Why Micron Is Suddenly Ditching 4F² DRAM As Rivals Double Down
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Buying for a business?Offer from Amazon

Get business pricing on networking and server gear

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

Micron has decided not to bring 4F² DRAM products to market, despite more than two decades of research, and plans to continue scaling its 6F² cell design. The company says capacitor spacing limits density gains from 4F², while the approach adds cost and complexity; rivals including Samsung, SK hynix and CXMT are pursuing 4F² designs.

Micron says it will not bring 4F² DRAM to market, ending more than two decades of work on the cell design and opting to keep advancing its established 6F² architecture. The decision sets the company apart from rivals pursuing 4F² and reflects Micron’s view that tighter cell geometry offers too little practical density improvement for its added cost and manufacturing complexity.

At a roadmap webinar on Wednesday, Micron outlined several upcoming memory technologies and said it is concentrating on continued 6F² development. The company reported that high-density 1δ (1-delta) DRAM is ramping yields. Its next listed step, 1ε (1-epsilon), is in development and uses advanced wafer bonding. Micron also named “ζÅ” and “ηÅ” as later stages in its planar DRAM roadmap.

Micron President and CTO Scott DeBoer said DRAM density is constrained by the spacing needed for capacitors, not just the footprint of the memory cell. In the company’s account, changing from 6F² to 4F² therefore does not produce the theoretical density improvement suggested by the geometry alone, while raising cost and complexity. Micron said it had worked on 4F² for “over two decades” before deciding not to commercialize it.

The roadmap also points toward 3D DRAM, in which storage cells themselves would be stacked vertically. That is different from bonding memory to logic. Micron presented 3D DRAM as a longer-term direction rather than an imminent product, given the engineering challenges posed by DRAM’s capacitors and the need to preserve signals and manage heat across stacked layers.

At a glance
updateWhen: Announced at a Micron roadmap webinar o…
The developmentMicron’s roadmap presentation set out a decision to forgo commercial 4F² DRAM and continue with 6F² scaling while developing future memory technologies.

Micron Bets on 6F² Economics

The choice may affect how Micron competes on DRAM density, manufacturing cost and investment as memory demand grows. If 4F² can be made economically at scale, companies that adopt it could gain a density advantage. Micron is instead betting that improvements to 6F² can deliver sufficient progress without taking on the expense and complexity it associates with 4F².

The consequences are not settled. Micron’s position could prove advantageous if rivals encounter cost or scaling limits, but it could leave the company at a disadvantage if competitors successfully manufacture 4F² memory with useful density and cost benefits. The source material offers no comparative production data or independent evidence establishing which approach will win.

Micron’s longer-term 3D ambitions add another dimension to the decision: the company is directing research toward stacking DRAM cells rather than relying on a 4F² layout. But 3D DRAM remains a development goal, not a near-term substitute that has been shown to solve current scaling pressures.

Products Worth Considering

Samsung Galaxy Tab A11+ 11-Inch Tablet, 6GB RAM, 128GB Storage, Expandable, Gray

Samsung Galaxy Tab A11+ 11-Inch Tablet, 6GB RAM, 128GB Storage, Expandable, Gray

Jackery Explorer 1000 v2 Portable Power Station, 1070Wh, 1500W

Jackery Explorer 1000 v2 Portable Power Station, 1070Wh, 1500W

As an affiliate, we earn on qualifying purchases.

Why DRAM Cell Geometry Matters

DRAM cells combine a transistor and a capacitor, unlike the switching-transistor structures used in logic chips. The 6F² and 4F² labels describe cell geometry and layout; they are not process-node names and do not, by themselves, specify the physical dimensions of a product. Actual scaling also depends on manufacturing processes and the spacing the components require.

Micron’s roadmap uses labels such as 1δ and 1ε for successive DRAM generations. Its choice to stay with 6F² is unusual in the competitive landscape described in the report: Samsung, SK hynix and China-based CXMT are pursuing 4F² designs. The report says CXMT claims it plans to have 4F² memory in server DDR5 RDIMMs by the end of the year; that is a company claim, not confirmation that such products are already available.

Three-dimensional stacking is familiar in flash memory, but DRAM’s storage cells have different requirements. Each bit depends on a capacitor that must retain charge, making vertical integration a substantial architectural challenge. Micron’s roadmap places 3D DRAM beyond its nearer-term planar development steps.

How Rivals’ 4F² Plans Compare

It remains unclear whether Samsung, SK hynix or CXMT can manufacture 4F² DRAM at scale and at a cost that makes the design commercially worthwhile. The source material gives no independently verified production yields, cost comparisons, density measurements or shipping schedules for their efforts. CXMT’s reported target for server DDR5 RDIMMs by year-end is a company claim and does not establish that delivery will occur.

Micron has also not provided a quantified comparison showing how much density 4F² would add in a finished product, how much more it would cost, or when 3D DRAM could become commercially viable. Its 6F² roadmap milestones describe development status, but the material does not give dates for all later generations or disclose product-level performance figures.

Micron’s Roadmap Milestones

The nearest milestone in Micron’s account is continued yield improvement for 1δ DRAM, followed by development of 1ε with advanced wafer bonding. The company has also identified ζÅ and ηÅ as future planar stages, though the source does not provide launch dates for them.

For the broader industry, the next evidence to watch is whether rivals move from 4F² development claims to products made at commercial scale, and whether those products show a measurable cost or density advantage. Micron’s longer-term test will be whether it can advance 3D DRAM from a roadmap ambition to a manufacturable technology. No timetable for that transition was specified in the source report.

Key Questions

What has Micron decided about 4F² DRAM?

Micron will not bring 4F² DRAM products to market, according to its roadmap announcement. It plans to continue scaling 6F² DRAM instead.

Why is Micron sticking with 6F²?

Micron says capacitor spacing, rather than cell footprint alone, limits DRAM density. It argues that 4F² offers limited practical density gains while adding cost and manufacturing complexity.

Are Micron’s rivals making 4F² DRAM?

The report says Samsung, SK hynix and CXMT are pursuing 4F² designs. It also reports CXMT’s claim that it plans to have 4F² memory in server DDR5 RDIMMs by year-end, but does not independently confirm a product launch.

What is 3D DRAM?

3D DRAM would stack multiple layers of DRAM storage cells vertically. Micron identifies it as a longer-term direction, but the source describes substantial technical challenges and gives no commercial timeline.

Source: hothardware.com

You May Also Like

Enterprise Cloud Backup Solutions: A Back to school Guide

Discover how enterprise cloud backup solutions secure, scale, and streamline your data. Learn recent trends, key features, and expert tips for choosing the right provider.

Jack Dorsey Launches Buzz To Combine Team Chat, AI Agents And Git Hosting

Jack Dorsey announced Buzz, a new platform integrating team chat, AI agents, and Git hosting, aiming to unify development workflows.

M-DISC – DVD/Blu-ray Compatible Discs That May Last Up To 1000 Years

M-DISC offers DVD and Blu-ray compatible discs claimed to last up to 1000 years, sparking renewed interest in long-term data storage solutions.