US chip export ban is hurting China’s AI startups, yet the giants haven’t been hurt as much


Before Washington banned Nvidia’s export of high-performance graphics processing units to China, the country’s tech giants were hoarding them in anticipation of an escalating tech war between the two countries.

Baidu, one of the tech companies building counterparts of OpenAI in China, has secured enough AI chips to train its ChatGPT counterpart Ernie Bot for “the next year or two,” company CEO Robin Li said in a statement this week. said on the earnings call.

“Furthermore, inference requires less powerful chips, and we believe our chip stock, as well as other options, will be sufficient to support a multitude of AI-native apps for end users,” They said. “And in the long run, difficulties in obtaining the most advanced chips will inevitably affect the pace of AI development in China. Therefore, we are actively exploring options.”

Other deep-pocketed Chinese tech companies are also taking active steps in response to US export controls. The Financial Times reported in August that Baidu, ByteDance, Tencent and Alibaba had collectively ordered the delivery of about 100,000 units of A800 processors to Nvidia this year, worth $4 billion. They also purchased $1 billion worth of GPUs with delivery scheduled in 2024.

Such huge upfront investments could easily deter many startups from entering the LLM race. Exceptions exist if the young business manages to quickly secure good investment. 01.AI, which was founded by lead investor Kai-Fu Li at the end of March, acquired a substantial number of high-performance inference chips through debt and has already repaid its debt after raising capital, which The price was $1 billion.

With its stockpile of GPUs, Baidu recently launched Ernie Bot 4, which Li claims is “in no way inferior to GPT-4.”

Rating LLMs is difficult due to the extreme complexity of these AI models. Many Chinese AI companies have resorted to increasing the rankings by diligently meeting the criteria of the LLM chart, but the verdict is still pending on the effectiveness of these models when applied to actual applications in real life.

Smaller AI players lacking the cash flow to accumulate chips will have to settle for less powerful processors that are not under US export controls. Alternatively, they may wait for potential acquisition opportunities. Li expects that with a confluence of factors including the shortage of advanced chips, high demand for data and AI talent, and huge upfront investment, the industry will soon move into a “consolidation stage.”

Source link

About Author

Leave a Reply

Your email address will not be published. Required fields are marked *