2026.09.23 | AC | Semis | MLCC Distributor Check 2

摘要

Key takeaways of this article include:

1. Shifts in trader hoarding strategies
2. Progress in production conversion across Japanese, Taiwanese, and South Korean suppliers
3. Price and shipment pace comparison between Chaozhou Three-Circle(Group) and Guangdong Fenghua Advanced Technology
4. Outlook for silicon capacitors and related technologies

Below are expert views:

1. What is the current price gap between traders and original manufacturers for high-capacitance versus medium- and low-capacitance MLCC products? What is the expected trajectory for this gap going forward?

During the peak shortage period in June and July 2026, the gap between trader market prices and factory-gate prices from original manufacturers approached nearly tenfold for high-capacitance products. For instance, an item priced at 90 yuan from the factory gate commanded 800 to 900 yuan in the trader market. Over the same period, the price gap for medium- and low-capacitance products stood at roughly four to five times; an item with a factory-gate price of 30 yuan traded in the open market between 120 and 150 yuan.

As market dynamics shifted, the price gap for high-capacitance products narrowed to six to eight times. Meanwhile, the gap for medium- and low-capacitance products declined to around three times, with certain specifications dropping to just two times. Looking ahead, prices of medium- and low-capacitance products are expected to continue falling. By late 2026, the price gap between traders and original manufacturers in this category may shrink to around one time.

2. In light of current market trends, how will traders adjust their future hoarding strategies? What is their stance toward medium- and low-capacitance MLCC products?

Traders' future hoarding strategies will decisively favor high-capacitance products. For medium- and low-capacitance products, once existing stockpiles are liquidated, traders face two primary options. First, they can pivot toward high-capacitance components. Second, even if they remain in medium- and low-capacitance lines, they will selectively stockpile only high-volume, mainstream market specifications such as the 104 series, while steering clear of other specifications facing anticipated price slides. Alternatively, traders can exit the MLCC market altogether and reallocate capital into speculating on other components, such as memory chips. Although certain Japanese manufacturers announced plans to discontinue specific standard component series, those phaseouts do not begin until 2027, with final shipments scheduled as late as 2029. This provides the market with a multi-year buffer. Domestic suppliers' production capacity is expected to be sufficient to fill market vacancies in the interim, making another severe shortage in medium- and low-capacitance products highly improbable. On balance, traders have an extremely low appetite for hoarding medium- and low-capacitance products through 2026.

3. What are traders' main rationales for shifting toward the memory chip market? Which types of memory products do they tend to target?

Traders view the memory chip supply chain landscape as more consolidated and stable than MLCCs, with potentially longer shortage cycles. They note that the global memory market is primarily dominated by a handful of overseas giants—namely Samsung, Micron, and SK Hynix—which have sustained prices at elevated levels and tend to allow supply deficits to persist. While domestic suppliers like CXMT can absorb a portion of orders, their pricing will not be set excessively low given broader global tightness. In contrast, the MLCC sector faces an influx of new capacity from domestic manufacturers, which could easily trigger renewed price competition. For traders who lack sourcing channels for high-end MLCCs, shifting funds into the memory market represents a superior allocation of capital. Their attention is centered primarily on DRAM and NAND Flash.

4. From an end-customer standpoint, are there clear performance or application differences between domestic high-capacitance MLCCs and overseas brands from Japan, Taiwan, and South Korea?

In the medium- and low-capacitance domain, consumer electronics clients across the board generally perceive virtually no difference between domestic and overseas brands. For high-capacitance offerings, however—particularly compact form factors with elevated capacitance values—gaps persist. Certain end customers, especially prominent brand owners, remain wary of utilizing high-capacitance parts from specific domestic brands. This caution stems from historical quality incidents involving domestic components, and large corporations maintain stringent reliability requirements that make them highly averse to such risks. While they are willing to accept and design in domestic parts, mature technical execution and proven quality consistency remain non-negotiable prerequisites. In today's hyper-competitive climate, any significant quality defect could precipitate a sharp drop in end-product unit sales.

5. Over the past one to two months, has AI-related demand fallen short of expectations?

For now, AI-related demand has not lagged expectations; rather, it has tracked sideways. Demand has remained broadly stable since Q2 2026. While a few isolated specifications registered marginal upticks, wide-scale, sharp acceleration has not materialized. Dialogue with server and compute accelerator card clients indicates that end-customer procurement roadmaps point to demand scaling up around 2027. Full-year 2026 demand is projected to stay near prevailing levels, without substantial incremental growth.

6. Is there a more concrete forecast regarding the growth of AI-related demand in 2027? What will the scale of increase look like compared with 2026?

AI industry demand is expected to expand by approximately 30% in 2027. This growth projection assumes that end-customer order forecasts are genuine and free of artificial inflation.

7. Compared to one or two quarters ago, has there been any meaningful change in suppliers' eagerness to expand capacity or convert production lines?

Expansion strategies across suppliers have diverged. Japanese vendors plan to expand overall capacity by roughly 15% in 2027, dedicating 80% of that new capacity to AI-related products while allocating virtually nothing to consumer lines. Samsung plans a 10% capacity expansion, choosing to convert select consumer production lines toward server products in response to growing AI market orders rather than adding further greenfield capacity. Domestic manufacturers have adopted more aggressive postures, albeit with internal divergence. Tier-one leaders like Guangdong Fenghua Advanced Technology and Chaozhou Three-Circle(Group) are projected to cap expansions at around 20%. In contrast, smaller and mid-sized domestic players, operating off smaller capacity baselines, are targeting twofold to threefold expansions in 2027, with some aiming as high as three- to fourfold increases.

8. Which customer segments and product types are domestic small and mid-sized MLCC manufacturers primarily targeting with their capacity expansions?

Expansions by these smaller vendors are aimed almost exclusively at consumer electronics clients, focusing heavily on medium- and low-capacitance items. Currently, only three domestic suppliers are capable of fulfilling overseas consumer-grade orders for high-capacitance products: Three-Circle, Fenghua, and Microgate. Other domestic vendors have yet to clear technical hurdles in high-capacitance segments, and their narrow product portfolios prevent them from capturing business from major Tier-1 accounts. Consequently, these players are forced to expand medium- and low-capacitance output to contest market share, which constitutes a primary catalyst behind anticipated downward price pressure in that category.

9. In terms of shipping strategies, fulfillment cadence, and pricing, do meaningful differences exist between Chaozhou Three-Circle(Group) and Guangdong Fenghua Advanced Technology?

Distinct strategic differences exist between the two, with Chaozhou Three-Circle(Group) exhibiting greater flexibility. As a state-owned enterprise, Guangdong Fenghua Advanced Technology operates under lengthier decision-making timelines and broader operational considerations. Furthermore, its portfolio spans resistors, ferrite beads, filters, and inductors alongside MLCCs, meaning MLCCs do not represent its exclusive priority. As a result, Fenghua maintains incomplete coverage or restricts capacity allocations across certain product specifications, particularly those plagued by suboptimal production yields.

Conversely, Chaozhou Three-Circle(Group) responds to market shifts with greater agility and aggressively pursues orders across diverse customer segments, affording it superior latitude in both allocation and pricing tactics. In the field, Three-Circle demonstrates substantially faster onboarding cycles for new accounts than Fenghua. Fenghua's revenue expansion relies more on deepening engagement with established relationships—namely key accounts and fellow central/state-owned enterprises—by expanding purchase volumes on existing part numbers or qualifying new catalog lines, whereas its pace in acquiring brand-new clients remains comparatively deliberate.

10. For high-capacitance MLCC lines, how do Three-Circle Electronics' prices compare with Guangdong Fenghua Advanced Technology's? How do their price adjustment tempos and commercial strategies differ?

At present, Guangdong Fenghua Advanced Technology's quotes for high-capacitance MLCCs typically run 10% to 20% higher than those of Three-Circle Electronics. In adjusting pricing, Three-Circle Electronics moves at a faster tempo and employs a more pliable approach. Driven first and foremost by order capture, Three-Circle is open to pricing concessions, especially when courting tier-one, high-volume clients. In contrast, Fenghua has historically pursued a more conservative posture in its MLCC operations, showing little interest in unconstrained volume growth. This strategic divergence has created a noticeable variance in the pace and nature of market expansion between the two vendors.

11. Looking at recent industry developments, what new steps have Murata, Taiyo Yuden, and Samsung Electro-Mechanics taken regarding capacity additions or product conversions? In the server space specifically, has there been any observable shift in Taiyo Yuden's and Samsung Electro-Mechanics' market influence regarding pricing leverage or order capture?

In 2026, Taiyo Yuden is stepping up capital expenditures for capacity additions. Its expansion programs, extending through 2027, are more aggressive than Murata's, particularly within AI server-grade MLCCs. This stance is largely a function of Taiyo Yuden's smaller baseline capacity: Murata commands monthly capacity of roughly 140 billion units, whereas Taiyo Yuden currently stands near 40 billion units per month. Despite this push, Taiyo Yuden will not pursue the multi-fold expansion multiples seen among Chinese mainland peers, given the concentration risk inherent in its server-heavy exposure. In the overseas server arena, the vendor hierarchy is led by Murata in automotive-grade lines, followed by Taiyo Yuden among Japanese suppliers, and Samsung Electro-Mechanics in third. As such, Taiyo Yuden's commercial strategy continues to anchor firmly to the server market.

Samsung Electro-Mechanics is likewise intent on gaining server market share. However, with incumbent Japanese leaders unwilling to cede ground readily, Samsung is seeking out alternative product categories and adjacencies while fulfilling select server allocations. For instance, Samsung has expanded into silicon capacitors, successfully partnering with server customers and booking design wins. Historically, the silicon capacitor landscape was governed primarily by TSMC and Murata. Overall, Samsung's roadmap focuses on migrating into premium MLCCs where viable; where blocked, it leverages alternative product form factors—such as silicon capacitors—to establish commercial footholds with server clients.

12. In server applications, could MLCCs be substituted by competing technologies like silicon capacitors, lithium-ion capacitors, or supercapacitors, resulting in a net reduction in future MLCC consumption?

While emerging options like silicon capacitors can substitute for MLCCs across specific functions, a net decline in overall MLCC consumption within servers is improbable. The underlying driver is that design power envelopes across forthcoming server generations will continue climbing, inevitably mandating higher total counts of MLCCs. Thus, even if alternate technologies displace select MLCC sockets, total MLCC unit consumption will continue expanding thanks to the fast-growing baseline demand across server architectures.

Over the next three years, the probability of an alternative technology achieving widespread displacement and meaningfully curtailing MLCC consumption is exceptionally slim. Taking silicon capacitors as an example, despite their profile as an emerging alternative, they remain incapable of matching MLCCs across select mission-critical parameters. Delivering equivalent performance characteristics results in manufacturing costs vastly exceeding those of MLCCs, leaving them uncompetitive on cost even following MLCC price hikes. As a result, no comprehensive technological substitute for MLCCs exists over the near-term horizon.