Hardware

The Chinese got involved in the creation of productive GPUs, but not for the sake of video cards, but for AI and bigdata

Over the past year, there has been a lot of news about GPU development in China – potential competitors to NVIDIA and AMD GPUs. But on closer inspection it turns outthat most of the developments are aimed at the GPGPU niche – general-purpose computing. More precisely, to the machine learning and artificial intelligence market, where NVIDIA solutions are now driving fashion. This is the market that many Chinese startups are targeting.

Image source: AP

Image source: AP

Chinese private and public investors are willing to invest and are pouring money into every new local developer who challenges NVIDIA products, even if only in words. One or two out of a dozen new companies will survive, but this does not bother the sponsors. The Chinese authorities are strongly encouraging such investments, hoping to create national platforms for AI, machine learning and big data independent of foreign technologies and developments.

According to the American company Katten Muchin Rosenman, in 2020, the semiconductor industry in China attracted more than 140 billion yuan from financial institutions, surpassing the Internet industry in terms of investment. In the first five months of 2021, about 164 Chinese semiconductor companies received investments with a total funding of more than 40 billion yuan ($ 6.17 billion), close to the level of all of 2019.

By 2025, AI accelerator chips are expected to account for over 80% of total computing power. In just 5-7 years, AI will surpass the computational capabilities of traditional computing, which has been developing for more than half a century. Is it any wonder that money is flowing into this industry like a river?

Image source: Iluvatar CoreX

Image source: Iluvatar CoreX

Among the young Chinese AI companies that have received investor recognition are Iluvatar CoreX, Biren Technology, MetaX Integrated Circuit, and Moore Threads Technology.

Iluvatar CoreX was previously mentioned as GPGPU manufacturer Big Island (BI) Zhixin Semiconductor Co., Ltd. It was founded in 2015 and can be considered one of the oldest Chinese AI startups. In March of this year, Iluvatar CoreX unveiled their 7nm GPGPU ready for mass production (read more in our archive).

Image source: Biren Technology

Image source: Biren Technology

Founded in 2019, Biren Technology has raised over RMB 4.7 billion since its inception from over 40 investors including Hillhouse Group, Walden International China and BAI Capital. Today, its capitalization exceeds 10 billion yuan ($ 1.5 billion). This developer promises 7nm GPGPU, which will be no worse than NVIDIA’s 5nm design.

In 2020, MetaX Integrated Circuit was born, backed by two people from AMD. Only this fact allowed the startup to attract investments from such serious financial institutions as Lightspeed China Partners, Sequoia Capital and ZhenFund.

The same worked for an absolute newcomer, Moore Threads Technology, which has raised several billion yuan of investments in 100 days since its foundation in October 2020. Moore Threads Technology was founded by natives of NVIDIA, which did not leave indifferent investors.

Image source: Moore Threads Technology

Image source: Moore Threads Technology

All of these Chinese startups are about to conquer the AI ​​market and challenge NVIDIA and its products. But it should be remembered that NVIDIA’s strength lies not only in GPGPU, but also in software support for developers and in promoting the CUDA platform, which it has been doing since 2006. This is a huge and entrenched ecosystem, the success of which will be not only difficult, but incredibly difficult to repeat. However, there is one recipe for accelerating progress in Chinese startups – these are sanctions, which will leave no choice. No matter how convenient CUDA is, if the supply of the latest GPUs to China is banned, local developers will have to fine-tune their developments.

About the author

Dylan Harris

Dylan Harris is fascinated by tests and reviews of computer hardware.

Add Comment

Click here to post a comment