How an MSG Factory Holds Nvidia by the Throat? What Is ABF Material?

TradingKey - The 2026 AI computing power race appears to be dominated by GPUs, HBM, and advanced packaging, but the real factor determining delivery pace is a Japanese company that started in MSG—Ajinomoto. Its ABF commands a market share of over 95% in the global high-performance chip packaging materials market, and NVIDIA ( NVDA ), Intel ( INTC ), Advanced Micro Devices ( AMD) are all unable to bypass this critical link.

What is ABF? Why is it indispensable for AI chips?

ABF stands for Ajinomoto Build-up Film. It is a high-performance organic insulating film used in FC-BGA (Flip Chip Ball Grid Array) packaging substrates to form multi-layer interconnect structures, acting as a "bridge" between the chip and the PCB circuit. In layman's terms, if the chip is the top floor of a high-rise building and the PCB is the ground, ABF is the insulation layer between floors—without it, high-frequency signals would experience crosstalk, and even the most advanced chips would fail to function.

The technical barriers for ABF are extremely high: it must simultaneously satisfy requirements for low thermal expansion, low dielectric loss, and high insulation, while maintaining exceptional flatness and yield during multi-layer stacking. A minute flaw in any layer can result in the entire packaging substrate being scrapped. Leveraging nearly 30 years of expertise, Ajinomoto has established a near-monopoly in the ABF material market for GPU and CPU packaging substrates, with a market share exceeding 95%. Its only competitor, Sekisui Chemical, entered the market in 2014 and currently holds a share of only about 5%.

How Ajinomoto Transitioned from MSG to Semiconductors?

Ajinomoto’s connection to the semiconductor industry dates back to the 1970s. The food giant, famous for Kikunae Ikeda’s discovery of "umami," began researching potential uses for byproducts from MSG production. During amino acid chemistry research, the team unexpectedly discovered that a byproduct could be transformed into a thermosetting film with exceptionally high insulation properties and low thermal expansion.

The true turning point came in 1996 when Intel proactively contacted Ajinomoto, seeking to leverage its amino acid technology to develop film-type insulators. The two collaborated to develop the FC-BGA packaging solution, and ABF has since become the primary insulating material for this packaging technology. The Ajinomoto team completed the development of ABF in just four months, and official mass production began in 1999 with Intel as its first customer.

In the nearly three decades since, ABF has gradually become an irreplaceable core material for high-end CPU and GPU packaging substrates.

How AI demand is fueling a surge in ABF usage

With iterations of NVIDIA's AI accelerators like Blackwell and Rubin, the complexity of packaging substrates is increasing exponentially. Traditional PC chip packaging requires only 4 to 6 layers of ABF, while AI accelerators have already increased this to 8 to 16 layers. ABF usage in high-performance CPUs is more than 10 times that of ordinary PC substrates, with top-tier AI accelerators reaching 15 to 18 times. ABF is gradually becoming the core bottleneck for AI chip production capacity.

Explosive demand stands in sharp contrast to the rigidity of supply. Reports from U.S. investment banks show that the ABF substrate supply-demand balance turned tight at the end of 2025 and is intensifying monthly: the shortfall is projected to reach 10% in the second half of 2026, widen to 21% in 2027, and rise to 42% in 2028. Fubon Securities offers an even more aggressive forecast, predicting a 40% annual demand increase in 2027 against a supply growth of just 12%, resulting in a 26% gap that widens to 46% by 2028. The demand structure is also shifting profoundly—the PC share is expected to shrink from approximately 70% in 2015 to about 15% by 2030, while AI-related chips will soar from roughly 10% to 75%.

Supply bottlenecks stem not only from Ajinomoto’s own capacity constraints but also from shortages of upstream raw materials. Tight supply of high-end glass cloth (T-glass) is expected to result in an actual supply gap exceeding 40% in the second half of 2026, potentially triggering market double-booking. Due to material constraints, ABF substrate manufacturers have generally postponed their expansion plans by 6 to 12 months.

Facing a supply crisis, Ajinomoto plans to invest at least 25 billion yen by 2030 to increase ABF capacity by 50%. However, whether a 50% increase can match the double-digit annual growth in AI computing power remains highly uncertain. Morgan Stanley expects a significant shortage of ABF substrates in 2027, with a market compound annual growth rate (CAGR) of approximately 16.1% from 2025 to 2027. Goldman Sachs predicts a 33% CAGR from 2025 to 2028, driven primarily by the size of AI GPU and ASIC substrates, which are projected to grow 2.5 to 4 times within two years.

Palliser intervenes, demanding a 30% price hike

Ajinomoto's monopoly in the ABF market has long been underpriced by capital markets. As one of Ajinomoto's top 25 shareholders, Palliser Capital released a value enhancement plan titled "The Most Undervalued AI Infrastructure Monopoly Goldmine" on March 31, 2026, explicitly calling for two key demands: a more than 30% price hike for ABF and the spin-off of the electronic materials business to improve transparency.

Palliser's logic is straightforward: ABF costs account for less than 0.1% of the total GPU unit price, meaning a 30% price increase would impose negligible pressure on customers like Nvidia while significantly boosting Ajinomoto's profit elasticity. The fund emphasized that addressing the valuation discount could unlock more than 70% upside for shareholders.

Capital markets reacted swiftly. On February 5, Ajinomoto raised its FY2025 revenue target for its electronic materials business from 84.9 billion yen to 97.9 billion yen (a 28% year-on-year increase) and its business profit target from 43.5 billion yen to 52.5 billion yen (a 31% year-on-year increase).

As of April 2026, Ajinomoto's stock price has risen by more than 40% cumulatively, hitting a record closing high of 4,968 yen on February 27. These financial signals indicate that, driven by the AI wave, Ajinomoto's electronic materials business is transforming from a "neglected side business" into the company's most valuable growth engine.

Notably, this is not the first time Palliser has targeted "hidden AI assets" within Japan's traditional manufacturing sector. In February 2026, the fund took a stake in Toto, a company famous for its sanitary products, demanding disclosure of the true value of its specialty ceramics business used in semiconductor manufacturing and labeling it as the "most overlooked AI memory beneficiary."

Which substrate manufacturers will be impacted by the ABF shortage?

ABF film must be processed by substrate manufacturers before delivery to chip companies like NVIDIA. With tight upstream material supplies coupled with surging AI demand, substrate manufacturers are entering a new supercycle.

Taiwan's three major substrate manufacturers delivered outstanding results: Unimicron's January 2026 revenue hit NT$12.767 billion, up 34.48% YoY; Kinsus reported monthly revenue of NT$3.961 billion, up 54.94% YoY, a record high; Nan Ya PCB's revenue reached NT$3.72 billion, up 44.98% YoY. In March, Nan Ya PCB's revenue climbed further to NT$4.29 billion, up 35% MoM and 39% YoY, a 36-month high. Goldman Sachs concurrently raised target prices: 'Buy' for Kinsus at NT$370 and 'Buy' for Nan Ya PCB at NT$655.

This cycle differs significantly from the 2020-2022 peak. US-based investment banks pointed out that while the previous round was driven by pandemic-induced supply disruptions, the current round is fueled by an explosion in real AI demand, with even tighter upstream material constraints; thus, the upcycle could last until the second half of 2028. Regarding pricing, ABF substrate prices are estimated to rise 3%-5% QoQ in Q1 2026, maintaining a roughly 10% QoQ increase for the subsequent three quarters, with the rally potentially widening in 2027. As the global leader, Unimicron is expected by Goldman Sachs ( GS) to see its 2026 substrate revenue grow 22% YoY; Kinsus plans to expand capacity by approximately 25% in 2027, highlighting superior supply flexibility; Nan Ya PCB is poised for double-digit growth in 2026, benefiting from the simultaneous volume ramp-up of two new ASIC customers.

Will NVIDIA’s next-generation chips also be constrained by ABF?

Every generation of NVIDIA's GPU upgrades translates to higher packaging density and more ABF layers, directly boosting demand for Ajinomoto. In 2026, the shipment mix of NVIDIA's high-end AI chips will evolve: the Blackwell series' share is expected to rise from 61% to 71%, as the GB300 replaces the GB200 as the flagship model in Q4 2025, reaching nearly 80% of shipments by 2026. The next-generation Rubin platform, slated for release around Q3 2026, will further push ABF density requirements. NVIDIA faces coordination challenges across the entire supply chain, from chip design to raw material sourcing.

A critical prerequisite for the successful mass production of the Rubin platform is whether ABF supply can scale in tandem. Market analysis suggests that as AI chip footprints expand and packaging layers proliferate, this once-obscure chemical material now directly influences delivery schedules and overall computing costs. Several hyperscale cloud providers, recognizing this latent crisis, have begun assisting Ajinomoto in constructing new production lines via prepayments and long-term contracts to lock in future capacity.

What are the potential risks of Ajinomoto's ABF monopoly?

Market focus initially centered on GPU design, HBM memory, and CoWoS packaging, but the true opportunities for alpha may lie with "hidden champions"—long-undervalued companies that control irreplaceable resources. However, the monopoly dominated by Ajinomoto is not without its share of concerns.

First, Ajinomoto's core business remains food, and electronic materials account for only a limited portion of revenue. Uncertainty persists regarding whether the company is willing to treat the segment as a core strategic pillar. Second, the sole competitor, Sekisui Chemical, holds only about a 5% market share; however, should it make progress in capacity expansion or technological breakthroughs, Ajinomoto's monopoly premium will face erosion. A deeper question remains: how long can this fragile balance be maintained when the critical lifeblood of the AI supply chain is held by a multinational food corporation?

From seasonings to insulation films, and from kitchens to wafer fabs, Ajinomoto's transformation has spanned half a century. This centennial enterprise has used its most inconspicuous "by-product" to secure a stranglehold on the world's most cutting-edge technology industries.