Edge AI startup Semron raises funds for memcapacitor processing
Cette publication existe aussi en Français
Semron GmbH (Dresden, Germany) has raised €7.3 million (about US$7.9 million) for edge AI processing based on memory-capacitive devices, according to reports.
The funding round was led by Join Capital and supported by SquareOne, OTB Ventures, and Onsight Ventures.
The company is claiming it use of an unusual device called a memcapacitor – analogous to a memristor – could allow a dramatic increase in processing density along with a reduction in power consumption for basic math operations at the core of AI.
The technology was researched by the founders CEO Aron Kirschen (left) and CTO Kai Uwe Demasius during their time at Technical University of Dresden and Max Planck Institute of Technology, respectively (See Energy-efficient memcapacitor devices for neuromorphic computing published by Nature Electronics in October 2021). A co-author on the paper was Stuart Parkin who is managing director of the Max Planck Institute of Microstructure Physics and was a 37-year veteran at IBM and an IBM Fellow.
Semron’s memcapacitors are made of conventional semiconductor materials laid down on silicon-on-insulator (SOI) wafers. They work by exploiting charge shielding. The memcapacitors control an electric field between a top electrode and bottom electrode via a “shielding layer.”
The CapRAM is formed by the use of planar cross-point arrays of memcapacitors that can also be stacked in the third dimension. This provides the potential for extreme processing density. Semron envisages stacking up to 100 layers in a manner similar to 3D-NAND flash memories.
To remain compact while generating minimal heat the architecture uses multi-bit-equivalent analog-in-memory computing. The CapRAM also stores the weights for matrix multiplication as a variable capacitance.
The company claims that CapRAM provides a fundamental benefit because it has capacitive read-out the noise is lower than in all other technologies. “No matter whether SRAM, emerging memories such as phase change memories (PCM) or other resistive technologies are used, they will never reach CapRAM’s energy efficiency, and thus intelligence density,” the company claims on its website.
Related links and articles:
News articles:
In-memory AI startup EnCharge raises another US$22.6 million
Ten more startups for Intel Ignite in Europe
Power breakthrough for neuromorphic AI
OpenAI’s Altman wants to build multiple wafer fabs
AI to drive strong chip market growth in 2024, says TSMC
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
