麻豆原创

The global market for Computing in Memory Technology was valued at US$ 268 million in the year 2024 and is projected to reach a revised size of US$ 175260 million by 2031, growing at a CAGR of 154.7% during the forecast period.
As a new computing architecture, storage-computing integration is considered to be a revolutionary technology with potential and has received great attention at home and abroad. The core is to fully integrate storage and computing, effectively overcome the bottleneck of the von Neumann architecture, and combine advanced packaging and new storage devices in the post-Moore era to achieve an order of magnitude improvement in computing energy efficiency.
According to the distance between storage and computing, the technical solutions of generalized storage-computing integration are divided into three categories, namely, Processing Near Memory (PNM), Processing ln Memory (PlM) and Computing in Memory (CIM). In-memory computing is storage-computing integration in a narrow sense.
Global key players of Computing in Memory Technology include Syntiant, Zhicun(Witmem) Technology, Reexen Technology, Graphcore and Mythic, etc. The top five players hold a share over 80%. North America is the largest market, has a share about 50%. In terms of product type, In-memory Computing is the largest segment, occupied for a share of about 88%, and in terms of application, Small Computing Power has a share about 90 percent.
Analysis of the market drivers of Processing-in-Memory (PIM) technology,
1. Explosive growth in computing power demand: the underlying pressure of AI and big data
Demand for AI training and reasoning:
The global AI chip market is expected to reach US$120 billion in 2025, of which 75% of computing power is consumed in data transfer (not computing itself).
Large-scale language models (such as GPT-5) have more than 10 trillion parameters, and processing-in-memory (PIM) can improve the efficiency of sparse matrix operations by 3-5 times.
Data center energy consumption crisis:
Global data center power consumption accounts for 1.5% of total power demand, and data transfer energy consumption accounts for 40% in traditional architectures. Processing-in-Memory (PIM) can reduce energy consumption by more than 50% by reducing the memory wall effect.
2. Moore's Law slows down: an inevitable choice for architectural innovation
Process bottleneck:
The cost of advanced processes (below 3nm) has soared, and the marginal benefits of increasing transistor density have diminished. Processing-in-Memory integrates computing units through 3D stacking processes (such as HBM3) to break through the limitations of planar processes.
Heterogeneous computing needs:
Scenarios such as AI and graphics processing require customized computing units. Storage and computing integration supports the collaborative design of the logic layer and the storage layer to improve the efficiency of dedicated accelerators.
3. New storage technologies mature: hardware foundation is ready
Non-volatile memory (NVM) rises:
New memories such as ReRAM, MRAM, and PCM have analog computing capabilities and are naturally adapted to the storage and computing integration architecture. For example, the resistance state of ReRAM can directly participate in matrix operations.
Storage-class memory (SCM) popularization:
SCM technologies such as Intel Optane and Samsung Z-NAND have been mass-produced, providing PIM with high-performance, low-latency storage media.
4. Edge computing and IoT scenarios: energy efficiency revolution
The computing power dilemma of end-side devices:
Devices such as autonomous driving, AR/VR need to process massive amounts of data locally (such as 8K video streams). Storage and computing integration can reduce power consumption by 70% and extend battery life by 2-3 times.
Real-time requirements:
Predictive maintenance in industrial IoT needs to respond within microseconds, and storage and computing integration reduces data processing latency from milliseconds to nanoseconds.
5. Software ecology and algorithm collaboration: application scenario expansion
Sparse algorithm optimization:
Sparse matrices account for more than 95% of neural networks, and storage and computing integration can skip zero-value calculations, improving efficiency by more than 10 times.
Programming model evolution:
PIM-oriented spatial computing paradigms (such as NDA and GenASM) are gradually maturing, and developers can call computing units in storage.
6. Policy and capital promotion: global technology competition upgrades
National strategic support:
The US CHIPS Act and the EU's European Processor Initiative both list storage and computing integration as key directions. China's "14th Five-Year Plan" clearly supports the development of storage and computing integrated chips.
Capital inflow:
In 2023, global PIM financing will exceed US$5 billion, and giants such as Samsung, SK Hynix, and TSMC will accelerate their layout, and start-ups such as Mythic and UPMEM will receive multiple rounds of financing.
7. Supply chain reconstruction: from vertical integration to open collaboration
Industry chain collaboration:
Memory manufacturers (Micron, Kioxia) and IP suppliers (Synopsys, Cadence) cooperate to develop PIM design tool chains.
Foundries (SMIC, UMC) launched 2.5D/3D packaging technology to support mass production of integrated storage and computing chips.
Summary: The integrated storage and computing technology market is driven by computing power demand, hardware innovation, and policy capital. The core competition will focus on process integration capabilities (such as 3D stacking), algorithm-hardware co-design, and ecological openness. Chinese companies need to overcome the shortcomings of memory media and EDA tools and accelerate the commercialization of AI and edge scenarios.
This report aims to provide a comprehensive presentation of the global market for Computing in Memory Technology, with both quantitative and qualitative analysis, to help readers develop business/growth strategies, assess the market competitive situation, analyze their position in the current marketplace, and make informed business decisions regarding Computing in Memory Technology.
The Computing in Memory Technology market size, estimations, and forecasts are provided in terms of and revenue ($ millions), considering 2024 as the base year, with history and forecast data for the period from 2020 to 2031. This report segments the global Computing in Memory Technology market comprehensively. Regional market sizes, concerning products by Type, by Application, and by players, are also provided.
For a more in-depth understanding of the market, the report provides profiles of the competitive landscape, key competitors, and their respective market ranks. The report also discusses technological trends and new product developments.
The report will help the Computing in Memory Technology companies, new entrants, and industry chain related companies in this market with information on the revenues for the overall market and the sub-segments across the different segments, by company, by Type, by Application, and by regions.
麻豆原创 Segmentation
By Company
Syntiant
Zhicun(Witmem) Technology
Reexen Technology
Graphcore
Mythic
Shanyi Semiconductor
AistarTek
Samsung
SK Hynix
Houmo Technology
Pinxin Technology
Yizhu Intelligent Technology
TensorChip
Segment by Type
Near-Memory Computing
In-memory Computing
Processing In Memory
Segment by Application
Small Computing Power
Big Computing Power
By Region
North America
United States
Canada
Asia-Pacific
China
Japan
South Korea
Southeast Asia
India
Australia
Rest of Asia
Europe
Germany
France
U.K.
Italy
Russia
Nordic Countries
Rest of Europe
Latin America
Mexico
Brazil
Rest of Latin America
Middle East & Africa
Turkey
Saudi Arabia
UAE
Rest of MEA
Chapter Outline
Chapter 1: Introduces the report scope of the report, executive summary of different market segments (by Type, by Application, etc), including the market size of each market segment, future development potential, and so on. It offers a high-level view of the current state of the market and its likely evolution in the short to mid-term, and long term.
Chapter 2: Introduces executive summary of global market size, regional market size, this section also introduces the market dynamics, latest developments of the market, the driving factors and restrictive factors of the market, the challenges and risks faced by companies in the industry, and the analysis of relevant policies in the industry.
Chapter 3: Detailed analysis of Computing in Memory Technology company competitive landscape, revenue market share, latest development plan, merger, and acquisition information, etc.
Chapter 4: Provides the analysis of various market segments by Type, covering the market size and development potential of each market segment, to help readers find the blue ocean market in different market segments.
Chapter 5: Provides the analysis of various market segments by Application, covering the market size and development potential of each market segment, to help readers find the blue ocean market in different downstream markets.
Chapter 6, 7, 8, 9, 10: North America, Europe, Asia Pacific, Latin America, Middle East and Africa segment by country. It provides a quantitative analysis of the market size and development potential of each region and its main countries and introduces the market development, future development prospects, market space, and capacity of each country in the world.
Chapter 11: Provides profiles of key players, introducing the basic situation of the main companies in the market in detail, including product sales, revenue, price, gross margin, product introduction, recent development, etc.
Chapter 12: The main points and conclusions of the report.

Please Note - This is an on demand report and will be delivered in 2 business days (48 hours) post payment.