もっと詳しく

China’s call to foster semiconductor independence has sent investors chasing after chip startups of all sorts. Moffett AI, a Shenzhen-based chip design company, has received a fresh Series A injection. The company did not disclose the exact amount, only saying that the proceeds were in the “several tens of millions of dollars.”

The round was led by CoStone Capital and the Greater Bay Area Homeland Development Fund, a financial vehicle set up to support startups in the GBA economic zone, which is China’s grand project to integrate Hong Kong, Macau, Shenzhen and several other cities in the southern province of Guangdong.

Other investors in the Series A financing include Co-Power, Grand China Capital, and Shenzhen Angel Fund of Funds, a strategic fund set up by Shenzhen’s municipal government to “hunt down the next Huawei, Tencent and DJI.” Moffett’s last round from March 2020 closed at 100 million yuan (around $16 million).

From autonomous driving vehicles to video streaming recommendations, artificial intelligence has become an essential part of our digital life. The skyrocketing demand for AI capabilities has put a strain on computing performance, so companies that provide AI acceleration, such as Moffett and Foxconn-backed Kneron, become coveted.

Moffett vows that its AI chips differentiate by using a technology called “sparsification“, a process that prunes redundant information from neural network models and ultimately leads to faster processing. The startup plans to spend its new funding on expanding the “ecosystem” of partners and clients that use its sparse technologies and mass-producing its first chip, Antoum, which is manufacturered by TSMC.

The startup claims that its chip has a sparsity rate of 32x, which makes its processing power 5-10 times higher than that of “international flagship products.”

Headquartered in Shenzhen, Moffett also keeps R&D teams in Beijing, Shanghai and Silicon Valley, where it was founded in 2018. The startup is run by a group of AI research scientists from Carnegie Mellon University and semiconductor veterans from Intel, Qualcomm, Marvel and Oracle.