SpaceX 将其 IPO 目标估值提高至 2 万亿美元以上，该公司正准备向潜在投资者推介这一可能是史上最大规模的 IPO。SpaceX 估值在数月内将增长近三分之二。今年 2 月，SpaceX 收购 Elon Musk 旗下 xAI 后，合并公司估值为 1.25 万亿美元。若以超 2 万亿美元估值上市，SpaceX 市值将超过标普 500 指数中除 Nvidia、Apple、Alphabet、Microsoft 和 Amazon 之外的所有公司，也将超过 Meta 和 Tesla。SpaceX 已秘密提交 IPO 申请，上市最早可能在 6 月进行，预计筹资至多 750 亿美元，远超此前最大 IPO——Saudi Aramco 2019 年 290 亿美元的纪录。所筹资金将用于太空 AI 数据中心和月球工厂项目。SpaceX 已选定 Bank of America、Citigroup、Goldman Sachs、JPMorgan Chase 和 Morgan Stanley 担任 IPO 主要承销商，并计划于下周一召开银行团电话会议，4 月晚些时候进行分析师简报。Bloomberg Intelligence 分析师指出，SpaceX 2026 年营收接近 200 亿美元，其中发射业务和 Starlink 贡献了大部分收入。

The U.S. nonfarm payrolls for March showed a significant increase, according to Jin10. The seasonally adjusted nonfarm employment rose by 178,000, far exceeding the market expectation of 60,000. This marks the highest level since December 2024 and represents a substantial improvement from the negative growth observed in the previous month.

非农公布后，美元指数 DXY 短线大幅拉升，最高报 100.1。非美货币对普跌，欧元兑美元 EUR/USD、英镑兑美元 GBP/USD 短线下挫近 30 点。(金十)

The unemployment rate in the United States experienced a slight decline in March, reaching 4.3%, according to Jin10. This figure was lower than market expectations, which had anticipated the rate to remain steady at 4.4% compared to the previous month.

The dollar index (DXY) experienced a significant short-term increase, reaching a peak of 100.1 following the release of nonfarm payroll data. According to Jin10, non-U.S. currency pairs saw a broad decline, with the euro against the dollar (EUR/USD) and the British pound against the dollar (GBP/USD) both dropping nearly 30 points in a short span.

美国 3 月季调后非农就业人口增加 17.8 万人，远超 6 万人的市场预期，创 2024 年 12 月以来最高水平，较上月就业“负增长”的情况大幅改善。美国 3 月失业率 4.3%，预期 4.40%，前值 4.40%。(金十)

美国 3 月季调后非农就业人口 17.8 万人，预期 6 万人，前值由-9.2 万人修正为-13.3 万人。(金十)

A recent staff paper from the Bank of Canada has highlighted the effectiveness of Aave V3's risk management model, which reported zero non-performing loans in 2024. According to Cointelegraph, the study attributes this success to the platform's overcollateralization and automated liquidation processes, which have been instrumental in preventing lender losses within its Ethereum lending market. The research utilized transaction-level data spanning from January 27, 2023, to May 6, 2025, revealing that positions were typically liquidated before collateral values fell below the outstanding debt, thereby containing lender losses across the sample period. However, the paper notes a tradeoff in Aave V3's model. While it effectively shields lenders from unrecovered losses, it shifts risk onto borrowers and limits capital efficiency compared to traditional lending systems. The design of Aave V3 relies on automated risk controls rather than conventional underwriting, requiring borrowers to post more collateral than they borrow and liquidating positions when risk thresholds are breached. The study also found that Aave V3's lending activity was not solely driven by liquidity needs. Recursive leverage, which involves repeatedly borrowing against collateral and redeploying borrowed assets as new collateral, accounted for over 20% of the total borrowed volume and 8.2% of borrowing transactions during the sample period. The paper highlights that this dynamic exposes borrowers to increased risk during market downturns. Liquidations on Aave V3 were observed to occur in concentrated waves, with four assets—Wrapped Ether (WETH), Wrapped Staked Ether (wstETH), Wrapped Bitcoin (WBTC), and Wrapped eETH (weETH)—accounting for 90% of the total liquidated value. The study estimates that borrower losses during major liquidation events could be substantial, with liquidation fees typically ranging from 5% to 10% of the liquidated value. Additionally, missed gains from subsequent price recoveries could push combined losses to about 10% to 30% in some cases. The paper concludes that while Aave V3's design helps prevent unrecovered bad debt, it does so by exposing borrowers to abrupt losses when collateral prices fall sharply. Cointelegraph reached out to Aave for comment but did not receive a response before publication.

据 CME“美联储观察”：美联储 4 月加息 25 个基点的概率为 0.5%，维持利率不变的概率为 99.5%。美联储到 6 月累计降息 25 个基点的概率为 7.8%，维持利率不变的概率为 91.7%，累计加息 25 个基点的概率为 0.5%。

A recent report by the Bank of Canada reveals that Aave V3 on Ethereum reported zero non-performing loans (NPL) in 2024, with a total loan volume of approximately $6 billion, a utilization rate of 40%, and a net interest margin of 0.64%. According to Foresight News, the report highlights that platform profits are highly concentrated, with WETH, USDT, and USDC contributing about 83% of the earnings. The study, which analyzed transaction data from January 27, 2023, to May 6, 2025, found that over-collateralization and automatic liquidation help prevent lenders in Ethereum's lending market from incurring losses. However, the model shifts risk to borrowers and limits capital efficiency compared to traditional lending systems. In terms of user behavior, recursive leverage trading accounts for 20.46% of total borrowing volume and 8.2% of borrowing transactions, with 58% of leverage traders being large investors. Liquidation activities show significant clustering, with WETH, wstETH, WBTC, and weETH comprising 90% of total liquidation value. Borrowers' combined losses during major liquidation events, including liquidation fees and the opportunity cost of missed price rebounds, account for about 10% to 30% of the total liquidation amount. Additionally, the research indicates that liquidation activities do not have a significant lasting impact on market prices.

由特朗普的两个儿子——长子小唐纳德·特朗普和次子埃里克·特朗普部分持股的无人机公司Powerus正在向海湾国家出售无人机拦截器。

机器如何为自己在现实世界中的行动买单、授权、并留下可信记录？

后量子迁移的关键是让系统具备应对密码学持续演进的能力。

Solana 生态衍生品与借贷一体化协议 Drift Protocol 出现异常资金外流，项目方确认正遭受攻击，最终导致协议约 2.8 亿美元资金被盗。

Robinhood,交易所,Coinbase,焦虑经济：交易为何爆发式增长 金色财经,未来十年金融市场的核心逻辑。

MARA正主动“瘦身”，将资源从传统挖矿业务中抽离，转向更具增长潜力的AI领域。

Anthropic亲手掀开AI大脑，发现了171个可怕的“情绪开关”。

人工智能,AI Agent,Token,算力,AI时代的省钱哲学：如何把每一个Token花在刀刃上 金色财经,AI时代下的省钱方法论。

看似平静之下却暗流涌动，往往都是暴风雨来临之前的假象。当流动性和波动率低到极至时，随后都会出现更大的波动。

目前而言高油价最大的影响并非经济层面，而是心理层面。关税、移民遣返、人工智能与机器人化带来的影响，远比高油价更持久。