According to BlockBeats, as of September 13th, approximately $7 trillion in short-term U.S. Treasury bonds are currently in circulation, of which about $6.1 trillion will mature within one year. Including the remaining maturing bonds, approximately $7.5 trillion in tradable U.S. Treasury bonds awaiting refinancing in 2026, with approximately $4 trillion and $3.5 trillion maturing in 2027 and 2028, respectively. If the Federal Reserve raises interest rates by another 75 basis points, the annual interest on short-term Treasury bonds alone could increase by approximately $50 billion, equivalent to 0.15% of GDP. Most of these bonds will be rolled over at a rate higher than the original market interest rate, meaning the U.S. is facing a more expensive and large-scale refinancing process.

According to BlockBeats, on September 13th, CryptoQuant analyst Darkfost stated that Bitcoin hasn't reached a new high for almost a year. It's currently about 342 days since the last high, nearing a full year. Previously, halvings would quickly lead to new highs, but this cycle is slowing down. The next halving is expected around April 2028.

Furthermore, the interval between the previous peak and the next new high is actually shortening: 1180 days from 2014 to 2017, 1094 days from 2017 to 2020, and 849 days from 2021 to 2024. Darkfost infers that if this shortening trend continues, although this round has dragged on for a year without breaking the previous high, the new high may come faster than in previous rounds, without having to wait until 2028 according to the old template of "a new high immediately after halving." The cycle template is becoming ineffective, but the interval between new highs is shortening and is inevitable.

According to BlockBeats, on September 13th, BitMEX co-founder Arthur Hayes stated that if the demand for AI computing power were to dry up, the only solution would be to print money to bail it out. The current so-called "AI safety first" approach essentially has only two paths: one is for the US government to step in, taking over the roles of Anthropic, OpenAI, and SpaceX, continuing as a major buyer of computing power to sustain data center and chip orders. The other is for these buyers to disappear, followed by a collapse of AI-related debt, with insurance companies initially bearing the burden, and then the Federal Reserve printing money to bail them out if they can't hold on any longer.

Arthur Hayes concluded that it doesn't matter which path is chosen, money will still be printed. Whether it's fiscal or monetary policy, liquidity will be released, and risky assets will ultimately benefit.