Bitcoin mining firm MARA's associated address transferred 1,000 BTC to a centralized exchange (CEX), sparking speculation of a sell-off.

According to a report by the Financial Times, U.S. stocks were poised to open higher on Monday, following their largest weekly decline since April last year when Trump announced tariff policies. During Asian trading hours, tech-heavy Nasdaq 100 futures rose 1.3%, while S&P 500 futures gained 0.8%. Analysts noted that after Friday’s sell-off sparked by market concerns over widening credit spreads for AI data center-related firms, investors are now reassessing market conditions, and this easing of sentiment also supported Asian markets. South Korea’s KOSPI index climbed 0.5% on Monday, after an 11% drop last week. Prashant Bhayani, Asia Chief Investment Officer at BNP Paribas, said: “We are seeing a rebound because there was no substantial escalation of the Iran conflict over the weekend.” He added that after last week’s “de-risking” moves, traders have adopted a wait-and-see approach toward AI-related trades. Japan’s TOPIX index fell nearly 1% in early Monday trading but rebounded in the afternoon, ending the day roughly flat. The tech-heavy Nikkei 225 index, meanwhile, declined 0.7%.

According to Bitget market data, WTI crude oil has fallen 2.00% intraday, trading at $96.97 per barrel, while Brent crude oil is down 2.1%.

Zcash’s new software team Zakura has released an engineering paper proposing to assign a separate viewing key for every ZEC swap executed via NEAR Intents, as a safeguard against future quantum computer privacy threats. When a user swaps ZEC for USDC, NEAR Intents will broadcast one of the user’s Zcash addresses on the NEAR public chain for refunds. This address holds a public key derived from the viewing key, which future quantum computers could exploit via Shor’s algorithm to reverse-engineer the key, potentially linking all of the account’s receipt history and memos. Zakura’s design grants each swap a unique viewing key, keeps the spending key in the user’s main account, and leverages Private Information Retrieval (PIR) to locate refunds. The team noted that wallets like Vizor, built on its open-source library, can integrate this feature within weeks—earlier than the January anti-quantum payment target set by Zcash developers.

Bybit CEO Ben Zhou recently told Wu Shuo in an interview that Bybit had covered the losses from the earlier North Korean hacker attack via operational revenue such as fees by the end of last year. He noted that crypto exchanges will continue to face hacker attacks in the future, so security systems should be designed on the premise of "assuming a breach is inevitable." On industry competition, Ben said the golden era of expansion for global crypto exchanges has passed, and the sector will see further consolidation going forward, while regional exchanges with licenses and local operational capabilities still have room to grow. He revealed that Bybit’s profits have declined year-over-year this year, and its traditional asset trading business (including stocks) is currently insufficient to offset the impact from its crypto operations. Regarding the U.S. market, Ben said Bybit is still waiting for the right entry opportunity and prefers to partner with local institutions that hold a retail user base. Over the next year, Bybit’s core strategy will focus on regional development and deep localization. In addition, he also discussed the differences between the Bitget security incident and Bybit’s own hack, as well as the large-scale application of AI within the platform and the resulting data security challenges.

JPMorgan Chase said it upgraded Intel (INTC.O)’s rating from "Underweight" to "Neutral".

Beating AI News: Interviews with multiple headhunters by Lei Feng Network reveal that major Chinese internet giants are tightening recruitment quotas, yet still willing to break salary caps for top AI talents. Headhunters disclosed that this year, outstanding fresh graduates specializing in foundational models have secured annual packages reaching the 2 million yuan mark. ByteDance’s Seed program and Xiaohongshu have offered the highest quotes, ranging from 4 million to 5 million yuan, while Baidu’s offers fall between 2 million and 3 million yuan. This high-salary talent race is exacerbating "salary inversion" within major firms. An internet industry insider shared that some foundational model employees, even at lower ranks, now earn more than colleagues at higher levels. Such issues emerged as early as 2017: a veteran engineer recalled that a returnee fresh graduate who joined one year later and held a lower rank earned 10,000 yuan more per month than him, leading to a monthly turnover rate of around 10% for his over-30-person team at the time. However, high salaries have not benefited all AI practitioners. Headhunters noted that application-side roles at major firms are shrinking, while foundational model teams still lack personnel capable of training large models. Coding Agents, model evaluation, pre-training data, and Agent infrastructure have become new key targets in the talent race. DeepSeek has recently been expanding its backend development and Agent infrastructure engineering teams. In contrast, recruitment enthusiasm for areas like AI search has declined. The threshold for this recruitment wave is rising. Companies no longer only focus on academic qualifications and papers, but also value whether candidates have hands-on experience training large models in top teams. For truly scarce talents, compensation and responsibilities can be negotiated separately; ordinary AI job seekers face increasingly fierce competition.