Micron Discusses Business Updates

Article By : Gary Hilson

Micron Technology unveiled a new DRAM roadmap and pricing model at its Micron Investor Day 2022 event.

Micron Technology execs discussed the company’s momentum at last week’s day of investor updates, unveiling plans to ramp up manufacturing of its sixth–generation 232–layer flash by the end 2022, among other updates.

Micron has spent the last few years attempting to stake its claim as a leader in DRAM and NAND flash development. The company announced in late 2020 it was shipping its 176–layer NAND based on replacement gate (RG) technology with its CMOS–under–array architecture, as well as unveiled its 1–alpha node DRAM in early 2021, improving memory density by 40% when compared with its 1z node DRAM. Micron opted not to pursue the 3D Xpoint market, however, and is instead focusing efforts on compute express link (CXL) development.

NAND technology roadmap

The most significant news on the technology front out of the session was that Micron expects to ramp up manufacturing of sixth–generation 232–layer flash by the end of calendar 2022, according to Scott DeBoer, executive VP for technology and products.

“This node will deliver increased density, power, and bandwidth relative to our 176–layer node.” He said external and internal controllers continue be a strong element of Micron’s vertical integration focus.

Micron is also continuing to focus on QLC (4bits/cell) NAND while extending its CMOS–under–array and two–stack process architecture, which stacks two 3D NAND dies one above the other. Also known as “string stacking,” the process can address semiconductor manufacturing difficulties, such as etching connection holes through multiple layers. This is because as the hole depth deepens, the sides of the holes can become distorted and prevent the NAND cells from operating properly.

DRAM advancements

On the DRAM front, DeBoer said Micron will ramping its 1–beta node in late 2022 using its advanced CMOS technolgy, with 1–gamma expected to follow in 2024 with the introduction of extreme ultraviolet (EUV) lithography. The company is extending its planar DRAM roadmap while making continued investment in 3D DRAM research and development.

Micron’s DRAM roadmap will see the company ramp its 1–beta node in late 2022 using its advanced CMOS, with 1–gamma expected to follow in 2024. (Source: Micron) (Click image to enlarge)

Micron CEO Sanjay Mehrotra explained the last five years has seen the company make tremendous progress to establish competitive advantages with respect to technology, product, and manufacturing. This makes the company well positioned for what he called the “next era of leadership” as digital transformation is creating more opportunities for data creation, which has been accelerated by the pandemic as well as hyper–connectivity enabled by 5G across data centers, intelligent edge devices, and smart user devices.

Add to that the amount of data being generated by artificial intelligence (AI) and machine learning (ML), and the amount of data created is expected to double by 2025. “As AI algorithms get more sophisticated, they’re working on more data, and they then need more memory to process and draw insights,” Mehrotra said.

Meanwhile, he emphasized “the train of 5G” is driving greater usage of DRAM content in smartphones that are doing computational photography and advanced video editing thanks to AI algorithms. These AI algorithms are also increasing the demand for memory content in autonomous vehicles as they become data centers on wheels.

Demand for DRAM and NAND

Mehrotra explained what’s even more exciting is that the demand for memory and storage is across multiple large end markets. “This diversity of large end markets has a favorable impact on the stability of demand environment.”

No longer is demand anchored by the development of the PC; its is also driven by mobile and data centers, while the PC market has seen a resurgence with the sudden pivot to remote work. He said the data center is growing faster than the industry average due to the cloud and the adoption of AI to support new business models. This will drive growth for DRAM as well as NAND, which will also have high growth demand in industrial and automotive, but slightly below industry average for PC and mobile, with smartphone units having stabilized on a total global sales basis.

All this diversified opportunity for memory and storage means they are going to grow faster than the broader semiconductor industry, Mehrotra explained, while industry profitability continues to strengthen. Demand is driving healthier industry fundamentals, while the slowing for Moore’s Law is increasing industry supply discipline. There’s increased capital intensity required to implement a new node because there’s more expensive equipment involved and advanced tools are needed to implement increasingly complex features, as well as all the materials and structure aspects.

Looking ahead

Mehrotra said Micron’s strategy is to grow base supply in line with demand and maintain the discipline it’s shown in the last few years, while also making necessary capital investments and bearing in mind rising industry complexity.

Sumit Sadana, executive VP and chief business officer at Micron, also emphasized the company’s focus on driving its overall supply growth in line with market demand. “Our overall bit supply share will be flat over time, but we are focused on driving preferential share gains and important portions of the market.”

Addressing the data center market means understanding current architectures can’t meet the needs of tomorrow’s workloads. As a result, the market is becoming more heterogeneous with the bandwidth between processor and memory becoming even more critical, Sadana explained. By putting memory closer to the processor, more specialized processors and the CXL protocol can address the data movement bottlenecks – CXL will “kickstart a whole new era of connectivity with memory,” Sadana said.

Micron expects CXL to kickstart a whole new era of connectivity with memory as customers seeks more memory pooling and disaggregation and an evolution toward more heterogeneous architectures. (Source: Micron) (Click image to enlarge)

A recent notable move by Micron was to exit the 3D Xpoint market with an eye on developing solutions for CXL. No details on specific products were provided, but Sadana said CXL 1.1 will start to be deployed late this year for proof of concepts.

“The real deployment of memory on CXL will start late in 2024 when CXL 2.0 gets introduced and then will accelerate even further when CXL 3.0 standard enables memory pooling in a very real and practical way,” he said. “Memory pooling and memory disaggregation are critical aspects that our customers are looking for.”

Even more important is the expansion of the amount of memory that can be connected to a processor, which is ultimately what CXL enables in a very material way. This will lead to a higher amount of average DRAM content in a server.

How much that DRAM will cost may be subject to different pricing for Micron customers, as the company launched a new pricing model called forward pricing agreements (FPAs). Rather than make its long–term agreements based on volume and not on prices — as they are currently — FPAs would see prices based on both volume and pricing. The goal of the model is to reduce pricing fluctuations, even though profit is traditionally made through price volatility.

The new pricing model comes at a time when Micron faces some strong “inflationary headwinds,” including higher energy prices, as does the entire semiconductor industry as it makes capital investments in tools to meet demand and navigate current supply chain challenges. This includes AI and ML platforms to help predict operational challenges.

Manish Bhatia, Micron’s executive VP of global operations, said the collaboration between its technology development and manufacturing organizations as well as equipment vendors allow it to identify and address variations in process and equipment early in production ramping.

During a period of high capital investment, Bhatia explained it’s especially important to drive for higher productivity. “Micron’s commitment to smart manufacturing has really enabled us to be able to drive that higher productivity, whether that’s tool throughput improvements or increasing the tool up time through implementation of predictive maintenance.”

With equipment delays or lead times longer and more common, Bhatia said it’s even more important to drive productivity of equipment so Micron can keep production ramps for leading edge technologies on track and enable swift deliveries for its customers.

This article was originally published on EE Times.

Gary Hilson is a general contributing editor with a focus on memory and flash technologies for EE Times.


Subscribe to Newsletter

Leave a comment