MyToken data monitoring shows that on 【July 2nd, 19:27】 (Beijing Time), 【Binance.DepositAndWithdraw_1】 transferred 【500000000 USDT】 to an address starting with 【0xdac1】, with a total value of approximately 【498.65 million】. Real time monitoring of large transfers：https://www.mytokencap.com/large-transfers/

MyToken data monitoring shows that on 【July 2nd, 19:22】 (Beijing Time), 【Bitfinex.DepositAndWithdrawAndGasfee_1】 transferred 【273800000 USDT】 to an address starting with 【0xdac1】, with a total value of approximately 【273.06 million】. Real time monitoring of large transfers：https://www.mytokencap.com/large-transfers/

MyToken data monitoring shows that on 【July 2nd, 19:20】 (Beijing Time), an address starting with 【0x5754】 transferred 【220000000 USDT】 to an address starting with 【0xdac1】, with a total value of approximately 【219.41 million】. Real time monitoring of large transfers：https://www.mytokencap.com/large-transfers/

Odaily Planet Daily reports that Goldman Sachs states investors are reducing their exposure to the "Big Seven" tech giants and favoring beneficiaries of artificial intelligence, such as semiconductor companies, rather than hyperscale cloud providers that fund large-scale AI spending. Goldman Sachs believes the market is rewarding companies that generate returns from AI investments while questioning those that bear the costs. This cautious stance on large tech companies is likely to persist until hyperscale cloud providers demonstrate stronger earnings growth.

According to CoinWorld, Google has open-sourced TabFM, a zero-shot large-scale table model. Users simply input a table containing historical data combined with the new row to be predicted, and TabFM can directly "read the table" and output the prediction result, skipping traditional model training and feature engineering. This model integrates alternating row and column attention with row vector compression mechanisms, enabling it to handle much larger datasets. Google used a structural causal model to generate hundreds of millions of virtual tables to pre-train TabFM. In authoritative evaluations using 51 real datasets, the model's overall prediction performance surpassed traditional supervised algorithms that had undergone meticulous manual parameter tuning. TabFM is open-source on Hugging Face and GitHub and will be directly integrated into Google Cloud Database BigQuery. In the future, users will only need to use a simple SQL query command (ai.predict) to invoke the AI for prediction.