Nvidia introduced a new series of chips designed to enhance artificial intelligence, specifically targeting the Chinese market. As reported by sources Reuters, the company of Jen-Hsun Huang introduced i Blackwell series chips, designed to comply with US export restrictions and comply with current regulations.
The chip, currently called B20, will be launched and distributed in collaboration with Inspur, one of the company's main distributors in China. This strategy allows Nvidia to continue serving the Chinese market without breaking regulations imposed by the United States.
The new chips from Nvidia: here is the Blackwell series
La Blackwell series, presented in March 2024, includes the model B200, which offers up to 30 times better performance than its predecessor in some applications. Designed to meet the growing computational demands of artificial intelligence and machine learning, the Blackwell series represents a major upgrade over previous models A100 and H100, with notable improvements in speed and processing power. Mass production of these chips is scheduled for the end of the year.
Blackwell chips, including the B200 and B20 models, are intended mainly to sectors that require high computing power, such as data centers, research institutes and large technology companies. For example, Tesla and Meta Platforms, already users of the A100 and H100 chips, could adopt Blackwells to manage increasingly complex and sophisticated workloads. Additionally, these chips are designed to meet the needs of large language models and other advanced AI applications.
The problem of duties and the impact on the market
Nvidia created new chips for overcome limitations imposed by the Biden administration in 2023, which has tightened export restrictions of advanced semiconductors to China. These rules aim to limit China's technological capabilities in supercomputing and forced the company to modify its chips, such as the H20, to match the requirements. Despite these efforts, the company has faced difficulties selling the H20 chip in China, due to strong local competition and lower-than-expected demand.
To meet these challenges, Nvidia has developed three chips specifically for the Chinese market, including the H20, whose sales, initially slow, are now growing and are expected to exceed one million units sold this year. The collaboration with Inspur will be crucial for the launch of the B20 chip and to maintain the company's competitiveness against local manufacturers such as Huawei.
Trade restrictions and geopolitical uncertainties are significantly impacting the global technology sector. The United States continues to apply pressure, seeking to further restrict exports of semiconductor equipment by countries such as the Netherlands and Japan, with the Biden administration considering new measures to limit the sale of products made with American technology .
Tensions between the two countries, along with semiconductor supply chain concerns, have contributed to market volatility and fluctuation of stock securities. Nvidia stock is at a one-month low. And then there are also Donald Trump's statements which have already had a destabilizing effect on the sector, and which, with his possible return to the White House, could further aggravate the situation.
To address these difficulties, Nvidia has formed strategic alliances, such as the one with AWS, which integrates Blackwell chips into Amazon's cloud services, offering users access to advanced technologies without requiring large hardware investments.
The chip war: competition is growing
Although Nvidia dominates the AI and GPU chip market, the competition is growing. Advanced Micro Devices (AMD) recently launched new Ryzen processors with Zen 5 architecture and announced the Radeon Instinct MI325X for the fourth quarter of 2024. Developments that position the company as a serious competitor, especially in challenging the semiconductor market for artificial intelligence.
AMD also plans to launch the MI350 and MI400 series in 2025 and 2026, respectively. These chips promise significant improvements in performance, with a predicted increase of up to 35x in AI inference, offering a serious alternative to Nvidia chips.
