According to Odaily Planet Daily, Nick Tomaino, a partner at 1confirmation, stated in his Q2 2026 letter to LPs that the fund's total cash dividends amounted to approximately $317 million, exceeding the total capital contributions of approximately $286 million across all funds.

MyToken data monitoring shows that on 【June 21st, 11:12】 (Beijing Time), an address starting with 【0xb0a2】 transferred 【87233126.1 USDT】 to an address starting with 【0xdac1】, with a total value of approximately 【87.13 million】. Real time monitoring of large transfers：https://www.mytokencap.com/large-transfers/

According to Odaily Planet Daily, since the beginning of the year, the semiconductor and optical communication AI hardware sectors in the US stock market have risen, with the upstream semiconductor equipment sector also experiencing a corresponding increase. Nine semiconductor equipment companies with a total market capitalization exceeding $10 billion in the US stock market have seen their share prices rise by more than 75% this year. Among them, Applied Materials, Lam Research, KLA-Tex, Teradyne, MKS Inc, Integrity, and Onto Innovation Inc have all seen their share prices more than double this year.

According to Bijie.com, Cai Yunda, Senior Principal Engineer of AI at Tesla, pointed out that while it's commonly believed that 99% of the work in machine learning projects involves training, the actual time spent on model parameter training is only 2%. In contrast, 50% of the effort is spent on evaluation and testing, 40% on data cleaning, and the remaining 8% on system integration. Cai emphasized that data cleaning and evaluation determine the limits of what AI can learn. If the original data is vaguely defined or the labels are contradictory, noise is introduced at the source. No amount of algorithmic magic or parameter tuning techniques can eliminate background noise because the model cannot correct its own errors; the final accuracy limit depends entirely on the amount of effective information in the data itself. To ensure consistent data standards from the source, Cai stated that he re-examines the definitions and classification systems of data concepts daily, even repeatedly reviewing historical labels. Many practitioners agree, pointing out that whether it's the rule setting for reinforcement learning or the precise labeling for model fine-tuning, the performance of AI is ultimately determined by data quality and evaluation level, not the model architecture itself.

Token | Trading Volume 【BTC】2.02 billion 【UNI】125.66 million 【SPX】1.34 million 【PI】1.20 million 【WLD】615.72 million 【NEAR】180.53 million 【AXS】29.61 million 【GWEI】2.62 million 【DCR】233,813.24 【OKB】13.96 million