Build1 publisher3 min readPublished
The chokepoint moved: ABF film, not lithography, now caps China's accelerator output
Ajinomoto has reportedly told mainland customers it will cut supply of high-end packaging film by 30 percent. It holds about 95 percent of that market; China supplies under 5 percent of its own.
The Engineer · Build desk
Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction

What happened
- Ajinomoto has reportedly told customers in mainland China that it will cut supply of ABF by 30%, according to a report from the Chinese outlet JW Insights, which cites unnamed supply chain sources.
- ABF is the insulating build-up film used in nearly every high-end processor package.
- Ajinomoto is reported to hold a 95% global market share of ABF film.
- China's ABF self-sufficiency rate is thought to sit below 5%.
- JW Insights attributes the cut to Ajinomoto prioritizing Japanese customers and core overseas accounts, which supply the FC-BGA substrates under Nvidia, AMD, and Intel accelerators, over mainland buyers.
Compiled by The EngineerSomething wrong?How this is made
Why it matters
Ajinomoto has reportedly told customers in mainland China that it will cut supply of ABF, the insulating build-up film used in nearly every high-end processor package, by 30 percent, according to the Chinese outlet JW Insights citing unnamed supply chain sources [1][2]. That matters because Ajinomoto is reported to hold roughly 95 percent of the global market for the film, while China's self-sufficiency rate is thought to sit below 5 percent [3][4] - which puts the binding constraint on Chinese accelerator and high-end CPU supply downstream of the wafer, in the substrate shop. JW Insights attributes the reported cut to Ajinomoto prioritising Japanese customers and core overseas accounts that supply the FC-BGA substrates sitting under Nvidia, AMD and Intel parts [5]. The named losers are Shennan Circuits, Xingsen Technology and Shenghong Electronics [6]. The volume cut is unconfirmed. The price move is not: Ajinomoto notified substrate makers in May of a roughly 30 percent price increase taking effect this quarter [7], two months after UK activist fund Palliser Capital disclosed a top-25 shareholding on 31 March and publicly demanded a price rise of more than 30 percent [8]. ABF is about 30 percent of a substrate's bill of materials [9], so a 30 percent material increase lifts substrate BOM cost by roughly 9 percent before anyone touches margin [10]. The arithmetic of relief is unfriendly. Ajinomoto ran about 2 million square metres a month at full utilisation in the second quarter [11], an annualised 24 million square metres [12]. It has committed 25 billion yen, around 156 million dollars, since 2023 to add roughly 50 percent of capacity by 2030, with a third plant on land in Kani City, Gifu Prefecture, not expected online until around 2032 [13][14]. Goldman Sachs models the ABF substrate supply-demand gap widening from about 10 percent in the second half of 2026 to 21 percent in 2027 and 42 percent in 2028 [15]. The new capacity therefore lands after the modelled peak shortfall [16]. Nor is there commercial pressure to discount: in the fiscal year ended 31 March, ABF sales grew 25 percent at margins above 50 percent, and the server and networking share of the film reached 70 percent, up from 40 percent in fiscal 2017 [17][18]. China has three films in qualification. Huazheng's CBF, developed with the Shenzhen Institute of Advanced Electronic Materials, uses modified epoxy resin with spherical silica filler to route around Ajinomoto's IP; Chinese media report mass-production yield above 85 percent, reliability testing passed inside Huawei Ascend systems, and validation underway at Xingsen and Shennan [19]. Its first line runs 3 million square metres a year at full utilisation, with a second doubling that at the end of 2026 [20] - about 12.5 percent of Ajinomoto's current annualised output today, 25 percent once line two ramps [21]. Shenzhen Newface's NBF, 51 percent acquired in April by MSG maker Lotus Holdings for about 103 million yuan, has qualified everything below nine build-up layers, with nine to 11 layers still in development [22]. Hongchang's GBF is in small-volume trial production with scale-up targeted for the fourth quarter [23]. The gating item is not R&D. Downstream reliability qualification takes one to three years of thermal cycling, damp-heat aging and electrical testing, often longer than development itself, and the highest layer-count films under flagship AI accelerators remain unmatched domestically [24]. Upstream specialty resins and spherical silica filler are themselves partly import-dependent [25]. One partial hedge already exists in silicon: the Ascend 910C reportedly links two compute dies on separate silicon interposers through an organic substrate, which SemiAnalysis has characterised as trading die-to-die bandwidth for yield and cost against Nvidia's CoWoS [26]. Watch whether the 30 percent volume cut gets corroborated beyond JW Insights, whether Hongchang's fourth-quarter scale-up holds, and whether Huazheng's second line lands on schedule at the end of 2026 [1][20][23].