Singapore – When Samer Kabbani was CEO AEM Holdingswas quick to cancel his daughter’s US$200 (S$256) monthly subscription after learning that she had signed up for ChatGPT without his knowledge. Within minutes, footsteps came rushing down the stairs. “Why did you cancel? My life depends on it!” he was told.
Meanwhile, my teenage son wanted a calorie counting app, so instead of buying one, he decided to try making one himself. Despite having no programming experience, the 16-year-old was able to use an artificial intelligence coding platform to create a working prototype that can analyze food photos and estimate calorie intake.
While walking through the ruins of the Parthenon in Greece during a family vacation, his other daughter used AI to generate an instant guided tour, pushing past the tour guides serving her around the tourist attraction.
“She simply pointed her phone at the ruins and asked ChatGPT to describe what she was seeing,” Cavani said. He regularly uses AI to brainstorm ideas, refine arguments, and more.
In the Qabbani family, AI is already integrated into daily life in many ways, including learning, fitness, travel, productivity, and entertainment.
For Kabani, who runs a Singapore company testing the high-performance AI chips that enable these applications, this says something important about where the world is heading.
“AI is now becoming as much a part of life as electricity, water and the internet,” he told The Straits Times in an exclusive interview.
The growing use of AI, from generating memes and entertainment content to supporting mission-critical transportation and medical systems, is driving enormous demand not only for high-performance computing chips, but also for the data centers and computing infrastructure needed to run that technology at scale.
This has driven trillions of dollars of investment into the field and triggered a rally in AI stocks from Singapore and South Korea to the United States and China.
Singapore-listed AEM, which operates systems and equipment to test these chips before they are deployed, is not left out.
Since the start of 2026, the company’s stock has ridden a wave of investment in AI infrastructure, rising about 450% from less than $2 in January to $9.56 at the close of trading on May 22nd.
Unlike traditional standardized components, AI chips now come in different configurations depending on the customer and application, Cavani explained.
For example, a chip designed for Meta might prioritize training AI models, while another chip for China might have a different configuration to comply with export regulations. Some are optimized for memory-intensive workloads, while others are optimized for processing speed or energy efficiency.
As demands for chips have increased, chips have also become physically larger, hotter, and more complex, and testing chips before assembly is now a critical part of the semiconductor manufacturing process.
Chipmakers such as Intel and Advanced Micro Devices (AMD) typically have to prove to customers like Meta that their chips can operate reliably in extreme temperatures such as 90 degrees Celsius. “Our mission is to maintain the temperature at 90 degrees Celsius at all times during the test,” Cavani explained.
He said AEM’s strength lies in its advanced thermal technology, which allows it to maintain constant temperatures during testing more effectively than its competitors.
“This is very important because when we test these chips and put them under high stress, they generate heat, so we need to control the temperature of the chips during testing, otherwise it could cause a deadly explosion.”
Another important advantage is that a much larger number of chips can be tested simultaneously.
“We have the equipment to test dozens of chips simultaneously, under extreme heat and power, with tightly controlled temperatures. We were testing 30 or 40 at a time, while others were testing eight,” Cavani said.
He added that speed and efficiency are increasingly valued in the AI era, and chips could become obsolete within two years as newer, faster versions emerge. The third advantage of AEM is that it can meet all necessary testing requirements.
“Customers can no longer afford to piece together test solutions from multiple vendors. They want someone who can deliver the entire solution quickly.”
Beyond testing PC chips for Intel, AEM has made significant investments over the past six years to position itself to support AI and high performance computing (HPC) chips made by companies such as Nvidia, AMD, and Taiwan Semiconductor Manufacturing Company (TSMC).
In 2025, the company said it began ramping up production for a second customer focused on AI and HPC chips. This customer is expected to overtake Intel in 2026 to become AEM’s largest customer.
At the same time, Intel itself is evolving into a foundry business, making chips for external customers such as hyperscalers and large technology companies building AI infrastructure and cloud computing systems. These companies include Meta, Amazon, and Microsoft.
This means AEM’s customer exposure through Intel foundries is also potentially more diversified, Cavani said. He added that since AI systems require large amounts of memory to process data, there is scope for AEM to further expand into testing advanced memory chips, which are in high demand.
In 2025, AEM announced that it had acquired Samsung Electronics, Micron Technology, and SK Hynix, one of the world’s three largest memory chip companies, as customers, with testing expected to begin towards the end of 2026.
“Currently, we have three different customers with three different profiles. They are very diversified. The market cycles are also different, so that helps stabilize the fluctuations for us,” Cavani said.
AEM’s latest advancement came in March, when it announced a partnership with Taiwanese semiconductor giant ASE Technology Holding, one of the world’s largest contract semiconductor assembly and test providers.
These companies work closely with foundries like TSMC to package, assemble, and test chips before shipping them around the world.
Cavani said AEM deliberately courted ASE because of its dominant market position and deep ties with TSMC. He added that it took more than a year of effort for AEM to win ASE to bring its technology and manufacturing ecosystem to Taiwan’s semiconductor industry.
ASE ultimately invested in $Investing $12 million in AEM to support expansion in Taiwan and integrating AEM’s technology into ASE’s manufacturing operations, with further investments leading to future revenue milestones.
AEM expects this partnership to help it reach more global AI chip customers and hyperscalers, and has revised its revenue outlook range upward by approximately 20%. $$550 million to $600 million in 2026.
Of course there are risks. First, the AI chip boom may not be able to maintain its current pace if demand fluctuates rapidly and hyperscalers cut spending, or if trillions of dollars of AI investments fail to produce the expected returns.
The company has successfully diversified beyond Intel, but at a time when AEM’s stock price has already soared and investor expectations are high, delays in orders, technology shifts by major customers, or competitors catching up could impact earnings.
What is clear, however, is that as AI becomes more deeply integrated into everyday life, demand for much larger and more complex chips will continue to rise, creating new engineering challenges and increasing demand for the specialized equipment needed to test them, Cavani said. He noted that an estimated US$7 trillion is being poured into the infrastructure needed to support AI applications, from generating itineraries and videos to personalized recommendations and coding, adding that there are “limitations” to what these chips will ultimately be able to do.
“These are exactly the conditions where AEM’s technology delivers maximum value,” Cavani said.
