Two separate announcements on May 29 landed in the same corner of the semiconductor industry, and their combined implications for South Korea's chipmakers are more significant than either story suggests on its own. Nikon's new chief executive told Nikkei Asia that the Japanese optics maker is in advanced talks with major chipmakers in the United States and Asia to supply lower-priced deep-ultraviolet lithography tools, using in-house component manufacturing to undercut a market where ASML currently holds more than 80 percent. On the same day, Intel, US-based startup 3D Glass Solutions, and the government of India's Odisha state signed a $3.3 billion agreement to build one of the country's largest high-technology manufacturing investments: a glass-core substrate facility in the Bhubaneswar–Khurda region. The two moves — one compressing equipment costs upstream, the other staking out next-generation packaging territory — are precisely the variables that Samsung Electronics and SK hynix will need to track as they compete in the AI hardware supply chain.
Nikon Challenges ASML: In-House Manufacturing Enables Lower ArF Prices
Nikon's new president, Yasuhiro Ohmura, who took office on April 1, 2026, told Nikkei Asia that the company's vertically integrated manufacturing model gives it a structural pricing advantage that ASML, which depends on a global supply chain including German optics supplier Carl Zeiss and American light-source specialist Cymer, cannot easily replicate. "We manufacture many parts in-house, giving us an advantage in cost competition," Ohmura said in the interview. ASML's advanced ArF immersion machines average approximately $82.5 million per unit, a figure that leaves room for a credible price competitor.
The context behind Nikon's announcement is a period of unusual weakness. The company shipped 11 ArF systems in its fiscal year ending March 2024 and recorded no ArF sales in the first three quarters of fiscal 2025, against ASML's grip on more than 80 percent of the lithography equipment market. A consolidated net loss of 86 billion yen for the fiscal year ending March 2025 underscored the urgency of Ohmura's five-year Medium-Term Management Plan, which repositions semiconductor lithography equipment as the company's growth engine through 2031. Nikon and ASML are currently the only two manufacturers that produce ArF immersion scanners, giving chipmakers limited leverage in a segment where AI-driven demand is putting pressure on available tool supply.
Why DUV Still Matters in an EUV World
Nikon's strategy is confined to the deep-ultraviolet segment — specifically ArF immersion systems, which handle nodes from roughly 7 nanometers upward and certain multi-patterning applications. The company walked away from EUV development in 2008, ceding that territory entirely to ASML, which now holds a monopoly on EUV tools. That limitation is real and should not be minimized: leading-edge logic at 3 nanometers and below requires EUV, and Nikon cannot compete there.
But the DUV segment is not a backwater. The majority of patterning steps in modern chip manufacturing — including in the production of 3-nanometer logic — still run on ArF immersion tools. Memory production and legacy-foundry lines rely heavily on DUV. Industry forecasters project the DUV lithography equipment market will reach approximately $23.7 billion by 2031, growing at a compound annual rate of roughly 6.4 percent. If Nikon's pricing lands at a meaningful discount to ASML's $82.5 million average, Samsung and SK hynix — both of which operate large memory lines requiring continuous DUV replenishment — could realize materially lower equipment-depreciation burdens, supporting operating margins in an already favorable demand environment.
Critically, Nikon's new ArF immersion platform, which it plans to launch in fiscal 2028 (April 2028 through March 2029), is being designed for compatibility with ASML's existing tools. That design choice reduces the integration risk for fabs considering a dual-sourcing strategy. Ohmura confirmed in the Nikkei Asia interview that discussions with major chipmakers in the US and Asia were nearing purchase orders — a meaningful step for a company that posted zero ArF sales across nine consecutive months in 2024 and 2025.
How Does Glass Core Substrate Technology Work?
The Intel–3DGS–Odisha deal addresses a different constraint: the packaging materials that connect and house the chips once they are manufactured. Conventional advanced packaging relies on organic substrates — plastic-based platforms that carry fine circuit lines. Glass-core substrates replace that organic base with a thin glass panel, which offers two specific physical advantages. Glass is flatter than plastic, enabling more precise fine-pitch circuitry; and its low coefficient of thermal expansion reduces warpage in large chip packages, a problem that becomes more acute as AI accelerators grow larger and denser.
For high-performance AI chips — Nvidia's GPU accelerators or AMD's Instinct series, for example — a substrate that stays flat under thermal stress and supports finer interconnect lines is not an incremental improvement; it is a prerequisite for the next generation of chip architectures. Intel first publicly outlined its glass substrate technology roadmap in 2023 and made early investments in developing the process. By mid-2025, industry reports suggested the company was reconsidering its in-house approach and looking toward external sourcing. The Odisha memorandum of understanding, signed May 29, 2026, represents a significant reversal of that direction — and a bet that positioning India as a glass substrate production hub, with Intel's process expertise and 3DGS's glass technology intellectual property, can be commercially viable at scale.
The Bhubaneswar–Khurda facility is planned to produce both glass-core substrates and high-density interconnect substrates for next-generation packaging, and is expected to generate approximately 1,800 direct high-skill jobs over five to six years. Mass production is estimated to begin between 2028 and 2030. The deal also reflects converging national interests: the United States is seeking to diversify critical semiconductor supply chains away from a geographic concentration in East Asia, while India's government has made semiconductor manufacturing a strategic priority.
Three Metrics That Will Determine Who Wins the Packaging Race
The May 29 announcements alter the competitive map for Samsung Electronics and SK hynix, but their actual impact will only become visible through three specific indicators.
The first is Nikon's share trajectory in DUV lithography. Lower prices only translate into financial benefit for chipmakers once Nikon's systems enter production lines at scale. The company's prototype delivery target — to a named but unidentified major semiconductor manufacturer — is expected before the end of 2027. Whether Samsung or SK hynix features among those initial customers will determine how quickly the equipment-cost advantage materializes.
The second is when the Intel–3DGS Odisha facility reaches target glass substrate yields. Glass is technically demanding to manufacture. Current industry estimates put glass substrate manufacturing yields in the 75 to 85 percent range, lower than the benchmarks needed for cost-effective high-volume production, due primarily to brittleness and the complexity of the through-glass via formation process. The Odisha facility is not expected to reach mass production until 2028 at the earliest. Who controls the yield curve at that point — Intel and 3DGS in India, Samsung Electro-Mechanics with its Sejong-based pilot line, or Absolics (an SKC affiliate) ramping output from its Georgia facility after beginning AMD qualification sampling in early 2026 — will shape which companies gain first-mover pricing leverage.
The third is how quickly Samsung Electro-Mechanics and Absolics can secure glass substrate qualification from major fabless customers. Samsung Electro-Mechanics is targeting commercial glass substrate prototypes in 2026. Absolics has been in AMD qualification sampling from its Georgia fab since early 2026. Landing qualification from Nvidia or AMD ahead of the Intel–3DGS ramp in India would be a prerequisite for any meaningful share gain — and would also complicate Intel's stated ambition to establish the design standards that fabless chip companies will be required to follow. If Intel succeeds in setting those standards, companies worldwide could face specifications derived from Intel's own process intellectual property, a scenario Korean industry observers have flagged as a structural threat to a Korea-led glass substrate ecosystem.
Frequently Asked Questions
Can Nikon realistically challenge ASML's dominance in lithography equipment?
Nikon's challenge is credible but limited in scope. The company can compete on price in the deep-ultraviolet ArF immersion segment, where it has a structural cost advantage from in-house manufacturing and ASML does not hold an outright monopoly. It cannot compete in extreme ultraviolet lithography, where ASML is the sole commercial supplier. The practical effect is downward pricing pressure on ArF tools — meaningful for memory and legacy-foundry lines at Samsung and SK hynix, but irrelevant for cutting-edge logic production below 7 nanometers.
What is glass-core substrate technology and why does it matter for AI chips?
A glass-core substrate replaces the plastic base layer in advanced chip packaging with a thin glass panel. Glass provides a flatter, more thermally stable surface than conventional organic materials, which reduces warpage in large AI accelerator packages and enables finer circuit patterns. These properties are critical for next-generation AI chips that need to connect multiple large dies in a single package without the distortion that heat and size create in organic substrates.
How does Intel's $3.3 billion India deal affect Samsung and SK hynix?
The Intel–3DGS Odisha facility, targeting mass production between 2028 and 2030, introduces a potential US-India supply node into a glass substrate market where Samsung Electro-Mechanics and Absolics currently hold the leading commercialization positions. If Intel and 3DGS establish the design standards for glass-core substrates, Korean producers may face a specifications framework derived from a competitor's process intellectual property. Conversely, if Samsung Electro-Mechanics or Absolics secures Nvidia or AMD qualification before the Odisha plant reaches scale, the Korea-led ecosystem retains its competitive advantage.
What three metrics should investors and supply-chain analysts track?
The three leading indicators are: first, Nikon's actual market share gains in DUV lithography — pricing matters only once tools are in production at Samsung, SK hynix, or TSMC fabs; second, when the Odisha facility achieves target glass substrate yields, with mass production expected between 2028 and 2030; and third, whether Samsung Electro-Mechanics or Absolics secures glass substrate qualification from Nvidia or AMD before Intel's India plant reaches commercial scale, which would be a prerequisite for any upward re-rating of Korean packaging names.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.
