Bank of America has released a research report initiating coverage on Knowledge Atlas (HKEX: 02513) and MINIMAX-WP (HKEX: 00100) with a "Buy" investment rating, setting price targets of HK$1,250 and HK$500 respectively.

The bank states that the global artificial intelligence (AI) market is developing a two-speed structure: US companies are likely to maintain a lead in frontier capabilities, while Chinese models are gaining market acceptance due to their low cost and high performance. Despite constraints on computing power, Chinese firms are optimizing their spending strategies. This value-for-money advantage is becoming increasingly important globally as the price of frontier models rises, particularly in high-volume, non-critical application scenarios.

Competitive Dynamics in China's AI Market

The report notes that China's AI model market is vibrant and exceptionally competitive, with a limited capability gap between established players (such as Alibaba (HKEX: 09988/NYSE: BABA) and ByteDance) and independent AI labs (like DeepSeek, Knowledge Atlas, and MINIMAX). In this environment, minor advantages in model performance are unlikely to yield lasting profitability. Instead, computing resources, ecosystem integration, and distribution channels are the key moats, which structurally favor the incumbents.

With stronger balance sheets and integrated cloud platforms, incumbents are better positioned to withstand lower profit margins at the model layer, thereby driving cross-selling of computing resources. Meanwhile, DeepSeek's rising capabilities and limited commercialization strategy are redefining the industry benchmark towards "high performance, low cost," compressing the profit potential for independent model providers.

Pathways to AI Commercialization

The commercialization of AI remains enterprise-led, with cloud and Model-as-a-Service (MaaS) being the clearest pathways. In the consumer AI space, not all users are equally profitable. Value is shifting towards high-intent "prosumers," while mass-market commercialization through subscriptions is still in its early stages. Advertising/transaction-based models may offer longer-term revenue options.

Bank of America anticipates that the relative outperformance of AI labs versus internet giants will gradually diminish as investor focus shifts to computing power and ecosystem strength. Consequently, numerous catalysts in the second half of 2026 are expected to drive individual stock performance.

Comparing the Two New Buys

Between the two, Knowledge Atlas's premium valuation reflects its faster Annual Recurring Revenue (ARR) growth, stronger talent density/public support, and its leadership in enterprise revenue. However, its expected 2027 price-to-sales ratio of 52x, compared to MINIMAX's 19x, appears high relative to its product breadth.

The bank believes MINIMAX could see catch-up trading after its lock-up expiry (July 8th), driven by its potential inclusion in the Southbound Stock Connect in early August and a potential catch-up in enterprise revenue following the M3 release.

Other Key Players in the Theme

Bank of America views Alibaba as the best proxy for the "China AI/Cloud Computing" theme. The bank is also positive on the accelerating momentum in models/consumer agents at TENCENT (HKEX: 00700). Given ByteDance and telecom operators' aggressive capital expenditure and their growing influence in the AI model/domestic chip ecosystem, they remain key swing factors.