The artificial intelligence (AI) semiconductor market is evolving into the “inference” stage, intensifying a shortage of a key component: package substrates. In response, Samsung Electro-Mechanics (009150.KS) and LG Innotek (011070.KS) are simultaneously accelerating large-scale facility investments and the recruitment of key talent, entering a race for market dominance. Samsung Electro-Mechanics is investing 8 trillion won (approximately $5.4 billion) in its Sejong plant to build an AI substrate production hub while finalizing the hiring of on-site staff. LG Innotek, targeting an eightfold increase in operating profit by 2031, has begun expanding its factories in Gumi, South Korea, and Vietnam.

“A Parking Lot Becomes a Cutting-Edge Fab”… 8 Trillion Won Investment in Sejong Materializes

According to industry sources on the 13th, construction of Samsung Electro-Mechanics’ new factory for AI server package substrates is becoming visible. The company plans to utilize an idle parking lot site within the Myeonghak General Industrial Complex in Yeondong-myeon, Sejong City, to build a new three-to-four-story fabrication facility (fab). Samsung E&A, which secured the construction contract, recently began recruiting on-site contract workers for the Sejong Project (PJT), covering fields such as construction, clerical work, general affairs, and facility management. With the hiring of on-site personnel complete, full-scale construction is expected to begin in the second half of this year.

This is a follow-up to the investment plan announced on the 2nd at the “Chungcheong Region Advanced Industry Development Vision Public Report” held in Asan, South Chungcheong Province. At the time, Samsung Electro-Mechanics unveiled a blueprint to develop its Sejong plant into a production hub for AI server package substrates. Industry analysts view this as an attempt by Samsung Electro-Mechanics to seize the initiative in the high-value substrate market by accelerating expansion amid preemptive supply-hoarding competition among Big Tech firms.

“People Are the Technology”… The FC-BGA Talent War

Alongside facility investments, competition for high-skilled personnel is also fierce. As the performance of graphics processing units (GPUs), which act as the brains of AI semiconductors, and central processing units (CPUs), which coordinate all tasks, increases, the technical difficulty of flip-chip ball grid array (FC-BGA) substrates—which electrically connect these chips to the mainboard—is skyrocketing.

According to industry sources on the 12th, both Samsung Electro-Mechanics and LG Innotek have begun recruiting experienced professionals covering everything from FC-BGA substrate development to mass production, quality, and sales.

Samsung Electro-Mechanics is focusing on securing product development personnel responsible for optimal process design and new process development, as well as process development personnel to handle new package product processing technology and method improvements. Key tasks include designing large, high-layer-count semiconductor package substrates, creating designs optimized for high-performance servers and AI products, and developing new processes for next-generation servers. The recruitment drive also includes reliability testing and failure analysis staff, customer quality response personnel, and even marketing and sales staff targeting Big Tech clients in the Americas.

LG Innotek is also recruiting experienced professionals in semiconductor substrate production technology and product development. It is specifically prioritizing candidates with FC-BGA substrate mass production experience and those from outsourced semiconductor assembly and test (OSAT) companies. The strategy is to strengthen overall technological capabilities encompassing products, materials, methods, and mass production, going beyond simply expanding the production workforce.

Big Tech Lines Up Amid ‘Substrate Shortage’… Preemptive Competition via LTAs

Component companies are focusing on substrates because they have emerged as a core component in the AI era. AI accelerators must process massive amounts of data at high speed, making the role of the substrate—which quickly transmits electrical signals from the chip and stably supplies power—absolutely critical. In particular, as the center of the AI market shifts from learning to inference, demand for substrates needed not only for GPUs but also for CPUs is surging.

Consequently, major Big Tech clients are moving to proactively secure substrates through long-term agreements (LTAs). There is a growing number of cases where Big Tech firms provide advance payments or collaborate from the facility investment stage. This creates a win-win structure where component makers can receive upfront investment from clients to stably expand production lines, while Big Tech companies secure mid-to-long-term supply.

Aligning with this trend, LG Innotek has set a goal to increase the operating profit of its package solutions business nearly eightfold by 2031. To this end, it is pursuing the expansion of its Gumi plant for the mass production of new optical solution models and to increase AI substrate production capacity. It also plans to invest 1 trillion won (approximately $669.4 million) in its Vietnam factory by 2028.

An industry official stated, “As demand for high-value substrates has recently exploded, the expansion movement will accelerate further.” The official added, “Since AI server substrates require considerable experience from product development to mass production stabilization, the competition among domestic companies to secure core technical personnel is expected to become even more intense.”