HAITONG INT'L: First granted "Outperform" rating to KNOWLEDGE ATLAS (02513) and MINIMAX-WP (00100) over the broader market.

HAITONG INT'L released a research report stating that they are first covering KNOWLEDGE ATLAS (02513) and MINIMAX-WP (00100), and giving them a "outperform" rating, with target prices of 860 Hong Kong dollars and 1,400 Hong Kong dollars respectively. The target price for KNOWLEDGE ATLAS is equivalent to a forecasted market-sales ratio of 70 times by 2028. The competition in the technology modeling field changes rapidly, some even consider it too fast, as seen from the constant changes in model rankings even on a quarterly basis. However, the report points out that compared to the emergence of numerous LLM startups to capture the new trend in the early days of ChatGPT, the current market participants have a much stronger foundation. The report believes that a more predictable competitive environment will increase opportunities for continued growth of MiniMax and KNOWLEDGE ATLAS, and may become part of the "endgame" of the artificial intelligence era. The report advises investors to not focus on quarterly ranking lists, but instead focus on the current success factors, innovation speed, and application of KNOWLEDGE ATLAS and MiniMax. Both companies are able to iterate their model versions at an accelerated pace, and the application of their various model capabilities is becoming increasingly evident, as seen from the recent price increase of encoding products for KNOWLEDGE ATLAS, and MiniMax's continued lead in the OpenRouter usage rankings. While valuation is certainly important, for MiniMax and KNOWLEDGE ATLAS, even though both companies are already listed, they seem to have more of a venture capital color at this stage. The report advises investors not to hesitate, but rather to cherish the opportunity to own these two top-notch model companies through the public market. Also worth noting is the rise in forecasted adoption rates of artificial intelligence in 2026, such as agent usage and multi-mode application cases, which will drive overall sentiment in the artificial intelligence market and token consumption of underlying models.