is not just executing as a single ramp customer. We're broadening our industry footprint at exactly the moment when the AI test market is entering a multi-year structural growth cycle.

Let me set the stage of why we have so much conviction in the duration of this cycle.

AI is driving the largest infrastructural build-out in history. By 2030, data center-related investments are expected to reach approximately US$7 trillion across compute infrastructure, equipment, networking, and power. For semiconductors, this points to a market that could roughly double from around US$800 billion today to approximately US$1.6 trillion by 2030. And the majority of this incremental growth is expected to come from leading-edge AI and HPC devices. This is very important for AEM because those devices are increasingly difficult to test. They are larger, hotter, more complex, and more heterogeneous.

As a result, production test is becoming more critical and is a central part of the semiconductor supply chain. AEM is positioned directly in that constraint. We provide the advanced thermal management coupled with high parallel test capabilities needed to qualify these advanced devices reliably and mostly at scale. As AI HPC complexity increases, the relevance of our platforms increases with it.

So I had the personal luxury of serving high-performance compute segment for over 25 years. And I want to take this moment to describe to you three distinct eras that this industry has gone through while addressing this high-performance compute market. In the PC era, a typical generational upgrade added roughly 40% more transistors, and about 10% more power consumption. And the product cycles were approximately between 12 to 24 months. Later, we transitioned to the mobile era. Release cadence accelerated to roughly an annual cycle. Package sizes and power removal requirements in test moved much more gradually. Test challenges evolved, but physical envelope remained manageable from generation to the next.

AI accelerators and where we are today are completely different. In recent generational transitions from a leading GPU, transistor count increased more than 200%, silicon footprint expanded also over 200%, thermal design power rose by 80%. And that all was coupled with doubling of chiplet count. All of these are happening while the cycle to produce these products was compressing down from six to nine months. That is a key point. AI/HPC is not a normal product cycle transition. The industry is being asked to handle larger, hotter, more complex devices at a faster cadence. That directly increases the importance of advanced thermal control, precision handling, contact integrity, and high parallel test capability.

It also expands the opportunity for recurring business and device-specific collaterals. This is exactly where AEM is focused. Test challenges are intensifying across three dimensions all at once.

First, larger form factor. Rapid growth of next-generation AI and HPC packages are expected to push the industry towards adopting new transport mediums with over two times the area of today's existing mediums. This shift is analogous to the front-end semiconductor industry when it transitioned from 200-millimeter to 300-millimeter wafers in the early 2000s. When a larger form factor is required, a new automation and handling on process control infrastructure was needed. For these upcoming packages, the industry will need to handle devices weighing more than 2 kilograms and warpage tolerances up to 25 times larger than older generation. Achieving accurate, repeatable contact at production throughput across all these larger geometries will become a precision engineering challenge that most legacy test solutions were simply not designed to solve.

Second, longer test time. Test time is rapidly rising as we transition into finer transistor nodes and increasing the chiplet counts per device. Every device of a leading edge node requires more validation, more coverage, and more test time.

Third, greater heterogeneity. As a modern AI processor is an assembly of compute chiplets, HBM memory stacks, co-package optics, and integrated voltage regulators, all on a single package, each has its own test requirements and thermal gradients across the assemblies, unlike anything we've seen in previous chiplet generations.

Against that backdrop, we see a US$3 billion serviceable addressable market, or SAM, that is available for AEM today. And based on our development roadmaps, for 2027 and 2028, alongside with market trends, we expect that SAM to expand to US$4.5 billion by 2028. That market spans four customer categories where we're active in engagement -- our established PC and Foundry relationship, the fast-growing AI and HPC segment, the emerging memory opportunities, and, of course, the OSATs. We're not trying to capture all of those overnight. Our target is to capture a meaningful share over the next three to five years. And we have five clear pillars, a pillar strategy to do exactly that.

The investment case for AEM rests on four interconnected layers, and I want to walk you through each one as we have conviction of the drivers.

The foundation of AI and HPC growth. Agentic AI is a main driver. And we see AI that acts autonomously, runs workflows, and orchestrated systems. We see that arriving at a faster pace and an extraordinary compute intensive. Bank of America forecasts 50% year-over-year growth in AI semiconductor devices. And I believe that forecast, if anything, is conservative. The reason is simple. AI compute architecture is broadening. Training clusters may have built around the CPU to GPU ratio of roughly one to eight, but agentic AI workflows are expected to move closer to one-to-one as CPU takes on a much larger role in orchestration, scheduling, memory movement, networking, and system control. This is a critical point.

The AI opportunity is not only about GPU. As AI moves from training to inference, reasoning, and autonomous execution, server CPUs re-emerge as a strategic layer in the AI stack. This plays directly to AEM's strengths. AEM commands a clear leadership position in CPU testing, and that capability becomes increasingly relevant as AI systems scale in complexity. The growth tailwinds are now not confined to one customer or one device category. It spans to CPUs, GPUs, and custom accelerators.

On top of this sits test complexity. Every device that ships into AI data centers has to be tested. And as I just walked you through just a few slides ago, each generation of these devices is getting harder, longer to test, and more expensive. Test intensity is growing faster than device volume.

The third layer is AEM's differentiation. AEM's pixel thermal technology and high-parallel test architecture are directly aligned with the technical requirements emerging from next-generation AI and HPC production. This is where AEM has built meaningful differentiation. Our engagements with leading semiconductor customers are a result of rigorous technical qualification and production validation. External recognition, including Intel's EPIC Award for performance, innovation, and collaboration, further reinforces that position. But the more important point is customers are selecting AEM for technically demanding production realms, where reliability and execution is critical. That is the clearest evidence of the value of our platforms.

The fourth layer is the earnings leverage. AEM's recent margin performance gives a clear indication of the leverage in the business. Over the past 12 months, PBT margin has improved from 4.4% to 15.2%. While some of that reflects cyclical recovery and operating discipline, it also demonstrates the margin potential of the model as volumes normalize and higher value platforms contribute to our mix. We're excited to see AEM's transformation to address highly diversified and growing portfolios.

Let me give you a snapshot of where we stand regarding our four growth pillars.

First, our PC Foundry customer. Our longest standing customer is showing signs of recovery. We're seeing improved order momentum, as we move into 2026, although we remain appropriately disciplined on how we think about the pace of that recovery. There are two developments worth highlighting. First, we're supporting this customer as they pursue hyperscaler-related opportunities, which could represent a new growth vector for this longstanding relationship. Second, we're actively working with them to migrate selected programs from older custom platforms onto our newer, high-performance tools. Both developments are strategically important. They also do not change the account overnight, but they will improve the quality and durability of the relationship over time. As our customer's business recovers and expands, these transitions should position AEM to participate in a broader opportunity set while improving the long-term economics of this account.

Second, fabless/AI HPC customer. This is the most exciting business for us now. This customer is ramping at a high volume and is on track to become our single largest revenue contributor in 2026. We're growing wallet share within the account and expanding into new OSAT relationships alongside of them. The ramp is real, and it's sustained, and we expect it to continue throughout the year.

Third, the memory segment. We're making great progress on the qualification of our new Xact final test handler with our first IDM customer. We remain on target for an initial revenue contribution in the second-half of the year, with the ramp predicted to start in 2027. In parallel, we're expected to engage with a second memory IDM customer for an evaluation in late 2026, which will broaden the opportunity of our Xact handler and support our longer growth ambition into this memory segment.

Fundamentally different

"AEM today is fundamentally different. We're no longer just serving a single customer. We have the technology, the customer relationship, the operational infrastructure, and now the strategic partnership (with ASE) to participate meaningfully in the most important technology investment cycle of our generation." -- Samer Kabbani, CEO

Fourth are OSATs. The OSAT pillar is critical to AEM's growth strategy. As AI and HPC architectures move deeper into advanced packaging and chiplet-based designs, OSATs are becoming increasingly central to this production ecosystem. Building a stronger position with leading OSATs is therefore a key part of how we intend to scale beyond individual device programs and participate more broadly in the next generation semiconductor growth. Our recent ASE partnership is a strong validation of AEM's capabilities. It demonstrates that our test handling thermal platforms are relevant not only to demanding device makers, but it's also critical to the advanced packaging and test leaders who are enabling AI and HPC production at scale.

Fifth, our contract manufacturing. Our contract manufacturing segment is playing an important role in supporting our semiconductor realm and is driving important improvement in our profitability throughout 2026. It's the operational engine that further enables our recurring business.

This slide captures why the OSAT pillar is critical to AEM's growth strategy. ASE is not just another customer or a partner. It's the world's largest OSAT with approximately 45% global market share and a central player in advanced packaging and testing. The partnership brings together ASE scale, manufacturing footprint, customer reach with AEM's differentiated pixel thermal technology and high parallel test architecture.

There are three important strategic benefits to this partnership.

First, ASE gives a strong pathway into hyperscaler-related opportunities through a partnership already embedded in that ecosystem. Second, our collaboration with ISE Labs positions AEM earlier in the development cycle during testing, validation, and characterization. Third, ASE's equity investment and revenue-linked warrants align incentives and provide strong validation and adoption of AEM's platforms.

Conclusion Let me close with our outlook. We're raising 2026 revenue guidance to S$550-600 million, representing a growth of approximately 38 to 50% versus our 2025 revenue. which was approximately S$400 million. This is a significant upward revision from our previous guidance, and it reflects the stronger than expected demand environment across our customer base. I want to leave you with a final thought. AEM today is fundamentally different. We're no longer just serving a single customer. We have the technology, the customer relationship, the operational infrastructure, and now the strategic partnership to participate meaningfully in the most important technology investment cycle of our generation. The AI/HPC market is in the early stages of a multi-year growth cycle. We're in the right place with the right solutions at the right time. We look forward to executing on that opportunity and continuing to deliver to our shareholders.

→ See also:AEM: $1.70 at Start of Year, Now Analysts' Average Target Price is $12.77