
MemryX
Designing neural network inference chips with unparalleled energy efficiency and compute density for edge computing, IoT, mobile, and more.
Date | Investors | Amount | Round |
---|---|---|---|
- | investor investor | €0.0 | round |
N/A | €0.0 | round | |
investor investor investor | €0.0 | round | |
* | $44.0m | Series B | |
Total Funding | 000k |
Related Content
MemryX is a startup that operates in the artificial intelligence (AI) industry, specifically focusing on Edge AI processing. The company has developed a unique solution that simplifies the integration of AI processing, making it as straightforward as adding Dynamic Random Access Memory (DRAM) to a system. This solution allows for the easy compilation and optimization of trained AI models without any loss in accuracy, eliminating the need for manual tuning or model retraining.
The company's technology is scalable, allowing for the linear addition of AI processing chips to increase performance or reduce latency. Each chip, or 'chiplet', can process multiple models or large models across multiple chips using the same software. MemryX's proprietary, configurable native dataflow architecture and at-memory computing set a new standard for Edge AI processing. This system architecture effectively removes the data movement bottleneck and is future-proof, supporting new hardware, processes, chemistries, and AI models with the same software.
MemryX's business model revolves around the sale of its AI accelerator chip, the MX3. This chip was recently demonstrated operating on the Lenovo ThinkCentre neo Ultra desktop PC at CES 2024, marking a significant milestone for the company. The MX3 is the first third-party AI accelerator card to be integrated into a PC, indicating a potential new market for standalone AI accelerator cards for PCs.
In summary, MemryX is a promising startup in the AI industry, offering a unique and scalable solution for Edge AI processing. Its business model is based on the sale of its AI accelerator chips, which have already demonstrated their potential in the PC market.
Keywords: Edge AI, AI processing, DRAM, AI models, scalable, chiplet, dataflow architecture, at-memory computing, AI accelerator chip, PC integration.