Sparse AI chip developer raises ‘tens of millions of dollars’ in Series A
The funds are marked for the production of the company’s first chip and continued development of its data and weight stripping or “sparsification” approach to neural network computation.
The round was led by CoStone Capital and Greater Bay Area Common Home Development Fund, followed by Co-power Capital, Zhongke Huachuang Investment Fund and Shenzhen Angel Fund-of-funds.
Moffett was founded in Silicon Valley in 2018 before setting up its headquarters to Shenzhen. It aims to support the development of a comprehensive sparse neural network architecture by optimizing computing models, and provide a universal AI computing platform.
Wang Wei, the founder and CEO of the company, has a master’s degree in electronics and computer engineering from Carnegie Mellon University and spent nearly 20 years working on commercial CPU designs. He worked on Intel processors the fifth to tenth generation.
The Antoum chip developed by Moffett supports algorithms for sparsification at a rate of 32x while supporting mainstream development framework and extensive AI operation bases and models.
Based on this chip, Moffett AI has developed two AI computing accelerator cards, SparseOne, which focuses on low power consumption and cost performance, and SparseMegatron, which focuses on high computing power. These two cards will be released in 1Q22.
Related links and articles:
News articles:
GraphCore joins €2.6m project for real world AI
Embedded computing kit for AI uses sparse modelling
Intel launches second generation neuromorphic chip on 4nm
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
