Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsThe April 25, 2018 headline “Chinese DRAM Industry Spreading Its Wings: Two More DRAM Fabs Ready” described two projects approaching production—not two fabs already shipping memory at scale. One project, Innotron in Hefei, became part of the story of CXMT, now China’s principal indigenous DRAM maker. The other, JHICC in Fujian, did not follow a simple, uninterrupted path from construction to major commercial output. By 2026, China has a more substantial DRAM industry, but projected wafer capacity is not proof of competitive yields, low costs, or market parity.
What the 2018 headline meant by “ready”
The original AnandTech report, published April 25, 2018, said construction was substantially complete at two 300-mm fabs and that equipment installation or production preparation was under way. It described expected trial or volume-production timetables, not demonstrated high-volume commercial output. AnandTech’s 2018 report is therefore best read as a snapshot of China’s ambitions and factory build-outs at the time.
Semiconductor manufacturing has several distinct milestones: constructing a cleanroom, installing and qualifying tools, running test wafers, improving yields, qualifying products with customers, and shipping reliable product in volume. Reaching one milestone does not guarantee the next. A completed fab shell is not the same thing as a productive, commercially competitive DRAM operation.
The two projects named in the report
JHICC: a Fujian fab with an uncertain production path
Fujian Jin Hua Integrated Circuit (JHICC) was building a 300-mm fab in Jinjiang, Fujian. The 2018 report said construction had been completed in November 2017 and equipment installation was under way, with volume DRAM production expected in the third quarter of 2018. It reported a first-phase investment of about $5.65 billion and an expected total investment of about $8 billion. Those were figures and plans reported in 2018, not proof that the expected production ramp took place.
#1 Best Overall
- Boosts System Performance: 32GB DDR5 RAM laptop memory kit (2x16GB) that operates at 5600MHz, 5200MHz, or 4800MHz to improve multitasking and system responsiveness for smoother performance
- Accelerated gaming performance: Every millisecond gained in fast-paced gameplay counts—power through heavy workloads and benefit from versatile downclocking and higher frame rates
- Optimized DDR5 compatibility: Best for 12th Gen Intel Core and AMD Ryzen 7000 Series processors — Intel XMP 3.0 and AMD EXPO also supported on the same RAM module
- Trusted Micron Quality: Backed by 42 years of memory expertise, this DDR5 RAM is rigorously tested at both component and module levels, ensuring top performance and reliability
- ECC Type = Non-ECC, Form Factor = SODIMM, Pin Count = 262-Pin, PC Speed = PC5-44800, Voltage = 1.1V, Rank And Configuration = 1Rx8
The report also described a partnership with Taiwan’s UMC for 22-nm DRAM process development. It said the product plan was not yet clear, including whether JHICC would prioritize LPDDR4. Its construction status and timetable should not be mistaken for evidence of sustained qualified output. Later capacity models again place JHICC among potential Chinese DRAM producers, but those projections do not establish that its 2018 production target became a stable commercial business.
Innotron: the Hefei project that leads to CXMT
The same report said Innotron Memory had completed construction of a 300-mm fab in Hefei in January 2018 and was moving equipment into the facility. It planned late-2018 trial production of 8-Gb DDR4 chips, volume production in early 2019, and an initial target of about 20,000 wafer starts per month. A second phase and a future 17-nm process-development target were also discussed. These were plans, not confirmed later capacity or output.
Current coverage centers on ChangXin Memory Technologies (CXMT), the DRAM company based in Hefei. The 2018 report used the name Innotron; the available official corporate history confirms CXMT’s Hefei development but does not reproduce every former corporate name used in contemporary reporting. It is safest to describe Innotron as associated with the Hefei effort that became the CXMT story, rather than assert a simple legal name change.
Why factory completion was only the beginning
DRAM production depends on more than the physical plant. A manufacturer must get process equipment working together, make dies at acceptable yields, test and package them, and demonstrate that products meet customer requirements. Capacity announcements usually describe wafer starts, not how many good memory bits can be sold. The usable output depends on factors including process generation, die size, yield, product mix, and packaging and test capacity.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Commercial competitiveness adds another layer: customers need compatible products, dependable supply, reliability evidence, and qualification for their systems. Server and automotive applications, for example, have different validation and support demands from commodity PC memory. The gap between an announced production date and a durable supplier can therefore be measured in years of yield learning and customer qualification.
Rank #2
- Requires overclocking/BIOS adjustments. Maximum speed and performance depends on system components, including motherboard and CPU.
- G.SKILL Flare X5 Series DDR5 U-DIMM Memory Kit, Model: F5-6000J3636F16GX2-FX5
- Non-ECC, DDR5 U-DIMM, 288-pin, for Desktop PC & Gaming
- Includes JEDEC default profile, and AMD EXPO & Intel XMP 3.0 memory overclock profile
- Do not mix memory kits. Memory kits are sold in matched kits that are designed to run together as a set. Mixing memory kits will result in stability issues or system failure.
How China’s DRAM effort changed after 2018
CXMT’s official timeline provides a clearer account of the Hefei project’s progress. The company says it was founded in 2016; construction of Hefei Fab A began in March 2017, and Fab A1 was completed in January 2018. CXMT records an 8-Gb DDR4 tape-out in July 2018, successful product testing in May 2019, a mass-production announcement in September 2019, and its first customer order and product sales in November 2019. It says Beijing Fab C1 began trial production in January 2022 and Hefei Fab A2 was completed in June 2022. These are company-reported milestones, documented in its official history.
The sequence shows a real outcome—but not the instant ramp implied by reading “ready” as “already producing at scale.” The Hefei effort took additional development and qualification steps before CXMT announced mass production and recorded first sales. In 2026, the more consequential questions are how quickly CXMT can scale, improve yields, compete on cost, and win customers for newer products.
Newer products are evidence of progress, not proof of parity
CXMT describes itself as a vertically integrated DRAM company serving mobile, PC, tablet, server, and other markets. Its newsroom lists an LPDDR5 product-line announcement in November 2023, an LPDDR5X mass-production announcement dated October 28, 2025, and DDR5 and LPDDR5X demonstrations at IC China in November 2025. Those company announcements show a product portfolio moving beyond the 8-Gb DDR4 focus of 2018. A product announcement or trade-show demonstration alone does not establish broad shipment, global customer penetration, or parity with Samsung, SK hynix, and Micron in cost, yield, reliability, or server qualification. See the CXMT newsroom for the dated announcements.
JHICC’s later prospects remain a separate question
A 2026 capacity model discussed in Tom’s Hardware reporting includes JHICC among possible future Chinese DRAM producers, with an initial 60,000-wafer-starts-per-month phase and longer-term cleanroom capacity modeled at 120,000. The report discusses planned equipment activity, not verified full-rate, qualified production. Equipment arrival, if it occurs, would still be only one step in a production ramp. The 2018 JHICC timetable should not be treated as an uninterrupted success story.
What 2026 capacity estimates do—and do not—show
One analyst model cited by Tom’s Hardware projects CXMT at approximately 350,000 wafer starts per month by the end of 2026, near Micron’s modeled capacity. It also models Swaysure at about 140,000 WSPM and YMTC/XMC at about 50,000 WSPM of DRAM capacity, and puts potential Chinese DRAM capacity at about 600,000 WSPM in the coming years, excluding Samsung and SK hynix facilities located in China. The same model projects CXMT could reach roughly 950,000 WSPM by 2030. These are forecasts and supply-chain estimates, not audited production figures or confirmed good-wafer output. The estimates and their assumptions are summarized in Tom’s Hardware’s capacity report.
Rank #3
- Capacity: 16GB(2 x 8GB)
- Tested Frequency Profile 1: PC5-48000 (6000MT/s)
- Tested Timings: 36-46-46-110
- Feature Overclock: XMP 3.0 & EXPO overclocking supported
- On-Die ECC
| Company or group | Reported or modeled figure | What the figure means |
|---|---|---|
| CXMT | About 350,000 WSPM by the end of 2026; roughly 950,000 WSPM in a 2030 projection | Analyst-model capacity estimates, not confirmed shipments or output. |
| JHICC | 60,000 WSPM initial phase; 120,000 WSPM longer-term capacity | Reported planned or modeled capacity, not verified production. |
| Swaysure | About 140,000 WSPM | Analyst-model estimate; the figure does not establish qualified commercial output. |
| YMTC/XMC | About 50,000 WSPM of DRAM | Analyst-model estimate, not evidence of a mature DRAM business. |
Wafer starts per month cannot be translated directly into market share, bits shipped, revenue, or usable server modules. Those outcomes depend on good-die yield, how many bits fit on each wafer, product mix, qualification, and packaging. Market-share estimates can also differ depending on whether they count revenue, bits, shipments, or capacity. The central point is that a large modeled factory footprint signals potential scale; it does not by itself prove global competitiveness.
How to judge whether China is a serious DRAM competitor
Capacity is one indicator, but a fuller assessment asks whether products can be made efficiently, qualified by customers, and supplied reliably. CXMT’s DDR5 cost position is one warning against equating expansion with parity: a 2026 report citing SemiAnalysis estimates its DDR5 cost per bit is more than 30% above Samsung, SK hynix, and Micron. The same report estimates CXMT will add around 85,000 monthly DRAM wafer starts during 2026. These are attributed estimates, not audited company disclosures. Tom’s Hardware’s report on CXMT’s expansion and cost position discusses them.
- Technology generation: DDR4 establishes manufacturing capability; DDR5 and LPDDR5/5X are evidence of movement into newer product categories. HBM requires separate evidence.
- Yield and cost: Compare cost per usable bit rather than nominal wafer capacity. A higher-cost supplier may still serve protected or strategically important domestic demand, but it faces a harder case in global competition.
- Customer qualification: Named customer adoption and qualified volume shipments are stronger evidence than a product announcement or demonstration.
- Product breadth: Mobile, PC, server, automotive, and HBM products have different design, validation, reliability, and support requirements.
- Equipment and process control: Restrictions on access to advanced lithography and other equipment can affect process development and ramp speed. The effect depends on the specific item, technology, entity, and transaction.
- Reliability and commercial support: Enterprise and automotive buyers expect validation, traceability, sustained supply, and long product lifecycles.
DRAM, NAND, and HBM are not interchangeable
| Memory type | What it is used for | What the China story establishes |
|---|---|---|
| DDR4 and DDR5 DRAM | System memory in PCs and servers | CXMT’s history records DDR4 sales; its newsroom reports DDR5 demonstrations. These milestones do not alone establish global server-market parity. |
| LPDDR | Low-power memory used in mobile and thin devices | CXMT lists LPDDR5 and LPDDR5X product announcements; announcement scope is not the same as broad market shipment. |
| NAND flash | Nonvolatile storage in SSDs, phones, and other devices | YMTC is primarily associated with NAND. NAND capacity is not DRAM capacity and does not prove DRAM capability. |
| HBM | High-bandwidth memory stacked and packaged for accelerators and other demanding systems | The cited 2026 report says CXMT’s expansion is focused mainly on conventional DRAM and identifies no HBM project in its IPO prospectus. Conventional DRAM growth should not be read as HBM leadership. |
YMTC’s established identity is primarily as a NAND producer, though 2026 reporting says it has sampled low-power DRAM, may allocate a substantial portion of Wuhan Phase 3 to DRAM, and is exploring through-silicon-via packaging for HBM. Those reports indicate development and plans, not an established mainstream DRAM supplier. Its reported Wuhan expansion is covered by Tom’s Hardware. A move from NAND into DRAM involves different product architectures, process expertise, and customer qualification; the categories should not be conflated.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Ownership matters when counting China’s memory capacity
Chinese-owned DRAM production is not the same as foreign-owned fabs located in China. Samsung and SK hynix have operated semiconductor facilities in China, but their presence there does not establish indigenous Chinese DRAM technology or ownership. Likewise, a facility list or a reported wafer-capacity estimate does not by itself show that a fab is producing qualified DRAM. The distinction matters when evaluating supply security, domestic industrial capability, and who controls the technology and output.
China’s broader memory ecosystem includes CXMT in Hefei, YMTC/XMC in Wuhan, and JHICC in Jinjiang, but those sites and companies are not at the same stage or focused on identical products. A location overview is available in this CMBI report.
Rank #4
- Disclaimer: Maximum Speed requires overclocking/PC BIOS adjustments. Maximum speed and performance depend on system components, including motherboard and CPU
- AMD EXPO & Intel XMP 3.0 Compatible Only: Dual memory profiles allow you to easily select optimized settings for your platform, whether you’re running an AMD or Intel processor
- Dynamic RGB Lighting: Individually addressable RGB lighting delivers vibrant effects through a sleek, understated panoramic diffuser
- Onboard Voltage Regulation: Onboard voltage regulation for reliable power at high frequencies
- Maximum Bandwidth and Tight Response Times: Optimized for peak performance on the latest AMD and Intel DDR5 motherboards
What expansion could mean for buyers and the global market
A larger Chinese DRAM base could give domestic PC, smartphone, server, and electronics manufacturers another source of memory and reduce reliance on overseas suppliers for some products. That may matter even before Chinese producers match the leaders on cost: local demand and state-backed procurement can help a supplier gain manufacturing experience and customer footholds.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →For the global market, substantial new output could add competitive pressure in commodity DRAM and contribute to more fragmented sourcing. But the effect depends on actual qualified bits reaching customers, not cleanroom capacity alone. Higher costs, uneven yields, equipment constraints, and customer qualification requirements could limit how quickly projected capacity becomes a global supply alternative. No single capacity forecast establishes self-sufficiency or predicts a particular price outcome.
Export controls are part of this constraint picture. U.S. licensing rules can apply to certain advanced-node DRAM fabs and equipment transactions, but applicability depends on the item’s origin and technology, the entities involved, and the transaction. The cited U.S. House hearing document provides context; it is not a substitute for transaction-specific legal analysis.
Where the 2018 forecast stands
The headline correctly captured the direction of travel: China was trying to build a domestic DRAM industry, and the Hefei effort ultimately developed into CXMT, a significant indigenous manufacturer with documented DDR4 sales and later DDR5 and LPDDR milestones. But “ready” did not mean that both projects were already operating at scale, and JHICC’s announced 2018 timetable is not evidence of a continuous path to mature production. By 2026, China’s DRAM expansion is more than a fab-construction story; whether it becomes a durable global competitive force depends on yields, cost per bit, qualified shipments, product breadth, and reliable execution—not projected wafer starts alone.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Recommended Free Tools
