Zhipu surges 33% as Wall Street raises bets on China AI after Anthropic curbs

Zhipu-related shares surged as much as 48% intraday and were last up 33% after JPMorgan raised its target to HK$1,400 from HK$950 and Bank of America initiated coverage with a HK$1,250 target. The bullish call reflects confidence in Zhipu's pricing power, enterprise exposure, and China’s share of the value-for-money AI market, while new U.S. restrictions on Anthropic model access may accelerate demand for Chinese alternatives. Zhipu's market cap reached HK$489 billion, nearly 4x MiniMax's HK$124.2 billion, reinforcing the widening valuation gap in Chinese AI equities.

Analysis

The immediate winner is not just Zhipu but the whole China AI monetization stack: policy-driven scarcity in frontier access is turning model quality into a distribution moat for vendors willing to stay open and cheap. That matters because enterprise buyers care less about absolute SOTA and more about a dependable cost/performance curve; if U.S. models become harder to access for global teams, the demand elasticities shift toward Chinese offerings faster than consensus expects. The second-order effect is a widening gap between “model hype” and actual billable usage, which should favor names with visible enterprise revenue and punish consumer-story peers still burning cash for mindshare.

For JPM and BAC, this is less about direct equity exposure and more about how quickly bank research can re-rate the private-to-public AI complex once a credible monetization path appears. The risk is that the current move front-runs a still-early revenue inflection: if the pricing increases choke off usage, or if open-source distribution commoditizes capability faster than expected, the premium multiple can compress just as fast as it expanded. In that scenario, the market will rotate from paying for model leadership to paying for distribution, cloud attach, and capital efficiency.

The biggest contrarian read is that export controls may be bullish for Chinese AI only in the near term; over 6-18 months they can also accelerate domestic substitution and reduce the willingness of multinational customers to build around China-linked infrastructure. That creates a barbell outcome: the strongest balance-sheet or enterprise-exposed platforms compound, while smaller or undifferentiated model vendors get trapped in a race to the bottom. The clean trade is therefore not a blanket long China AI basket, but a relative-value expression favoring the highest-quality monetizers over the most promotional peers.

Key Decisions for Investors

• Long Zhipu / short MiniMax as a 3-6 month relative-value pair: the market is already rewarding visible enterprise monetization and pricing power; stop if the valuation gap narrows without confirmation from revenue prints or if MiniMax shows faster API adoption.
• Buy BAC on any post-initiation weakness over the next 1-2 weeks, but treat it as a sentiment read-through rather than a fundamental driver; risk/reward is favorable if bank research continues to catalyze rerating in the China AI complex.
• Avoid chasing the broad Chinese AI basket after the move; instead, wait for a 10-15% pullback to build longs in the names with the clearest enterprise revenue exposure and strongest balance sheets.
• For a tactical hedge, buy 1-3 month puts on the more crowded AI momentum names if headline risk around export controls intensifies; the trade works if the market shifts from scarcity premium to monetization skepticism.
• Consider a long/short basket: long quality AI monetizers, short cash-burning model vendors, targeting a 15-20% relative spread over 2-4 months as investors separate ‘model access’ from ‘real ARR.’