NVIDIA releases a new generation of open-source large language models!
Tianhong's Sci-Tech Chip Design ETF traded nearly 50 million yuan yesterday, with institutions optimistic about the growth potential of domestic AI chips.

robot
Abstract generation in progress

Yesterday (March 12), the three major A-share indices all declined collectively, with technology sectors such as semiconductors pulling back. The Shanghai STAR Market Chip Design Theme Index closed down 1.64% yesterday, with constituent stocks like Lixin Micro, Sirui Pu, and Jingchen Shares rising over 2% against the trend, and Xinxiang Micro and Nanchip Technology rising over 1%.

Among related ETFs, the Sci-Tech Innovation Chip Design ETF Tianhong (589070) had an average daily turnover rate of over 8% yesterday, with a trading volume close to 50 million yuan, showing active trading throughout the day. Data shows that on the previous trading day (March 11), this ETF experienced a net capital inflow of 1.91 million yuan.

The Sci-Tech Innovation Chip Design ETF Tianhong (589070) has a 20% fluctuation space. This product closely tracks the Shanghai STAR Market Chip Design Theme Index, which focuses on the core links of chip design, with related industry content approaching 95%, high industry concentration, and prominent “chip content.” Its constituent stocks include 50 leading companies in the STAR Market chip field.

On the news front, according to Securities Times, on March 11 local time, Nvidia announced the launch of the new generation open-source large language model Nemotron 3 Super, designed specifically for enterprise-level multi-agent systems. With a brand-new Mixture of Experts (MoE) architecture, it has increased inference throughput to more than five times that of the previous generation model.

Bohai Securities believes that by 2026, the artificial intelligence industry is expected to officially enter the era of intelligent agents. Currently, AI applications are continuously being catalyzed, and the industry’s commercialization inflection point may arrive earlier. Driven by accelerated model iteration and the promotion of applications by major internet companies, the computing power industry chain presents relatively certain investment opportunities.

Western Securities (002673) also pointed out that the rapid iteration of AI inference chips may drive unit token calculation costs to continue decreasing, promoting downstream procurement demand, and potentially driving significant and relatively certain growth for domestically produced AI chips.

View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pin