DRAM Archives - Counterpoint

Mobile Demand Raises DRAM Revenue by 30% YoY to $19 bn in Q1 2021

London, Hong Kong, Boston, Toronto, New Delhi, Beijing, Taipei, Seoul – Jun 25, 2021

Global DRAM revenues rose to $19 billion in Q1 2021, increasing by a solid 30% YoY and 9% QoQ. Distance education and work from home (WFH) continued propelling a substantial demand for smartphone and laptop DRAM, resulting in a 6% growth in bit shipment and a 3% rise in ASP over the previous quarter.

Associate Director Brady Wang said the Chinese handset makers stepped up their smartphone shipments in the first quarter, intending to gain shares from the beleaguered Huawei. In addition, the high-density LPDDR4x memory prices softened in H2 2020, increasing smartphone DRAM content and making 6GB the minimum standard for mid- to high-end smartphones since 2021. As a result, the average DRAM capacity in smartphones clocked at 5.3GB in Q1 2021, rising impressively by around 21% YoY and 7% QoQ. Similarly, servers experienced recovery in demand, and the adoption of a new data-center CPU bumped up server content per box. Therefore, the server segment’s DRAM demand rose as well.

According to Wang, DRAM is already an oligopolistic market with an Herfindahl-Hirschman Index (HHI) of 3,138. A significant capacity expansion by any player will soon turn around market status and reduce the overall profitability. Therefore, this year, all three major players will spend most of their resources in migrating to advanced nodes, a process that is bound to reduce production capacity. In addition, transportation and component shortage concerns will force device vendors to place orders earlier than usual. When coupled with the growing demand for personal computers, games and servers, these factors signal the possibility of the DRAM market turning to a shortage this year. The smartphone market is recovering at present, but its cost sensitivity means that a DRAM price spike may put the brakes on smartphone DRAM content growth.

Competitive Landscape

The DRAM industry is dominated by three major players that collectively account for about 95% of the market’s bit shipments and revenue.

Samsung Electronics

With $7.9 billion in revenue, Samsung led the DRAM market in Q1 2021. The South Korean semiconductor giant continued commanding over two-fifths of the DRAM market revenue, overshadowing its nearest competitor by over 41%. Samsung’s DRAM bit shipment growth came from actively responding to the (i) 5G-related surge in smartphone demand, (ii) rising server demands for data centers and (iii) rise of home entertainment culture that increased memory content in TVs and STBs to support 4K UHD content and streaming.

Research Associate Siddharth Bhatla sees 1Z nm or 15 nm as Samsung’s most advanced mass-produced DRAM node from the technology perspective. The company’s plans include beginning mass production of 14 nm node in H2 2021. Samsung aims to differentiate its DRAM offerings using multi-layer EUV on its 14 nm node, building upon the single-layer EUV in its current 15 nm node.

SK hynix

Ranking second, SK hynix accounted for over 29% of the DRAM industry’s Q1 2021 revenues. Surpassing Micron by over 25%, its revenues rose by over 28% YoY in that period, in line with the industry’s overall growth. In addition, the company’s bit shipments rose 4% QoQ, thanks to its ability to actively cater to the surging demand for mobile and PC memory. By the year-end, SK hynix aims to (i) ramp up the production of its 1Z nm DRAM and (ii) complete development and begin mass production of its 1α-EUV node.

Micron Technology

Micron achieved a 44% YoY jump in its DRAM revenue in Q1 2021, continuing to grow fastest among the big three since the preceding two quarters. Micron was the first among the big three to begin mass production of 1α DRAM, accounting for one-fourth of the industry’s DRAM revenue. However, Micron’s 1α node is based on DUV, an older technology that may face severe cost competition once the EUV-based 1α DRAMs reach the mass market.

Feel free to contact us at press(at)counterpointresearch.com for questions regarding our in-depth research and insights, or for press enquiries.

Background:

Counterpoint Technology Market Research is a global research firm specializing in products in the TMT (technology, media and telecom) industry. It services major technology and financial firms with a mix of monthly reports, customized projects and detailed analyses of the mobile and technology markets. Its key analysts are seasoned experts in the high-tech industry.

Brady Wang

Siddharth Bhatla

Counterpoint Research
press(at)counterpointresearch.com

Related Reports

Smartphones Beat DRAM Drum to Meet Performance Demand

Smartphones are getting more performant with each passing day, thanks to the confluence of user demands and technological advancements. One of their key components is DRAM, a high-speed-low-latency memory module that serves as a temporary memory for an application processor (AP). Commonly known as RAM, DRAM stores the OS and running applications’ working data. A larger DRAM enhances a phone’s capability to host more applications simultaneously, enabling users to switch between apps seamlessly without needing to reload them back from the flash.

Like every other smartphone component, DRAM has improved over time in several aspects: capacity, speed and affordability. On average, smartphone DRAM capacity has risen consistently over time. Several reasons can explain this phenomenon: (i) Intense megapixel race in cameras, (ii) Surging performance needs of big apps and games, and (iii) Increasing prevalence of multi-tasking and high-resolution-high-framerate displays. Users now expect razor-sharp photography, high framerate gameplay, and the capability to drive large applications parallelly without any lag.

Currently, the DRAM sizes embedded in smartphones vary greatly, starting from 2GB for the low-cost segment to 18GB for the flagship models. Every two in five smartphones sold in 2020 had 4-6GB DRAM. Therefore, most phones of the day support moderate gameplay along with some degree of multi-tasking. Even though it may seem so, more DRAM is not always merrier. Bigger DRAM consumes more power irrespective of whether it is fully or partially used. So, smaller DRAM is favourable for budget phones from the battery and cost perspective.

Android needs more DRAM than iOS

According to our Mobile Handset Sell-through Tracker, the global average DRAM capacity almost reached the 5GB mark in Q4 2020, touching 4.5GB and 5GB for iOS and Android smartphones respectively, and growing 22% YoY in 2020 overall. The Android smartphones’ average DRAM sustained stable growth in 2019-2020, while the iOS segment had it relatively uneven. For the latter, high growth phases primarily centre around the new iPhone releases, scheduled in the fourth quarter every year.

The Android segment’s growth faced a slump during Q2 2020-Q4 2020 due to the bearish market following COVID-19. For the iOS camp, the release of iPhone 11/Pro/Max and iPhone 12 Pro/Max resulted in a steep rise in Apple’s memory numbers, owing to their 4GB and 6GB DRAMs respectively.

Overall, the DRAM content for iPhones was consistently dwarfed by the other brand families in this list. This disparity stems from the far more efficient memory management approach that iPhones employ. The iOS uses reference counting, a mechanism that needs much lower working memory than Android’s garbage collection approach. Also, iPhone’s higher bandwidth NAND flash reduces the time required to reload data from storage. Therefore, iPhones can typically rely more on flash storage to cope with DRAM overflow compared to their Android counterparts. As per our Component Price Tracker, a significant drop in DRAM prices facilitated the smartphone OEMs’ transition to larger memory units in 2019-2020. The LPDDR4x DRAM prices tumbled by over 20% in the H2 2019-H2 2020 period. By the end of this period, 6GB emerged as the DRAM size variant with the minimum cost per GB.

DRAM Size Ranking by Brand Family

According to our Smartphone DRAM Status Update for Q4 2020, bigger average DRAM numbers signal a greater proportion of high-end phones in a brand’s portfolio, especially in the Android segment. This section compares the growth in DRAM capacities of the top six best-selling smartphone brand families during Q1 2019-Q4 2020.

Apple recorded the highest DRAM growth numbers, followed by Huawei and Samsung. The DRAM content for the bottom three brands grew half as fast as the ones on the top. Early adoption of larger DRAM units justified the lower growth for smartphones under the OPPO and vivo umbrella. The same was the case with Xiaomi’s cash-cow sub-brand Redmi that focuses on the budget segment.

Over half of Huawei’s top three best-selling phones – P30, P30 Lite and P30 Pro – were sold in 8GB variants, while the 6GB variants accounted for another one-third. Among Apple’s top three best-selling model families – iPhone 11, iPhone XR and iPhone 11 Pro Max, the 4GB RAM variants made up for 70% of sales while the 3GB ones took the remaining 30%. Other notable OEMs in this aspect were Black Shark, OnePlus and Razer with their 8.5GB, 8.4GB and 8GB average RAM densities respectively.

Our quarterly report on smartphone DRAM presents a more profound analysis in this direction, featuring (i) quarter-wise DRAM numbers and personalised insights for major smartphone OEMs, (ii) change in market share of smartphone DRAM densities and (iii) correlation heatmap for the primary camera and DRAM capacity.

Conclusion

Growing affordability of LPDDR4x memory, rising user expectations and ballooning memory footprints of modern apps have played a crucial role in driving up smartphone DRAM sizes over time. The increasing prevalence of multi-tasking can certainly inflate the users’ memory needs.

Users invest in large DRAMs, hoping to future-proof their phones. But little are they aware of the raised power consumption of these large memory units — as smartphones in the present need just 4GB DRAM for optimal performance. On the other hand, the gaming-oriented ones need a minimum of 8GB DRAM for smooth multi-tasking alongside heavy gaming. The smartphone DRAM of the budget segment should see good growth in the near term, and it will be interesting to see how far up the premium segment pushes the DRAM capacities.

Counterpoint Recommended Reading

Expected Memory Market Rebound in 2020 and Continuous Investment Can Strengthen Samsung's Leading Position in Memory

Samsung Electronics announced that it achieved memory revenue of 13.26 trillion won (USD 11.48 billion) in Q3 2019, up 7.8% sequentially, but down 37% year on year, mainly due to the weak memory market.

Unlike other memory companies that are planning on reducing memory CAPEX in 2019 and 2020, Samsung’s 2019 Capex will reach 23.3 trillion won (~USD 20 billion), which is at a similar level to that of 2018. The accumulated investment in the first three quarters was 16.8 trillion won and will invest the other 12.2 trillion won in the fourth quarter. Most of the remaining CAPEX in 4Q will be used to build memory infrastructure, including the second phase of Samsung’s fab in Xi’an’s and Pyeongtaek. Both are expected to be in mass-production in 2020 to meet the mid- to long-term demand. It is in line with Samsung’s long-term strategy to increase investment as the market cycle reaches the bottom.

Exhibit one is Counterpoint’s estimate of revenue for both DRAM and NAND. Exhibit 2 shows our estimate for bit shipment and ASPs of both.

Exhibit 1: Historical revenue of both DRAM and NAND flash, 1Q18-3Q19

Exhibit 2: Historical bit shipment and ASPs of both DRAM and NAND flash, 1Q18-3Q19

Samsung will continue to drive the technology conversion of both DRAM and NAND to reduce the production costs and widen the gap to its other competitors.

The percentage of Samsung’s 1Xnm DRAM capacity will be close to 80% by the end of 2019; higher than other players. Also, the next 1y-nm DRAM will become mainstream in the first half of 2020.

From the perspective of NAND flash, Samsung is also converting 96-layer fifth-generation V-NAND technology to the 136-story sixth-generation V-NAND, which can increase the bit density and reduce the cost of a single-chip.

Samsung expects bit consumption of both DRAM and NAND in smartphones will both increase because 5G smartphones will increase the demand for high-density and low power DRAM.

The recovery of the data center and the launch of new CPUs from Intel will both drive high-density, and high-performance SSDs. At the same time, the adoption rate of SSDs in PCs will also increase steadily.

The inventory of both DRAM and NAND continues to drop. However, DRAM inventory may increase again as Chinese manufacturers start to build inventory to avoid being impacted by trade wars.

Although the demand for both DRAM and NAND is gradually increasing, Counterpoint expects that the ASPs of both will still see a mild decline in 4Q 2019. However, prices are expected to rebound from the 1Q 2020 because of pre-orders for Chinese New Year and the release of new 5G smartphones. The inhibitors to higher prices, however, are the pre-established chip inventory by Chinese companies.

Counterpoint suggests memory vendors carefully check whether smartphone OEMs have placed double orders. Also, they should understand the operating conditions of OEMs via Counterpoint’s tracking services to mitigate the risk of oversupply and excessive inventory.

SEMICON West 2019: New Non-Volatile Memory Technologies MRAM, PCM and RRAM to Revolutionize Future AI Workloads

Counterpoint Analysts were invited by Applied Materials, one of the leading semiconductor manufacturing equipment vendors, to participate in the AI Design Forum at SEMICON West in California, USA recently. The event was hosted by SEMI Americas, the Electronic System Design Alliance and Applied Materials. The keynote speakers included key executives in the semiconductor industry from Applied Materials President and CEO Gary Dickerson to other CEOs from AMD, Synopsys and Xilinx, along with technology presentations from arm, Google, Qualcomm, and the Embedded Vision Alliance.

We interviewed a number of key stakeholders driving the AI in the cloud and at the edge and understand the challenges the semiconductor industry is about to face as Moore’s Law, the rule of progress in semiconductor design, is no longer valid. As a result, traditional chip design and the associated manufacturing processes cannot fulfill the fast-developing demand for Artificial Intelligence (AI) and will have to change.

During the meeting, Victor Peng, the President & CEO of Xilinx, said the processor performance is saturating and is facing three major problems:

Power density increases
End of Power, Performance & Area (PPA) improvements
Multicore hits a limit

Given these facts, semiconductor technology must adapt to meet the growing needs of emergent technology areas such as AI & IoT.

In addition, Gary Dickerson, chief executive of Applied Materials, discussed the data explosion that AI and the Internet of Things will cause, thanks to the expected hundreds of billions of connected devices by 2030.

The amount of data generated by a person is currently about 1GB per day. The amount of data generated by an autonomous vehicle in 2023 will reach 4000GB per day with average annual growth rates of about 5% and 70% between 2018 and 2023, respectively.

Also important, but often overlooked, is the energy consumption needed to train AI models. This is expected to increase exponentially. It is estimated that training a single AI model can emit as much as carbon as five cars in their lifetime. Energy consumption will likely become a key constraint to the growth of AI unless more energy-efficient computing paradigms are developed.

And while the power consumption of most IoT edge devices is low, the total energy consumption is largely due to the massive number of them. For example, each IP camera consumes only 5 to 8 watts, but in 2020 all IP cameras combined will consume more power than the energy generated by a standard power plant in the United States.

Dickerson also mentioned AI will take 80 percent of the workforce in future datacenters for neural nets and training, and also consume more than 10% of global electricity supply in 2025. Therefore, in order to reduce the excessive power consumption and the surge in data volume, future semiconductor design and manufacturing must consider four key aspects: Performance, Power, Area-Cost (PPAC). And start with five new technologies:

New architectures
New structures/3D
New materials
New ways to shrink
Advanced packaging.

In addition, semiconductor innovation is needed as you can see in the examples below in the cloud and edge.

Exhibit 1: Promising pipeline of new hardware innovation

Source: AI Design Forum at SEMICON West 2019

As pointed out by the exhibit above, the new role of memory in hardware innovation is getting important. Today’s mainstream memory consumes too much power, so a “new architecture” for chip design and new memory chips are needed. He also listed several types of memory, including MRAM, PCM, and ReRAM and demonstrated Applied Material’s capability in manufacturing equipment for memory.

Counterpoint has long-term research for emerging memory. In the past 20 years, several next-generation memories have debuted, but none can challenge the mainstream DRAM and NAND. The power consumption of volatile DRAMs is much higher than non-volatile memories. Although the production cost of NAND flash is low, its speed and delay are poor. However, as DRAM approaches its physical limits, production and R&D costs have increased significantly relative to the past. The technology lifetime of NAND flash will be extended with 3D structures. However, its shortcomings have not been addressed during the migration to 3D. RRAM, MRAM, and PCM are currently considered to be promising replacements for DRAM and NAND, but they still need to prove their capabilities in full-scale production, as well as further reduce production costs.

Exhibit 2 illustrates the current memory classification, including mass production memory, niche memory, and emerging memory.

Exhibit 2: Memory Technology Classification, 2019

Source: Counterpoint Research – Memory Research

PCM: One of the most promising PCMs is 3D XPoint which is promoted by Intel and Micron. So far, the manufacturing cost of 3D XPoint is lower than that of server DRAM, but it is still higher than NAND flash and commodity DRAM. 3D XPoint is superior to NAND flash in terms of read/write speed, latency, input/output operations per second (IOPS) and endurance. Compared to DRAM, 3D XPoint is still slower but, it has lower manufacturing costs and power consumption. As a result, 3D XPoint can be used as a non-volatile dual in-line memory module (NVDIMM), solid state drive (SSD) storage and cache in servers.

MRAM is a memory that stores binary data in an electronic spin state. MRAM is faster than DRAM and requires neither management nor refresh. MRAM is one of the most promising non-volatile emerging memory types and is now in mass production. However, it is still expensive, so its application is limited to industrial and server uses. Currently, the maximum density of a single-chip STT-MRAM is 1 Gb. Once the price falls further, it has an opportunity to be used as a cache in consumer SSDs.

The density of ReRAM could be comparable with 3D NAND because 3D structures are possible. However, it still has reliability issues and is expected to see more adoption after the production cost of 3D NAND increases significantly.

In conclusion, one of the biggest challenges to implement AI today is to overcome the memory speed and power bottlenecks in the current architecture to achieve faster data access while reducing energy costs. Adopting the new non-volatile memory, MRAM, PCM and RRAM, can make it possible to achieve large-scale energy savings and performance improvements, thus greatly extending battery life and better user experience.

SK Hynix to Cut DRAM Production and Reduce Wafer Start of NAND Flash

SK Hynix will begin to cut DRAM production from Q4 2019 as well as reduce its NAND flash wafer start by 15% as compared to its 2018 output. Last quarter, the company had said it would reduce NAND flash wafer starts by 10%.

The fresh set of announcements come after the company reported an 89% year-on-year (YoY) decline in operating profit in Q2 2019. This was despite a 13% quarter-on-quarter (QoQ) increase in DRAM bit shipments, mainly due to the growing demand from the mobile and PC DRAM markets. However, DRAM prices are still weak. In fact, average selling prices (ASPs) have dropped by 24%. It was a similar story in the NAND flash market where the company’s shipments increased 40%, but ASPs were down 25%.

Exhibit 1: SK Hynix DRAM bit shipment and Blended ASP from 1Q18 to 2Q19

Source: Counterpoint Research – Memory Tracker and Forecast

Usually, both DRAM and NAND flash have the characteristics of a boom-bust cycle. As a result, these two products normally do not reach the bottom at the same time. Therefore, the production capacity is interchangeable to can make up for each other’s losses. However, this time, the situation is very different. Both DRAM and NAND flash are at the bottom of their bust cycle, which is hurting Samsung, SK Hynix, and Micron. It is still unclear how long the threat of the trade war between Japan and South Korea will last. In order to reduce the risk of raw material shortage as well as limited resources for investment, SK Hynix must choose between a cut in spending on DRAM or NAND flash.

Currently, spot prices of both DRAM and NAND flash have rebounded. However, contract pricing has not seen a significant rebound because of high inventory in the channels. Therefore, SK Hynix’s DRAM investment cuts can increase its bargaining power. However, this takes time and has risks since other competitors take this chance to grasp more share from SK Hynix. The two main applications of DRAM are smartphones and servers. The growth of both smartphones shipment and DRAM content will be limited while the server market is still in a downturn. Therefore, we expect that DRAM prices will rise marginally in the short-term but will continue to be weak afterward.

The situation for NAND flash is different. The demand for NAND flash in smartphones is still growing. The average storage size in smartphones in 2020 is expected to reach 72GB, up from 64GB in 2019. The storage size of the flagship smartphone will be at least 128GB in the H2 2019. In addition, the demand for NAND flash in SSDs is also growing. Both PCs and servers require larger and faster NAND flash. The mainstream capacity of SSD in PC is moving to 512GB from 256GB. The average capacity has also risen to 321GB. As the unit price of NAND flash declines, the penetration rate of SSD in both PC and server will also increase significantly. Therefore, we expect to see the penetration of SSDs in PCs continuing to rise and cross the 50% mark this year. The NAND flash market is expected to recover in Q4. Demand will increase as prices decline. As a result, it can help companies like SK Hynix increase the bit growth and improve the utilization rate of their production lines.

In conclusion, it makes sense for SK Hynix to cut DRAM investment. In the case of limited resources in raw material and CAPEX, the bit growth of DRAM will be lower than that of NAND flash. SK Hynix should immediately stop producing 36 and 48 layer products as they are not economically viable. Instead, the company can do better by focussing on improving the yields, increase the capacity of NAND flash, its 96 layer products, and even 128 layer NAND flash. This can increase utilization and effectively reduce costs.

Top

Insights Tag: DRAM