Hong Kong stocks closed, with the Hang Seng Index down 0.73% and Zhipu gaining 8.90%.

Goldman Sachs’ latest research shows that over the past three weeks, semiconductor stock exposure explains around 54% of the performance divergence across different asset classes, while oil price exposure accounts for just 2%. As the influence of AI-related trading continues to grow, tech stocks have become a key driver of global market risk appetite and asset performance divergence. Goldman Sachs notes that investor risk appetite has recently been highly correlated with the relative performance of large-cap tech stocks. Last week, mounting concerns over AI investment returns, financing costs, and capital expenditures led to sell-offs in tech stocks and corporate bonds, pushing Goldman Sachs’ risk appetite indicator to its lowest level since April 2025. South Korea’s stock market serves as a prime example of AI trading’s market impact. Given semiconductor stocks’ heavy weighting in the Korea Composite Stock Price Index (KOSPI), shifts in global AI investment expectations directly influence local market performance. On Tuesday, the KOSPI closed up 3.1%, with SK Hynix and Samsung Electronics rising 5.2% and 4.2% respectively. However, Goldman Sachs emphasizes that the 54% and 2% figures reflect a statistical relationship within a specific time frame, and do not mean AI factors will drive global market movements in the long term. While the semiconductor sector’s influence has outpaced oil prices recently, the impact of high oil prices on inflation, monetary policy, and asset pricing remains significant. As large tech firms including Microsoft, Google, and Amazon prepare to release their earnings reports, the market will focus on whether AI demand can sustain growth and whether massive capital expenditures will translate into actual profits.

Modular blockchain network Initia has shared the latest fixes and recovery progress for the earlier MoveVM vulnerability security incident on social media. The project’s official team announced that the security patch for the vulnerability has been formally finalized. To guarantee the absolute safety of the fix, the team is currently conducting a rigorous joint audit of the plan with external security partners. Upon successful completion of the audit, the team will release the specific follow-up execution plan to the community. Per the official roadmap, the upcoming detailed plan will cover two core aspects: first, outlining how to safely and compliantly restore support for assets affected by the incident; second, clarifying the orderly restart of Initia’s L1 mainnet and all Rollup networks. The Initia team projects that the entire audit and network restart process will wrap up in the coming days. Previously, the security incident resulted in approximately $2.7 million in assets being illicitly transferred, leading the Initia network to proactively suspend operations to mitigate risks. This latest update means the network, after a brief shutdown, is gradually progressing toward full recovery.

Hong Kong-listed domestic large model stocks rally sharply in morning trading: Zhipu AI (ZHIPU) rose around 13.3% at one point, while MiniMax’s gain neared 6%. An on-chain whale built long positions in both companies amid the rally, purchasing a total of around $3.85 million, but has since incurred unrealized losses as gains faded. According to TradingBeats’ monitoring, the account started buying ZHIPU in batches at noon today, and has so far built 40,000 long positions worth around $2.8305 million at an average price of $70.76; it also bought 45,000 MiniMax positions in the same period, worth around $1.0221 million at an average price of $22.71. The whale’s new positions account for a large share of both contracts’ open interest (OI). By position count, the account’s ZHIPU and MINIMAX long positions equal around 20.6% and 26.4% of the respective contract’s OI. The account currently only holds these two contract positions, both with 1x isolated leverage. The Hong Kong market’s rally failed to fully extend into the afternoon. As of press time, Zhipu’s gain narrowed to around 8.9%, while MiniMax turned lower, down around 1.1%. The combined value of the two on-chain long positions is around $3.7983 million, with unrealized losses of around $54,300.

To protect users and maintain a high-quality trading market, Binance conducts regular reviews of all listed spot trading pairs and may delist some pairs for reasons including insufficient liquidity and low trading volume. Per the latest review results, Binance will delist the following spot trading pairs and halt trading in them starting at 03:00 UTC on October 30, 2026: AAVE/BTC, ATOM/BTC, CAKE/BTC, DASH/BTC, ETC/BTC, ICP/BTC, PENDLE/BTC, RENDER/BTC, TAO/BTC, THETA/BTC, XAUT/BTC, and XTZ/BTC.

Per official announcement, Huobi HTX launched perpetual contracts for EIDO/USDT, EQT/USDT, STM/USDT, TAIYOYUDEN/USDT, and SPOT/USDT on October 27, supporting long and short positions with leverage from 1x to 20x. Additionally, from now until 15:00 UTC+8 on October 29, Huobi HTX is hosting a new token contract trading campaign. Users who register, trade the eligible token contracts, and meet the specified thresholds will have a chance to split a total prize pool of 1 billion $HTX and 5,000 USDT in contract trading bonuses.

Jump Crypto’s Solana client Firedancer has updated its bug bounty program rules, requiring researchers to permanently burn 1 SOL (valued at roughly $100) for each vulnerability report submitted. No refunds will be issued if a report is invalid, out of scope, or a duplicate of an existing flaw; reports without the burned SOL will not be eligible for rewards. The move aims to address the surge in AI-generated vulnerability reports. In an audit contest held from April 9 to May 9 with a total prize pool of $1 million, 162 researchers submitted 809 reports, but only 127 unique security issues were identified, with nearly half of valid reports being duplicates. Jump noted that AI agents were involved in nearly all report discovery or drafting. Jump Crypto’s Ishan Bhatt previously stated that AI has reduced the cost of vulnerability detection, but auditing bugs remains highly resource-intensive. The large volume of duplicate or unusable reports once led to a backlog of 299 pending reviews, forcing engineers to spend more time verifying issues and delaying the investigation and patching of real vulnerabilities. Meanwhile, Jump has also increased bug bounties: eligible vulnerabilities will receive double the standard amount by November 20; for flaws in Solana’s new Alpenglow voting system, rewards go up to triple, with a maximum payout of $300,000 for critical-severity bugs.