ChipMOS Q2 Revenue and Profits Hit Record Highs! EPS of NT$15.02, HPC Revenue Up Nearly 80% YoY, Additional Capital Expenditure for Capacity Expansion

Rising demand for AI and high-performance computing (HPC) chip testing has propelled ChipMOS Technologies (6510) to another operational peak in the second quarter of 2026. ChipMOS held an online business update today (29th), reporting consolidated revenue of NT$1.64 billion and net profit attributable to parent of NT$494 million, with earnings per share (EPS) reaching NT$15.02. Revenue, profit, and EPS all achieved new single-quarter highs. Notably, HPC-related revenue grew nearly 80% year-on-year, and its share of total revenue jumped from 45.1% in Q1 to 56.3%.

Profit growth outpaced revenue growth, with EPS exceeding one full share value for the second consecutive quarter. ChipMOS’s Q2 consolidated revenue reached NT$1.64 billion, up 20.8% quarter-on-quarter and 34.9% year-on-year. Gross margin improved to 57.5%, up 0.7 percentage points from Q1 and 1.7 percentage points from the same period last year. Benefiting from increased revenue scale, improved product mix, and cost control, operating margin rose to 34.7%, a 5.4 percentage point increase from the previous quarter.

Leveraging operational leverage, Q2 net profit attributable to parent reached NT$494 million, up 44.5% quarter-on-quarter and 129.1% year-on-year—significantly outpacing revenue growth. EPS increased from NT$10.43 in Q1 to NT$15.02, marking the second consecutive quarter of earnings exceeding one full share value.

For the first half of 2026, ChipMOS achieved consolidated revenue of NT$2.997 billion, net profit of NT$836 million, and cumulative EPS of NT$25.45.

HPC Revenue Share Rises to 56.3% – U.S. Customer Demand as Key Driver

Regarding Q2 growth drivers, General Manager Huang Shui-ke stated that continued expansion of AI infrastructure investment by global cloud service providers has driven demand for high-speed computing chips and related test interfaces. HPC-related revenue share increased from 45.1% in Q1 to 56.3% in Q2, now exceeding half of total revenue. This segment’s revenue grew nearly 80% year-on-year, primarily driven by expanding demand from U.S. customers. This indicates that AI and HPC businesses are no longer just short-term orders but are becoming a core pillar of ChipMOS’s revenue structure.

Huang noted that structural demand from AI and HPC is pushing the semiconductor industry into a multi-year expansion cycle. "AI-related demand has transformed from short-term rush orders into long-term growth momentum." The test interface market will also benefit from rising AI chip performance and advancements in advanced packaging, leading to sustained long-term demand.

AI Race Shifts from Training to Inference – Cloud Providers Accelerate In-House ASIC Development

As large language models and agentic AI become more practical, AI computing demand is expanding beyond model training to broader, continuous inference workloads. Achieving high-volume inference at lower power and cost has become the next competitive focus for cloud service providers.

In addition to continued GPU adoption, major global cloud providers are accelerating the development of customized ASICs with advantages in low power consumption and high cost-performance, driving AI computing architectures toward diverse paths including CPU, GPU, APU, ASIC, and TPU.

Huang emphasized that for test interface providers, different AI chip architectures, computing performance, and advanced packaging designs mean test products must handle higher frequencies, more complex signals, and stricter signal integrity requirements. The AI race is not only boosting chip shipments but also raising the technical barriers and product value of high-end test interfaces.

Additional Capital Expenditure for 2026 – New Capacity Coming Online

To meet rapidly growing AI testing demand from customers, ChipMOS’s board has approved additional capital expenditure for 2026 to further expand advanced process capacity, upgrade equipment, and strengthen smart manufacturing infrastructure. Huang stated that new capacity is already being brought online, enhancing production flexibility and delivery capability.

ChipMOS continues to integrate AI-driven smart manufacturing into production processes to improve process technology, production efficiency, and cost control. The rise in Q2 operating margin to 34.7% reflects not only a higher proportion of high-end products but also the initial impact of smart manufacturing and economies of scale on profitability.

However, the company did not disclose the original capital expenditure budget or the additional amount. Market focus will now shift to the speed of new equipment deployment and capacity ramp-up, and when expansion benefits will further reflect in revenue.

Next-Gen AI Chips Gradually Passing Validation – Advanced Packaging Raises Test Complexity

On the technology front, Huang stated that the company is continuously investing in R&D for high-end test interface products and gradually passing customer validation for next-generation AI computing chips. It is also introducing next-generation materials and precision manufacturing techniques to meet testing demands for high-speed computing and advanced packaging chips.

As AI chips evolve toward chiplets, heterogeneous integration, and more complex advanced packaging architectures, the number of computing and memory components integrated within packages increases. Test interfaces must now support higher pin counts, high-frequency transmission, signal integrity, and precision. The more complex the chip design, the higher the difficulty of verification and mass production testing—favoring suppliers with advanced testing technology, precision manufacturing capabilities, and customer validation experience.

Global Semiconductor Market Projected to Reach $1.51 Trillion – AI Infrastructure and Memory as Key Drivers

Citing the latest forecast from the World Semiconductor Trade Statistics (WSTS), ChipMOS noted that driven by memory and AI infrastructure demand, the global semiconductor market is projected to reach $1.51 trillion in 2026, a year-on-year increase of 89.9%. It is expected to grow further to approximately $1.91 trillion in 2027, up 26.6% year-on-year.

The HPC market is also entering a rapid expansion phase. According to Yole’s forecast cited by ChipMOS, under a base scenario excluding TPU shipments, the global HPC market—including APU, CPU, GPU, and AI ASIC—is expected to reach $523 billion in 2026, up 46.2% year-on-year. It is projected to exceed $725 billion in 2027, up 38.6% year-on-year, with a CAGR of 27.1% from 2025 to 2029.

Including TPU shipments, ChipMOS’s own estimate suggests the global HPC market could reach $614 billion in 2026, up 47.2% year-on-year.

Huang believes AI applications are expanding beyond cloud data centers into factory automation, robotics, smart manufacturing, and daily life, broadening the variety of AI chips and testing scenarios. Future operational growth will depend not only on HPC demand from U.S. customers but also on the ramp-up of new capacity, progress in validating next-generation AI chips, and whether the high-end product mix can continue to support gross and operating margins.