| 1. |
EXECUTIVE SUMMARY |
| 1.1. |
The Impact of Data Growth and Energy Consumption |
| 1.2. |
Modern Applications are Demanding High Performance Storage |
| 1.3. |
Memory bottlenecks for High Performance workloads |
| 1.4. |
Hierarchy of computing memory |
| 1.5. |
What are HDDs? How Do They Work? |
| 1.6. |
Hard Disk Drive Market in 2024 |
| 1.7. |
Forecast: Hard Disk Drive by Application |
| 1.8. |
SSDs Cell Types |
| 1.9. |
Key Manufactures of SSDs and Market Size |
| 1.10. |
Evolution of Capacity in QLC SSDs & HDDs |
| 1.11. |
Capacity Density Comparison of QLC SSDs & HDDs |
| 1.12. |
QLC SSDs & HDDs Best Metric Comparison Table |
| 1.13. |
Forecast: SSDs & NAND by Application |
| 1.14. |
What is DRAM? |
| 1.15. |
HBM |
| 1.16. |
Forecast: Yearly Unit Sales and Market Size of High Bandwidth Memory (HBM) |
| 1.17. |
Forecast: Memory by Application |
| 1.18. |
Emerging memory technologies: PCRAM, FRAM, RRAM, MRAM. |
| 1.19. |
Emerging Non-Volatile Memory Technology |
| 1.20. |
Companies involved within the Emerging Memory Space |
| 1.21. |
Emerging Memory is Taking the Embedded Route |
| 1.22. |
Emerging Memory Still Has a Path to Viability as Storage-Class Memory |
| 1.23. |
Forecast: Market size of Emerging Memory by Type |
| 1.24. |
IDTechEx Analysis: Future Outlook for MRAM |
| 1.25. |
IDTechEx Analysis: Future Outlook for ReRAM |
| 1.26. |
IDTechEx insight: Commercial FeRAM should adopt HfO₂ to move out of niche |
| 1.27. |
IDTechEx Outlook & Comments for FeRAM |
| 1.28. |
Failure of PCM as Storage Class Memory (2015-2023) |
| 1.29. |
IDTechEx Outlook & Comments for PCM |
| 1.30. |
Memory and Storage Technology Readiness Level |
| 1.31. |
Access More With an IDTechEx Subscription |
| 2. |
INTRODUCTION |
| 2.1.1. |
Understanding the Memory Hierarchy for General Computing |
| 2.1.2. |
Trends for storage and memory in Todays tech climate |
| 2.2. |
AI & HPC |
| 2.2.1. |
HPC – overview |
| 2.2.2. |
AI as a Leading Driver for Memory solutions |
| 2.2.3. |
Data movement through storage tiers in clusters for AI workloads |
| 2.2.4. |
Modern Applications are Demanding High Performance Storage |
| 2.2.5. |
Memory bottlenecks for High Performance workloads |
| 2.2.6. |
Overview of trends in HPC chip integration |
| 2.3. |
Cloud Storage |
| 2.3.1. |
The Impact of Data Growth and Energy Consumption |
| 2.3.2. |
Rising Data Storage Costs |
| 2.3.3. |
On-premises, cloud, and hybrid storage solutions: Shift towards cloud & hybrid |
| 2.4. |
Embedded Memory |
| 2.4.1. |
What are Embedded Memory and Embedded Systems? |
| 2.4.2. |
Types of Embedded Memory |
| 2.4.3. |
Embedded Flash Struggles with sub-28nm |
| 2.4.4. |
Scaling Embedded Memory to Advanced Nodes is Important for Key Metrics |
| 2.5. |
Edge Devices & IoT |
| 2.5.1. |
Edge Devices also is a Driver for Memory Solutions |
| 2.5.2. |
Edge vs Cloud characteristics |
| 2.5.3. |
Embedded Memory in Automotive Vehicles |
| 2.5.4. |
Edge AI in Smart Appliances |
| 3. |
FORECASTS |
| 3.1. |
Forecast Methodology |
| 3.2. |
Forecast: Market size of Hard-Drive-Disk by Application |
| 3.3. |
Forecast: SSDs & NAND by Application |
| 3.4. |
Forecast: Market size of Storage Cloud/Data Center Market |
| 3.5. |
Forecast: Market size of Storage Edge Market |
| 3.6. |
Forecast: Memory by Application |
| 3.7. |
Forecast: Yearly Unit Sales and Market Size of High Bandwidth Memory (HBM) |
| 3.8. |
Forecasts: Memory and Storage for Servers for AI/HPC |
| 3.9. |
Forecast: Memory & Storage Market by Application & Type |
| 3.10. |
Forecast: Top Level Forecast Memory & Storage Market |
| 3.11. |
Forecast: Market size of Emerging Memory by Type |
| 4. |
STORAGE |
| 4.1. |
Overview |
| 4.1.1. |
The Impact of Data Growth and Energy Consumption |
| 4.1.2. |
Rising Data Storage Costs |
| 4.1.3. |
Modern Applications are Demanding High Performance |
| 4.1.4. |
Trends for Storage in Todays Tech Climate |
| 4.1.5. |
Understanding the Memory Hierarchy for General Computing |
| 4.1.6. |
Storage in Datacentres |
| 4.1.7. |
Flash storage is the leading storage technology for HPC and AI applications |
| 4.1.8. |
HPC and AI require large-scale and high-performance data storage |
| 4.1.9. |
Storage requirements varies depending on the AI workloads |
| 4.1.10. |
Data movement through storage tiers in clusters for AI workloads |
| 4.1.11. |
Example of SSD configurations and solutions for AI and HPC workloads |
| 4.1.12. |
Examples of SK Hynix NAND Flash storage for AI and data centers |
| 4.1.13. |
Solidigm (SK Hynix subsidiary) offers SSDs previously manufactured by Intel |
| 4.1.14. |
Micron has a range of SSDs for applications in datacenters and AI |
| 4.1.15. |
Micron’s 9550 SSDs are designed for AI-critical workloads with PCIe Gen5 |
| 4.1.16. |
KIOXIA offers a range of datacenter and enterprise SSD solutions |
| 4.1.17. |
Storage in Edge Computing Devices |
| 4.2. |
Hard Drive Disks (HDDs) |
| 4.2.1. |
What are HDDs? How Do They Work? |
| 4.2.2. |
Advancements in HDD Technology |
| 4.2.3. |
Energy-Assisted Magnetic Recording (EAMR) Technologies |
| 4.2.4. |
Data Centre HDD match up |
| 4.2.5. |
Benefits and Drawbacks to HDDs relative to QLC SSDs |
| 4.2.6. |
Hard Disk Drive Market in 2024 |
| 4.2.7. |
HDDs Market Historically |
| 4.3. |
Solid State Drives (SSDs) |
| 4.3.1. |
What are SSDs? How Do They Work? |
| 4.3.2. |
NAND Flash memory uses floating gates or charge traps to store data |
| 4.3.3. |
Advancements in SSD Technology |
| 4.3.4. |
NAND Layer Stacking |
| 4.3.5. |
SK Hynix – NAND technology development |
| 4.3.6. |
SK Hynix: Overcoming stacking limitations to increase capacity using 4D2.0 |
| 4.3.7. |
KIOXIA uses BiCS 3D FLASHTM Technology to increase storage density |
| 4.3.8. |
SSDs Cell Types |
| 4.3.9. |
SLC SSDs |
| 4.3.10. |
SSDs for storage class memory bridging gap to volatile memory |
| 4.3.11. |
TLC SSDs |
| 4.3.12. |
QLC SSDs |
| 4.3.13. |
Increasing SSD capacity through emerging lower cost QLC NAND |
| 4.3.14. |
QLC affords higher capacity at a lower cost per bit but with performance deficits |
| 4.3.15. |
Benefits and Drawbacks to QLC SSDs |
| 4.3.16. |
Use Cases of HDDs & QLC SSDs |
| 4.3.17. |
Data center and enterprise SSD form factors transitioning towards EDSFF |
| 4.3.18. |
Step change in sequential read bandwidth with each generation of PCIe |
| 4.3.19. |
Evolution of PCIe generations in the SSD market |
| 4.3.20. |
Key Manufactures of SSDs and Market Size |
| 4.3.21. |
SSD Market Historically |
| 4.3.22. |
Storage Market |
| 4.4. |
Database Comparison – QLC SSD & HDD |
| 4.4.1. |
Why Compare QLC SSDs & HDDs |
| 4.4.2. |
Important KPI’s for Comparison |
| 4.4.3. |
Evolution of Capacity of QLC SSDs & HDDs |
| 4.4.4. |
Capacity Density Comparison of QLC SSDs & HDDs |
| 4.4.5. |
Sequential Bandwidth of QLC SSDs & HDDs |
| 4.4.6. |
Capacity-to-Power Ratio of QLC SSDs & HDDs |
| 4.4.7. |
Capacity Density / Power of QLC SSDs & HDDs |
| 4.4.8. |
Capacity Density / Power of QLC SSDs & HDDs |
| 4.4.9. |
QLC SSDs & HDDs Best Metric Comparison Table |
| 4.5. |
Going Forward – Improving Current Technologies |
| 4.5.1. |
SK Hynix Unveils Penta-Level 3D NAND Flash Memory in 2024 |
| 4.5.2. |
Macronix Introduced Compute-In-Memory 3D NOR Flash technology for AI Applications in 2024 |
| 4.5.3. |
SK Hynix introduced Accelerator-in-Memory for LLM Inference |
| 5. |
MEMORY |
| 5.1. |
Overview |
| 5.1.1. |
Hierarchy of computing memory |
| 5.1.2. |
Memory bottlenecks for HPC/AI workloads and processor under-utilization |
| 5.1.3. |
What is DRAM? |
| 5.1.4. |
What is SRAM? |
| 5.1.5. |
Types of DRAM and Comparison of HBM with DDR |
| 5.1.6. |
HBM vs DDR for computing – market trend |
| 5.2. |
DDR Memory |
| 5.2.1. |
Developments in double data rate (DDR) memory |
| 5.2.2. |
DDR5 memory in AMD’s 4th Gen EPYC processors for HPC workloads |
| 5.2.3. |
DDR5 MRDIMM increases capacity and bandwidth for high CPU core counts |
| 5.2.4. |
NVIDIA’s Grace CPU uses LPDDR5X memory to lower power consumption |
| 5.2.5. |
GDDR7 announced by major players targeting HPC and AI applications |
| 5.2.6. |
Comparison of GDDR6 and GDDR7 modules |
| 5.3. |
High Bandwidth Memory (HBM) |
| 5.3.1. |
HBM |
| 5.3.2. |
High bandwidth memory (HBM) and comparison with other DRAM technologies |
| 5.3.3. |
Demand outgrows supply for HBM in 2024 |
| 5.3.4. |
HBM (High Bandwidth Memory) packaging |
| 5.3.5. |
HBM packaging transition to hybrid bonding |
| 5.3.6. |
Benchmark of HBM performance utilizing µ bump and hybrid bonding |
| 5.3.7. |
SK Hynix has started volume production of 12-layer HBM3E |
| 5.3.8. |
Micron released 24GB HBM3E for NVIDIA H200 and is sampling 36GB HBM3E |
| 5.3.9. |
Samsung expects production of HBM3E 36GB within 2024 |
| 5.3.10. |
Overview of current HBM stacking technologies by 3 main players |
| 5.3.11. |
Evolution of HBM generations and transition to HBM4 |
| 5.3.12. |
Benchmarking of HBM technologies in the market from key players (1) |
| 5.3.13. |
Benchmarking of HBM technologies in the market from key players (2) |
| 5.3.14. |
Examples of CPUs and accelerators using HBM |
| 5.3.15. |
Intel’s CPU Max series for HPC workloads has HBM and optional DDR |
| 5.3.16. |
AMD CDNA 3 APU architecture with unified HBM memory for HPC |
| 5.3.17. |
Three main approaches to package HBM and GPU |
| 5.3.18. |
Drawbacks of High Bandwidth Memory (HBM) |
| 5.4. |
Memory Expansion |
| 5.4.1. |
Samsung’s CMM-D memory expansion for AI and datacenter server applications |
| 5.4.2. |
Micron’s CXL memory expansion modules for storage tiering in datacenters |
| 5.4.3. |
Memory Market |
| 5.4.4. |
DDR memory dominates CPUs whereas HBM is key to GPU performance |
| 5.5. |
Memory Market |
| 5.5.1. |
DDR memory dominates CPUs whereas HBM is key to GPU performance |
| 5.5.2. |
Memory market |
| 6. |
EMERGING STORAGE & MEMORY |
| 6.1. |
Overview |
| 6.1.1. |
Memory bottlenecks for HPC/AI workloads and processor under-utilization |
| 6.1.2. |
Embedded Flash Struggles with sub-28nm |
| 6.1.3. |
Scaling Embedded Memory to Advanced Nodes is Important for Key Metrics |
| 6.1.4. |
Emerging memory technologies: PCRAM, FRAM, RRAM, MRAM. |
| 6.2. |
Magnetoresistive RAM (MRAM) |
| 6.2.1. |
What is Magnetoresistive RAM (MRAM)? How Does it Work? |
| 6.2.2. |
Types of (MRAM) |
| 6.2.3. |
Benefits and Drawbacks to MRAM |
| 6.2.4. |
Current Applications of MRAM |
| 6.2.5. |
MRAM Specific Companies & Startups |
| 6.2.6. |
MRAM Specific Companies & Startups |
| 6.2.7. |
Everspin Technologies is the leading supplier of discrete MRAM components |
| 6.2.8. |
Everspin xSPI STT-MRAM sets new benchmark MRAM |
| 6.2.9. |
Everspin Expands MRAM Portfolio for Edge AI and Embedded Systems |
| 6.2.10. |
Everspin Target Markets is Growing With New applications |
| 6.2.11. |
Avalanche Technology’s MRAM Adoption in Aerospace Applications |
| 6.2.12. |
TSMC’s Involvement in MRAM |
| 6.2.13. |
TSMC and NXP MRAM in Automotive Industry |
| 6.2.14. |
TSMC: STT-MRAM Co-Optimized for AI Edge Devices |
| 6.2.15. |
Samsung’s Role in MRAM Research and Development |
| 6.2.16. |
Samsung Unveils World Most Write Energy 14nm eMRAM Technology for Automotive Applications |
| 6.2.17. |
Samsung Reveals Smallest-Cell eMRAM Compatible With 8nm Logic Node for Automotive Applications |
| 6.2.18. |
Netsol Uses Samsung Foundry 28nm Process to produce MRAM Products |
| 6.2.19. |
Kioxia Introduces World Smallest 1Selector-1MTJ Cell for 64 Gb Cross-Point MRAM |
| 6.2.20. |
MRAM Market: Segmentation by Company Type |
| 6.2.21. |
IDTechEx Analysis: Future Outlook for MRAM |
| 6.3. |
Resistive RAM (ReRAM) |
| 6.3.1. |
What is Resistive Ram (ReRAM)? How Does it Work? |
| 6.3.2. |
Benefits and Drawbacks to ReRAM |
| 6.3.3. |
Current Applications of ReRAM |
| 6.3.4. |
ReRAM Market: Segmentation by Company Type |
| 6.3.5. |
ReRAM Market Historically |
| 6.3.6. |
ReRAM Specific Companies & Startups |
| 6.3.7. |
ReRAM Specific Companies & Startups |
| 6.3.8. |
Weebit Nano Developing and Licensing ReRAM Technology |
| 6.3.9. |
Weebit Nano’s Roadmap for ReRAM in AI Applications |
| 6.3.10. |
CrossBar Inc Licensing ReRAM Technology |
| 6.3.11. |
CrossBar Inc Provides High Performance Embedded and 3D High Density ReRAM |
| 6.3.12. |
4DS Memory Develops Area Based Interface Switching ReRAM |
| 6.3.13. |
RAMXEED ReRAM Technology and Development |
| 6.3.14. |
GlobalFoundries Demonstrates ReRAM in its 22FDX Platform |
| 6.3.15. |
TSMC integrates ReRAM into its nRF54L Series SoCs AT 22nm |
| 6.3.16. |
IDTechEx Analysis: Future Outlook for ReRAM |
| 6.4. |
Ferroelectric RAM (FeRAM) |
| 6.4.1. |
What is Ferroelectric RAM (FeRAM)? How Does it Work? |
| 6.4.2. |
Benefits and Drawbacks to FeRAM |
| 6.4.3. |
Current Applications of FeRAM |
| 6.4.4. |
FeRAM Market: Segmentation by Company Type |
| 6.4.5. |
RAMXEED FeRAM Technology and Development |
| 6.4.6. |
Infineon is a leading supplier of FeRAM |
| 6.4.7. |
Micron FeRAM Achieves Industry-Leading Density |
| 6.4.8. |
Ferroelectric Memory Company Targets HfO2 FeRAM Commercialization |
| 6.4.9. |
SK Hynix Unveils Ultra-High-Density 3D FeNAND Arrays for Analog Computation of Hyperscale AI Models |
| 6.4.10. |
TSMC Showcases Ferroelectric FET Memory with Smallest Cell Area and High Endurance |
| 6.4.11. |
IDTechEx Insight – Commercial FeRAM Needs HfO₂ to Stay Competitive |
| 6.4.12. |
IDTechEx Outlook & Comments for FeRAM |
| 6.5. |
Phase Change Memory (PCM) |
| 6.5.1. |
What is Phase Change Memory (PCM/PCRAM) How Does it Work? |
| 6.5.2. |
Benefits and Drawbacks to PCM |
| 6.5.3. |
PCM Market: Segmentation by Company Type |
| 6.5.4. |
PCM Market Lessons from Intel Optane Failure |
| 6.5.5. |
Micron 3DXPoint |
| 6.5.6. |
Failure of PCM as Storage Class Memory (2015-2023) |
| 6.5.7. |
STMicroelectronics produces ePCM for Microcontrollers in automotive controllers |
| 6.5.8. |
STMicroelectronics presents Single-Ended ePCM Memory Array for Neural Network Weight Storage in Edge-AI Applications |
| 6.5.9. |
PCM Market |
| 6.5.10. |
IDTechEx Outlook & Comments for PCM |
| 6.6. |
Comparison of Emerging Memory Platforms |
| 6.6.1. |
Emerging Memory Still Has a Path to Viability as Storage-Class Memory |
| 6.6.2. |
Emerging Memory is Taking the Embedded Route |
| 6.6.3. |
Comparison of Emerging Technology |
| 6.6.4. |
IDTechEx Comparison of Commercialized Emerging Tech Products |
| 6.6.5. |
Companies involved within the Emerging Memory Space |
| 7. |
COMPANY PROFILES |
| 7.1. |
Company Profiles Included with this Report |
