In the first 6 months of the year, chipmaker Cambricon saw its revenue increase by over 4,000% to 2.88 billion CNY (402.7 million USD). Net profit also hit a record 1.04 billion CNY.
Cambricon is one of many Chinese companies working to replace Nvidia in creating chips for training AI models and applications. Their business results are still small compared to the American chipmaker (46.7 billion USD in Q2).
However, the latest figures show that Chinese tech companies are urgently seeking alternatives to Nvidia, as they constantly face the risk of the US cutting off their technology supply.
 |
Cambricon's booth at the World AI Conference in 2021. Photo: Cambricon |
Cambricon's booth at the World AI Conference in 2021. Photo: Cambricon
This year, Nvidia was banned from selling its H20 chips to China. These were specifically designed for the Chinese market to comply with AI chip export regulations from the Joe Biden administration. Currently, they are allowed to sell again, but must share 15% of their revenue in the Chinese market with the US government.
However, the Chinese government is said to discourage domestic companies from buying H20 chips. The country's tech giants are using both domestic and foreign chips.
Cambricon's stock has more than doubled this year, boosting its market capitalization by over 40 billion USD, according to S&P Capital IQ. The company's current market value is 80 billion USD.
 |
Cambricon's stock price has soared this year. Chart: Google Finance |
Cambricon's stock price has soared this year. Chart: Google Finance
However, Nvidia's strength lies in its hardware and software, which have become familiar to developers. Cambricon says it is also improving its software and developing a new generation of hardware.
However, Nvidia's Chinese competitors face obstacles if they want to compete. Their technology still lags behind the American chipmaker, while the long-term outlook is more difficult due to export controls aimed at preventing Beijing from accessing the most advanced chip manufacturing techniques. This will affect efforts to develop domestic AI chips.
Ha Thu (via CNBC)
