Orbit Post Sitemap
山寨板块轮动周期是多长？？？
存量行情下，短线热点板块炒作周期大多3‑7天，牛市增量行情时可持续2‑4周。
当前市场流动性偏弱，轮动速度明显加快。以近期盘面举例：先是BEAT妖币行情持续4天热度快速退潮，随后资金短暂切换到存储映射SNDK，最近2天游资又突袭$GPS脉冲拉升，热点切换间隔越来越短。
长叙事赛道周期会更长，像AI存储、AI Agent这类具备产业故事的板块，轮动周期可以拉长至2‑4周，但中间也会不断穿插短期妖币分流资金。
大盘增量资金充足的牛市阶段，一个主流叙事板块往往可以持续1个月以上，板块内多只标的轮番上涨；一旦市场变成存量博弈，游资快进快出，板块行情往往刚被散户发现，热度就已经临近尾声。
轮动节奏快慢核心由市场总流动性决定，BTC、ETH走稳，场外资金进场时，板块持续性会拉长；大盘震荡、资金不足时，热点一日游、3日行情会成为常态。
本文仅行情复盘，不构成任何投资建议。#财报观察员：小米即将发布财报，你更看好哪条业务线？ #30年期美债收益率创2007年以来新高 #闪迪收涨逾8%，长期协议受关注 $BTC $ETH $SNDK Uniswap's founder stated that AMMs will take over the financial markets
Uniswap's founder is once again making big promises, but this time the pie is quite real. He said that AMM, that automated market maker mechanism, could become the core engine of financial markets in the future. Even after stock ETFs are listed online, they will no longer be quoted against the US dollar, but will be traded directly between assets. This sounds wild, but on closer thought, it actually makes some sense.
Here's the contrast. We're used to everything being priced in dollars, and buying ETH is based on USDT prices. But once stock ETFs move on-chain, the market may shift from all assets to dollar-to-dollar swaps directly for related assets. With lower holding risk and market-making costs, AMMs have the chance to penetrate the heart of traditional finance. It's not crypto catering to Wall Street, but Wall Street's trading methods being restructured on-chain.
In terms of practical significance for us, in the short term, it means the DeFi market will be expanded. Stocks going on-chain means on-chain liquidity will have to catch traditional assets, increasing opportunities for market makers and liquidity providers, and upgrading the logic of the LP business. This kind of narrative in the swing market repeatedly sends buzz to DeFi tokens, but don't treat it as an immediate positive factor—slow implementation has always been an old problem on the chain.
Some might say that the depth on the chain is just a small puddle compared to the NYSE. That's true, but every financial infrastructure upgrade in history started from a marginal pool. Back then, electronic trading was laughed at by traders on the market for a long time.
Of course, AMM isn't a cure-all. Its biggest concern is slippage when liquidity is insufficient; if the pool is small, a single large order can throw the price off the rail. Traditional exchange-style order book matching still has an advantage in depth. So in the short term, it's more likely that both will coexist, rather than one immediately replacing the other. But signs of the trend are already there: stablecoin settlement, RWA on-chain, stock tokenization—once these pieces are assembled, the variety of on-chain trading options will be unimaginable to us today.
In the long term, I believe in this direction. Asset on-chain is a slowing but irreversible trend. By then, wallets won't just hold coins, but the entire financial market will be held. Uniswap may not be as outrageous as it sounds. Do you think ten years from now, will we buy stocks in brokerage apps or complete exchanges with one click in wallets? This shift won't happen overnight, but the direction is already written into the code.$SNDK 暴涨的真正含义：AI行情已经从GPU扩散到“数据怎么存”
这几天美股最有流量的一条线，不是普通AI软件，也不是某个小盘概念股，而是存储股重新被市场追捧。$SNDK 单日大涨接近9%，五个交易日累计涨幅一度超过30%，这不是单纯因为NAND涨价，而是市场突然意识到：AI基础设施的瓶颈，可能不只在GPU，也在存储。
过去大家看AI，第一眼看 $NVDA。因为模型训练要GPU，推理也要GPU，数据中心预算最大头都被GPU吃掉。后来市场开始看HBM，因为GPU再强，没有高带宽内存，性能就上不去。现在轮到NAND和企业SSD被重新讨论。原因很现实：AI模型不是训练完就结束，真正大规模商业化之后，每天都要处理推理请求、读取上下文、调用数据库、保存用户数据、做缓存、做检索。数据不是躺在那里不动，它要被不断读取、写入、迁移和调用。
$SNDK 的故事就在这里被改写。以前它是典型存储周期股，手机、电脑、SSD补库存时涨，库存过剩时跌。投资者不愿意给高估值，因为存储行业太周期，价格一涨，厂商扩产；客户提前备货，未来又去库存。这个循环太熟，市场自然打折。但AI推理时代把NAND带进了新场景。尤其是KV cache、企业SSD、数据中心闪存这些关键词出现后，$SNDK 就不再只是“消费电子存储供应商”，而是被包装成AI推理降本基础设施。
这点很关键。AI训练阶段拼的是算力峰值，推理阶段拼的是成本效率。模型如果每天服务数亿用户，单次推理成本就会变成生死线。昂贵内存不能无限堆，低成本、高密度、可规模化的闪存就有机会进入更重要的位置。存储不再只是后台配角，而开始参与AI系统成本结构的优化。
不过，$SNDK 这波也不能只写成无脑利好。涨得越快，市场后面越挑剔。投资者会追问：长期供货协议是不是真的锁住利润？AI客户需求能不能持续到2027甚至2030？NAND涨价会不会刺激产能扩张？消费端如果被高价挤压，会不会拖累整体出货？这些问题都没消失。
所以 $SNDK 当前最值得写的，不是它今天涨了多少，而是它的身份正在切换。过去它吃库存周期，现在它想吃AI数据周期。一个公司最重要的变化，不一定是价格先变，而是市场解释它的方式先变。$SNDK 现在被重新解释成AI推理时代的数据底座，这才是它最近有流量的核心。
AI第一阶段买GPU，第二阶段买HBM，第三阶段开始买存储。$SNDK 这波上涨，说明市场已经不满足于只看最前排的AI龙头，而是在寻找AI基础设施里还没被完全定价的零件。 OP's 540 million token airdrop will be confiscated
Those still holding OP tokens, don't rush to swipe away. A 540 million token airdrop is facing the risk of confiscation, and its fate will be decided by a fierce civil war over governance. Currently, about 9.1 million OP votes are in favor and about 4.25 million OP are opposed; neither side has reached the point where a decisive decision is made.
What's the fault with this? Airdrops were originally a favor from the project team to the community, but on Optimism, unclaimed tokens might actually be taken back to the protocol. Supporters say the money should be used to build the ecosystem, while opponents see it as exploitation of early contributors. The same coin, wanting to confiscate and distribute it on one hand, is not about the money, but about the right to interpret the rules.
From the perspective of memes and altcoins, this is actually an old problem in the crypto world. Many projects have governance tokens with voting thresholds so high that retail investors can't participate; only a few big addresses really influence the outcome. OP just put the conflict on the surface this time, showing you that so-called decentralized governance sometimes just moves power from one company to a few wallets. You think you're participating in governance, but the decision-making power has already been divided.
Ultimately, this kind of drama damages the reputation of the entire governance token. Investors were already skeptical, and now even airdrops can be bought at will—who would dare hold long-term? If a project can't even uphold the most basic contract spirit, no matter how high the token price jumps in the short term, it's just a castle in the air.
Someone even did the math: if these 540 million coins were really confiscated and redistributed, the short-term selling pressure wouldn't be huge, but the symbolic meaning would be very negative. It essentially tells everyone that the governance power in your hands could be rewritten by the rules at any time. This kind of uncertainty is even more damaging than a ten-point drop in coin price. Once community trust is broken, it's much harder to repair than to pull up the market.
In the short term, this kind of internal friction is a real drag on coin prices; funds fear uncertain rules the most. For the swing market, I suggest not touching these unresolved governance coins; wait until the results are settled, so you don't get shaken out by the tug-of-war between the two sides. In the long run, whether the governance mechanism can be fixed determines whether OP is still trustworthy. Are you on the confiscation camp or the distribution camp? Tell me your reasoning; I think that's much more interesting than guessing the coin price. The fear of rules being rewritten is more sleep-inducing than price fluctuations.Bitmine added $18.9 million worth of ETH in one week
Tom Lee never gives a top or bottom statement, yet Bitmine never stops buying ETH. The Data Nerd monitored that in the past week, Bitmine bought another $18.9 million worth of ETH plus $30.8 million in BMNR. This adds up to about 4.81% of his ETH holdings across the entire network—a staggering proportion for any institution.
What's even more interesting is the contrast. Retail investors in the market are still debating when BTC will return to 90,000, whether ETH will continue to grind bottoms, while institutions are already investing real money. BMNR is Bitmine's own ticker symbol; it buys ETH while buying itself, effectively binding the company and ETH even tighter. This approach is not about short-term volatility but betting on ETH's long-term position.
Bitmine has a premise behind this approach: it believes ETH staking yields can cover funding costs. Currently, ETH staking annualized returns are about 3% to 4%, which sounds small, but rather than just holding cash on accounts, institutions prefer to let it generate interest. Combined with RWA and stablecoin settlement narratives, ETH's story is deeper than just hoarding. Another point: BMNR is originally an ETH proxy target; buying it means institutions indirectly add positions without directly touching coins, making it especially friendly to compliant funds.
The most common mistake retail investors make is rushing in just because they see institutions buying. But you have to think carefully: Bitmine buys on a half-year-year cycle, and it can afford to lose time, but you might not be able to afford it. Its holdings are open and transparent on-chain addresses, basically laying their cards on the table. By the time the day they actually exit, retail investors often see it too late. Learning direction by watching institutional actions is not about blindly copying homework.
For the market, holding 4.81% of the supply means ETH's chip structure is more concentrated than it appears. In the short term, if these whales don't sell, ETH will have an invisible bottoming line. But conversely, if it really reduces its holdings, selling pressure won't be small. On the swing side, I think ETH is actually more resilient to declines compared to Bitcoin, but don't treat it as a risk-free asset; institutional positions are a double-edged sword.
I agree with the long-term logic—ETH staking and RWA narratives are still growing. Right now, this pace of continuous institutional buying provides psychological support for those holding spot assets. But do you think monopolizing nearly 5% of the supply is just stabilizing the market's foundation, or is it planting a landmine that could explode at any moment? This kind of concentration is a double-edged issue, and every holder should keep a close eye on their on-chain addresses.#BTC 低交易量流动性陷阱：$64K 反弹的背后
比特币在月初价（Monthly Open）处强烈反应，上涨至 $64K 水平。月初价是机构高度重视的基准线。通过分析主要交易所数据，可以在低交易量情况下把握此次急涨的结构性原因。
资金费率及持仓分歧
在月初价时点，主要交易所的资金费率明显分化。
- 币安、Bybit、OKX、Deribit：资金费率进入 -0.00 水平的负值，空头持仓优势明确显现
- HTX：维持 +0.05 的正资金费率，多头持仓占优
反弹机制
在 $62.7K 的月初价支撑线附近，价格横盘期间，以币安为中心的各大交易所资金费率转为负值。这种过度拥挤的空头持仓成为空头挤压的主要催化剂，推动价格上涨。
当前市况及风险
- 资金费率趋势：币安、OKX、Bybit、Deribit 的资金费率正从中性再次向负值方向移动。如果这种环境持续，短期上行泵动可能进一步发生。
- 交易量及波动性风险：此次上涨缺乏坚实的现货交易量支撑。市场流动性不足会放大波动性，使价格变动更加剧烈。
- 流动性错觉：最近的剧烈双向波动更接近短期流动性扫荡，而非可持续的有机趋势。
关键阻力位：$65K 水平仍作为强劲阻力发挥作用。在低交易量环境下，此区间下方需保持警惕。
✏️ 一行总结
#BTC 月初价 $62.7K 反应引发空头挤压，反弹至 $64K，但现货交易量支撑不足且资金费率再度转为负值，因此需同时关注 $65K 阻力位突破与交易量恢复情况。 He shorted $16 million but bet that Bitcoin would not collapse
A trader named LucasMeow recently did something particularly difficult on Hyperliquid. He used 40x cross-margin to short 100 BTC and 25x cross-margin short 5,000 ETH, with the total nominal value of the two short positions about $15.89 million, currently holding a profit of about $503,000; At the same time, he went to the prediction market Polymarket, investing nearly $470,000, betting that these coins wouldn't crash at all. Simply put, his left hand is betting on a crash, the other is betting on whether it will crash.
Looking specifically at his Polymarket positions, the logic is even more intriguing. He spent about $104,000 betting that BTC would not fall to six low levels between $15,000 and $45,000 within the year; He also spent $48,200 betting that ETH wouldn't fall to $800; The largest bet was $314,500, betting that Satoshi Nakamoto would not touch that batch of dormant coins this year. The cumulative cost of the eight prediction positions was about $466,700, roughly 86.5% of the $539,000 margin on his two short positions.
That's quite ironic. Someone who bet their fortune on a crash then spends almost the same amount of principal and gambles that the crash won't happen. What's even more intriguing is that out of his insurance payout of over 460,000 yuan, the largest sum of 314,500 was entirely on Satoshi Nakamoto's untouchable coins, showing that what he truly fears isn't the market, but that ghost wallet that has been dormant for over a decade and no one can figure it out.
What's even more awkward is that this insurance is useless. Because when he buys, these events are highly likely No positions; even if all eight sets are cashed out, the profit ceiling would only be $50,800. But if his short positions reverse, BTC rises to $96,715 liquidation, ETH to $2,542, totaling about $5.993 million. That predicted profit doesn't even cover a fraction of the liquidation loss, accounting for only 0.85%.
Cross-platform hedging isn't new in the industry; veterans often predict the market to insure on-chain positions. But in normal play, insurance scale should match risk, but this trade is clearly misaligned. Going back and forth between Hyperliquid and Polymarket has become standard move for many leveraged players, but LucasMeow's move feels more like a placebo in a gambler's rai.
Interestingly, he still hasn't stopped. On-chain data shows he placed two non-reduced sell orders at 62,277 and 61,500 USD in BTC, preparing to add another 125 BTC to short, with a nominal amount of about 7.715 million USD, still stuck as a dead short.
Someone who only spends a little hedge dare to keep increasing positions on the road to a crash. With 40x and 25x leverage right there, any decent rebound could swallow him. Whether he truly saw the trend or was already stuck in a trap, only he knew.说自己是堡垒的巨头把三万亿藏进了附注里
Meta在路易斯安那州建的那个Hyperion数据中心，占地大概相当于1700个足球场。租约要到2029年才开始起算，一签四年，最长能续到20年，Meta还额外承诺，如果提前退租就给债券持有人补足差额。这么大一摊子，它的资产负债表上一分钱负债都没记。理由很简单，公司认为真要掏这笔补差钱的概率不高。
华尔街日报把九家公司最近一期证券文件的附注全翻了一遍。Alphabet、Meta、微软、亚马逊、甲骨文、英伟达、博通、SpaceX、AMD，附注里披露的表外承诺加起来大约3万亿美元，绝大部分直接跟AI基建有关。这九家过去一年合计资本开支6000亿，也就是说，藏在附注里的东西是它们真金白银花出去的5倍，也差不多是未偿租约加长期借款的3倍。更扎眼的是，两个月前这个数还是1.8万亿，两个月涨了五成。
拆开看，一头是1.2万亿的未开始租约，Meta一家就占3470亿；另一头是1.9万亿采购承诺，主要用来提前锁英伟达芯片和存储产能。按现行准则，这两类在交割或者开始付租之前都不用进表。合法，但也确实看不见。
还有一层更绕的东西叫剩余价值担保，规模约700亿。SPV先借钱买芯片，还款靠AI公司的使用合同现金流；AI公司要是停付，芯片转租或者卖掉抵债；还有缺口，就由芯片厂补上。谁在最后兜底？英伟达、博通这些卖铲子的。可这份担保在它们账上通常也不记。博通把这套逻辑推到了极致，代号Big Sky的项目给一笔350亿债务做担保，Apollo和黑石出钱买定制芯片租给Anthropic用，靠这层担保拿到投资级评级，融资成本压了下来。美银估算，博通的AI平台到2029年中期可能背上3700亿高级债务。黄仁勋那边的说法是，最高提供25%的残值支持，逐案评估。
评级机构已经坐不住了。穆迪说主要风险是这类交易在短期内密集发生，即便账面杠杆不高，也会限制博通的财务灵活性。标普干脆把残值支持定性为或有债务类义务，要算进调整后债务里。有位DoubleLine的组合经理话说得更直白，这就像在钻系统的空子，从评级机构那儿讨个优待，我们正在进入金融工程的时代，而搞金融工程就是在掩盖财务现实。CreditSights的比喻更狠，这相当于卖出一个看跌期权，繁荣期几乎不花钱，真到严重下行、客户违约、硬件跌价的时候，它才最要命。
当然也有人不当回事。Janus Henderson那边说，要触发残值支持得是极端情况，token使用量增速断崖式下跌，现在完全没看到，这些公司不是想藏，是想融资。
我更在意另外两个细节。Alphabet和亚马逊的自由现金流已经转负，资本开支超过了经营现金流入，这意味着接下来还得持续找市场要钱。同时中国的开源模型正在用低得多的价格提供接近的性能，相关token价格过去几周累计跌了超过五成。收入这头有变数，账单那头基本不可撤销。
这堆东西现在还不是要命的偿付危机，更像一堵被递延的折旧墙，和一张还没见底的账单。你们觉得，附注里的这三万亿最后会怎么落地？Whale crude oil short positions pile up to eight million dollars
Today, an interesting address popped up on-chain. The data comes from TradingBeats' on-chain monitoring, and every move of such wallets is closely watched. This account had been shorting storage stocks in the previous stage. Just one short position at SanDisk cost nearly $9 million, but the book still had a floating loss of over 300,000. Normally, when you're losing money, you should stop, but this one did the opposite, shifting new chips to crude oil.
Just this early morning, this address opened over 27,000 short contracts of Brent crude oil on Hyperliquid at an average price just over $88, with a turnover of about $2.46 million. Even worse, it placed over 60,000 sell orders near $91, with orders worth $5.67 million. If all were sold, its short positions in crude oil would have soared to $8.2 million.
I checked the details of its position and felt this position was a bit uncertain. Crude oil short positions had 20x leverage, and the current floating loss is only 25,000 yuan, which doesn't seem like much, but the liquidation price is set at 98.81 USD. Brent crude has already risen 3.4% in the past two days, at one point reaching 91 USD. With some signs of stirring in the Middle East, oil prices are pushing upward, and this layer of position is actually quite thin.
What really surprised me was the pace of their short selling. SanDisk, Seagate, and Western Digital all plunged before the market opened; it was already short on storage, and now it's turning its bullets to crude oil. It's rare for an address to bet on both storage and energy lines moving downward at the same time. It's like betting on the same thing: macro demand is about to collapse.
The problem is, the direction of its bet is twisted with its account status. SanDisk's short positions have already lost over 300,000 yuan, and it's still holding a sell order of $1.57 million at $1,821, refusing to reduce it. On the crude oil side, leverage is piling up again. Once oil prices and storage stocks rise instead of falling, losses under twenty times leverage will be spread out very quickly. People on the chain are all staring at this address, wanting to know what information it holds or if it's just holding on.
In on-chain perpetual markets like Hyperliquid, betting on whale addresses has become a daily routine for many people. Some copy it, some do the opposite; every time this address adds positions, it feels like it's publicly livestreaming its own cards. But what is made public is not just its profits, but also its pressure.
There is always a type of player in the market who refuses to admit losses and instead raises bets hoping to turn things around. The position this whale is standing in is either a foresight that others can't understand, or the next sample to be liquidated. Do you think it really has internal logic, or has it already been held hostage by its position?跌破四万九千四死拿也要亏
昨天有个数据看得我后背发凉。CryptoQuant的分析师Axel Adler Jr算了一笔账，比特币长期持有者的成本基数在四万九千四百美元。现在大饼站在六万四千一附近，看着离得远，可一旦价格掉到这个数以下，这批人整体上就要从浮盈变成浮亏。
什么叫长期持有者？就是那些买了币死死攥着、跌了也不卖、被叫做钻石手的那群人。他们手里握着一千六百三十五万枚比特币，只比七月三十日一千六百四十一万枚的历史峰值低了五点八万枚。过去七十八天，大饼一直在他们的成本基数和一点五倍之间晃荡，一点五倍对应着七万四千一百美元。说白了这帮人现在也就赚个三成左右，远没到该狂欢的时候。
最耐人寻味的是抛售频率。数据显示，过去十五天里只有两天出现长期持有者供应下降，也就是说他们几乎没怎么出货。按理说这种不肯卖的结构对价格是好消息，说明最坚定的筹码稳住了。可分析师也泼了冷水，这只能说明供给端干净，不代表新一波上涨已经开始，真要往上走还得需求端和价格端一起确认。
再换个角度想，现在价格相对成本基数只有一点三倍，放在整个周期里其实还偏低。历史上每次大饼跌穿长期持有者成本基数，往往不是软着陆，更像是踩踏的前奏。真到那个关口，钻石手们是继续死扛还是被迫松手，谁心里都没底。
问题就出在这个微妙的平衡上。一边是钻石手们纹丝不动，市场找不到砸盘的筹码；另一边他们也不是铁板一块，四万九千四就是那道账面的分界线。真摔下去，连最死忠的持有者都要开始算自己亏多少，那种情绪传导从来都不是小事。
我们现在处在什么位置？大饼还在六万四上方晃，离那条线还有不小的安全垫。可这两周宏观那边又是美债收益率创新高、又是地缘局势紧绷，风向说变就变。你是愿意相信钻石手能一直扛，还是觉得这道防线迟早要被考验？火币用户账户突现神秘转账官方火速否认
今天下午社区里冒出一件怪事。最早是几名用户在社交平台上晒出截图，说自己地址凭空多了一笔钱，随后越来越多人跟帖，说自己也中招了。而转账来源在区块浏览器上赫然标记着HTX。金额从几美元到几十美元不等，不是同一个数，但标签全都指向同一家交易所。钱不多，却架不住它来路不明，群里一下就炸了，有人担心自己账户出了岔子，有人怀疑是不是被什么人盯上了。
HTX这边反应不算慢。市场负责人六爷很快出来回应，说内部第一时间做了核查，目前可以确定官方没有开展任何相关转账或测试行为，这些钱具体从哪来、为什么打到你账上，还在进一步核实，事实确认之前不做推测。
这话等于把悬念甩回给了市场。既然不是官方主动转的，那链上那个HTX标签又是怎么来的。懂行的人都知道，浏览器上的交易所标记，本质是第三方数据公司根据资金流向做的标注，出错和滞后一直都有。一种可能是地址标记搞错了，把本不属于交易所的钱包误标成了HTX；另一种更阴险，是有人故意用一个带着交易所标签的地址往外打钱，制造官方在动的假象，好让收到的人放松警惕。
在加密圈，钱包里莫名其妙多出一笔钱，从来不是什么好兆头。早年不少钓鱼套路就是先往你账户塞点代币，等你去授权、去交互，私钥和资产就被悄悄卷走。更常见的是，攻击者先用小额转账测试这个地址活不活跃、有没有人在盯，确认有人上钩后再放大招。这次金额虽小，但这种反常动作本身就在试探你的反应。
眼下真相还没浮出水面。但有一点很明确，看到陌生转账别急着高兴，更别手痒去点任何附带链接或授权什么合约。你的资产安不安全，往往就卡在这种不起眼的小动作上。等HTX把来源查清楚，咱们再看看这出戏到底是谁在导。Polymarket, which claims to be decentralized, was instantly blocked by South Korea
Today, Korea dealt Polymarket a harsh blow. The Korea Broadcasting and Communications Review Committee held a meeting and directly determined that this overseas prediction market platform provided illegal betting, and decided to block its access. In plain terms, in the future, when Korean users want to open Polymarket, the door will be welded shut from the outside.
The regulators' reasoning is very straightforward. What Polymarket does is open trading based on outcomes beyond users' control—politics, sports, elections, and weather. Whoever guesses correctly takes it all—a typical winner-takes-all structure. The committee believes this is fueling speculation. Moreover, the platform itself is responsible for opening the market, setting rules, managing operations, and providing acceptance and settlement, making money by charging fees through share trading. In their view, this has already formed a complete closed loop for raising and delivering user funds.
What's interesting is Polymarket's defense. The platform has always emphasized that it is a non-custodial peer-to-peer trading platform running on smart contracts, without direct fundraising, managing users' money, or issuing sports lottery betting rights. We're all too familiar with this logic: decentralization, on-chain autonomy, code is law, and it's very effective as a free card for immunity.
But South Korean regulators shattered this shield with just one sentence. They say you can't use these technical features to circumvent domestic laws just because you don't provide Korean services, use decentralized technology, or have interfaces that look like centralized exchanges. To put it bluntly, how you package your technology is your business. As long as you actually provide illegal betting to Korean users, you should be regulated.
This incident has actually hit a weak spot for the entire industry. In recent years, many projects have treated decentralization as a universal shield, thinking that as long as contracts run on-chain and teams don't touch users' money, they can stand where regulators can't reach. But regulators are increasingly uninterested in this. They look at what you actually do, who is profiting, and where users' money ends up, not how beautifully the structure in your white paper is.
South Korea's blocking may not truly block those who want to bypass it, but the signal is very strong. When a platform is directly labeled by the authorities as illegal speculation, many more things will follow. Do you think the shield of decentralization can still withstand the hammer of regulation in the future?The Truth About Bitcoin's $64K Rebound: Short Squeeze and Liquidity Trap under Low Trading Volume
Bitcoin rebounded to $64,000 supported by the monthly opening price ($62.7K).
However, this rally was not a solid spot buy, but rather a strong short squeeze characterized by the liquidation of short positions on major futures exchanges.
Funding Rates: An indicator in the futures market that measures the skew between long and short positions. A negative value (-) indicates that the bear (put bet) is overheated.
Exchange funding rate divergence: Binance, Bybit, OKX, and Deribit fell into negative territory, forcing liquidations of overcrowded short positions and driving prices higher
Loss of trading volume: The lack of substantial spot trading volume supports the market, which only leads to a sharp increase in volatility in a liquidity environment
$65K Resistance Level: Below the strong resistance line near $65,000, it is still difficult to see this as an organic trend reversal
This rebound is closer to a "short-term liquidity sweep" rather than a sustainable organic rise.
With a lack of spot buying inflows, an attempt to break through $65K requires caution for sharp volatility.5.4亿枚OP空投或将被充公提案引爆治理内战
Odaily 8月18日消息，Optimism社区正被一场治理投票撕成两半。核心就一件事，一笔约5.4亿枚OP的空投激励，可能要被提案方直接收走，重新划进协议金库。
这批OP原本是留给早期用户和生态贡献者的奖励，按计划分批释放。但最近一份治理提案提出，要把这批还没发出去的额度重新归集，理由是提升资金效率、聚焦核心建设。目前赞成票约910万枚，反对约425万枚，赞成暂时领先，投票还没结束，社区已经吵到不可开交。
一边是协议方说要把钱花在刀刃上，另一边是早期用户觉得自己被单方面改了规则。很多人当年为了这波空投，真金白银做交互、刷任务、锁仓，如今说收走就收走，情绪可想而知。链上还能看到有人把已申领部分和待释放部分分开计算，真正的争议其实是那笔还没释放的大头。
Optimism是头部Layer2，OP既是 gas 代币也是治理权，这批空投当年承载着拉新和留住用户的承诺。如今Superchain叙事铺开，协议方更想把钱集中投到基建设施和生态补贴上，个人用户的零散奖励就显得碍事。这个逻辑站不站得住，取决于你站在哪边。
这已经不是OP第一次在激励上闹争议。上一轮回溯性空投就有人抱怨门槛太高、女巫被砍得太狠，这次直接动了待释放额度，等于把矛盾摆到了台面上。
换个角度看，这其实是整个Layer2赛道的缩影。过去两年项目方靠空投撒钱拉新，用户靠撸毛赚奖励，双方各取所需。如今TVL和活跃地址成了硬指标，单纯给用户发币的性价比越来越低，钱更想流向能带来真实使用和锁仓的基建。OP这5.4亿枚要是真收归金库，按当前流通盘算是一笔不小的弹药，足够协议方重做一套补贴逻辑。社区里有人提议折中，比如只收回长期无人认领的部分，把活跃用户的那份保住，但提案方似乎更想要整笔拿走。
更微妙的是，赞成票里不少来自协议关联地址和金库本身，散户的反对声被稀释得很厉害。当一份写在治理文档里的承诺可以被多数票推翻，用户还敢把未来奖励当真吗。你们觉得这种充公式治理，到底是项目方的理性止损，还是对早期支持者的背刺。A SOL whale that has been dormant for two years
GvHYQQLUwnb6cJpJuxBtAPPC2uWp84WxVEjC1idXgUce
Today bought 47,535 SOL, about $3.6 million
Interestingly, this guy bought 290,000 SOL around $23 in 2023, then sold a large portion around $128, making over $20 million
Then went silent for two years, now that SOL has dropped to around $75, he's back
Combined with last night's chaos, could it be... a bullish signal for SOL? $SOL NAVI上新借贷金库历史峰值破10亿
Sui 链上的 DeFi 项目 NAVI Protocol 这回整了个新活。它正式推出 NAVI Prime，是 Sui 生态头一个上线的策展人借贷金库，走的类似 Morpho 那套模式，把不同的风险资产隔开，各自配独立的风险框架和抵押品评估。官方晒的数据挺唬人：历史 TVL 峰值破 10 亿美元，累计用户超 110 万。
这东西解决的是老 DeFi 借贷的一个老毛病。传统资金池把各种风险资产混在一起，一个雷炸了容易传染全池。NAVI Prime 让每个市场像独立小房间，参数和风控各管各的，资金效率上去了，跨市场传染的风险下来了。它说自己历经四个月开发、做了三次以上独立安全审计，这周还会跟 Sui 官方联合发完整公告。
对机构资金来说，这种模块化、可隔离风险的设计才是它们敢进来的前提。过去 CeFi 借贷出过太多事，机构要的是接近传统金融风控标准的基础设施，而不是一个池子赌全场。NAVI 把门槛往机构级靠，显然是想吃那部分更稳的钱。
放到 Sui 整个生态看，NAVI 算是锁仓靠前的老牌协议了，Move 语言主打安全和并行执行，机构相对吃这套。但借贷生意的命门永远是抵押品质量和清算机制，隔离风险只是把雷分装进不同房间，房间里的雷该爆还是会爆。TVL 峰值 10 亿听着光鲜，可那是在牛市高点攒下的，市场一冷，杠杆仓位先撤，TVL 掉起来也快。对咱们来说，这种新金库上线短期可能带一波生态热度，但真要参与，得看清每个独立市场的抵押率和清算线，别被峰值数字晃了眼。
短线上，TVL 峰值 10 亿是历史高点，不代表现在还在那个水位，Sui 整体锁仓受大盘影响很大，别看到数字就上头。长线看，Move 生态能不能靠这类专业借贷工具留住真实使用，才是 NAVI 值不值得跟的关键。DeFi 的叙事早过了随便发个池子就能涨的阶段，拼的是风控硬不硬、机构认不认。
所以抛个问题给大家：这种隔离风险的专业借贷金库，能成为下一波 DeFi 回暖的真正抓手吗？还是说又是一个好看但用的人不多的新壳子。欧元稳定币两年飙四倍破4亿欧
很多人盯着美元稳定币打得火热，却没人注意欧元这边悄悄起了量。Circle 这两天放话，旗下欧元稳定币 EURC 流通量已经突破 4 亿欧元，而在 2025 年 6 月，整个欧元稳定币市场才 4 亿欧元的规模，现在光 EURC 一家就到了这个数，等于两年左右翻了好几倍。
这背后的推力是合规。EURC 按欧盟的 MiCA 法案以电子货币代币标准运营，由法国机构发行、受法国监管，储备和 Circle 自有资金完全隔离还定期审计。说白了，它踩中了欧盟最看重的那根线：合法、透明、可监管。Visa 和 Mastercard 都扩大了 EURC 的结算支持，让它从交易玩具变成了能跑跨境支付和卡结算的基建。
EURC 也不是没对手。Tether 的 EURT、以及一众欧洲本土稳币都在抢这块蛋糕，但 Circle 胜在合规身位最干净，法国牌照加定期审计，机构敢碰。从使用场景看，EURC 已经从单纯的交易对，延伸到支付、结算和机构资金管理，Bitpanda、Bitstamp、Bybit、Coinbase、Kraken 等主流所都上了 EURC 交易对，法币出入金通道也接了 Mercuryo、MoonPay、Ramp、Transak 这些服务商。地基铺得越广，后面起量越快。
宏观层面，这件事和大环境绑得紧。全球稳定币总供应约 3000 亿美元，美元稳定币一家独大，但欧元稳币已经稳居第二大类。欧元区的 M2 超过 16 万亿欧元，EURC 才占极小一块，空间看着还早。可一旦欧洲真把链上支付当成战略方向推，EURC 这种合规尖子生会最先吃到红利。
短线上，EURC 破 4 亿对加密盘面没直接刺激，它不改变 BTC 和 ETH 的方向。但它给了一个信号：稳定币的竞争正在从美元一家独大，走向多币种合规化，RWA 和链上支付的叙事会被持续喂料。
长线看，稳定币是连接法币和链上世界的桥，谁先在某个货币区里站稳合规身位，谁就握住了那个区的支付入口。对咱们做波段的人来说，这虽不是立刻能交易的标的，却是判断资金往哪条链、哪种合规框架聚集的重要风向。
所以抛个问题：美元稳币一家独大的局面，会被欧元这波合规突围慢慢撬动吗？还是说监管再严也改不了美元的底色。Wall Street is playing dead, increasing its position in the IBIT sovereign fund against the trend
The latest disclosed 13F filing laid bare the public's attitude toward Bitcoin ETFs. On one side, Abu Dhabi's two sovereign wealth funds remained silent: Mubadala held 14.72 million shares of IBIT, and the Abu Dhabi Investment Board held 8.22 million shares, neither adding nor decreasing a single share in the entire second quarter. On the other hand, Wall Street giants surged against the trend: JPMorgan Chase increased its IBIT holdings from 8.46 million shares to 10.62 million shares, a 25.5% increase, and Morgan Stanley added 16.5 million shares, up 23%.
The most outrageous case is UBS. Its actions on the options side are clearly bullish: IBIT call options jumped from 80,000 to 1.95 million, while put options dropped from 303,000 to 143,000. This asymmetric position change is more of a signal than net long/short quantity. Sovereign funds choose to stay silent most likely because Bitcoin has fallen and their holdings have shrunk, so rather than trade, it's better to lay down and hedge safely; Wall Street, on the other hand, treats ETFs as tools for bottom-fishing and allocation.
However, 13F also has blind spots: it only discloses long positions and held options, excluding short and short options. So when UBS's call options surged, we could read the bullish tendency but couldn't determine their net direction. Combined with CME leveraged funds holding 5,000 long Bitcoin futures and 12,000 short positions, with more than double the shorts, the internal divisions among big money are actually more chaotic than the surface. Retail investors tend to get caught up in Wall Street buying positions, but aside from real money, hedge positions have also been heavily invested.
The disagreements behind this are worth pondering. Sovereign funds are large and have long assessment cycles; when prices drop, they just play dead and wait for the cycle; Wall Street trading desks are more flexible, and in the midst of volatility, they actually see opportunities. Looking at overall capital flows, Bitcoin ETFs saw a net outflow of about $4.9 billion in Q2, with a net inflow of $850 million for five consecutive days in the first week of August, and another net outflow of $390 million last week, showing frequent short-term capital inflows and outflows.
In the short term, this institutional divergence provides a point for the swing market: Wall Street buying real money into ETFs often corresponds to underlying buying support, but sovereign funds remain unmoved, indicating that big money remains cautious overall. When the two are fighting, prices are more likely to wear down within a range. In the long run, ETFs are becoming the main channel for institutions to access Bitcoin; who's secretly increasing positions is more valuable than who is just shouting more.
So I want to ask: do you think Wall Street's counter-trend buying is a bottom-up purchase, or is it just taking the knife?8月12日，香港持牌稳定币发行方碇点金融科技正式推出港元稳定币 HKDAP（HKD At Par），首阶段面向机构分销商及专业投资者开放，HashKey Exchange、OSL集团成为首批认可分销商。 这意味着港元稳定币正式从监管框架、沙盒测试，走向商业化落地。 很多人期待稳定币一上线就带来资金狂潮，但现实恰恰相反： 没有散户狂欢，没有短期暴涨，却可能正在发生一场更重要的基础设施革命。 一、神话退潮，真正的价值才开始出现 过去市场对稳定币的想象非常宏大： 跨境支付秒到账、传统金融被重构、海量资金进入链上…… 但真正落地后，最先出现的往往不是“颠覆”，而是一些看起来很普通的业务流程。 这其实很正常。 1858年，第一条跨大西洋电报电缆铺设完成时，人们曾认为它能够改变世界。 它确实改变了世界，但并不是因为它立刻带来了所谓的“文明革命”，而是因为它把信息传递从几周压缩到了几分钟，最终重塑了全球金融市场的运行方式。 稳定币可能也是如此。 真正值得关注的，不是它今天有多大的市场声量，而是它能否逐渐成为数字金融的清结算基础设施。 二、第一个真实应用，可能比一场“散户狂欢”更重要 目前已经出现了一57000美金成多头屠宰线
夜里刷持仓的人估计都看到那个数字了。分析机构 Alphractal 的 Joao Wedson 把 57000 美元划成了一条线，说这是杠杆多头能不能活命的分界线。翻译成大白话就是，大饼真要跌到这个位置，一大批开了高倍多单的人会被自动平仓，连锁踩踏可能直接把价格砸穿。
这事儿最吓人的不是那个价格本身，而是现在的盘口实在太薄。Wedson 点出一个矛盾：活跃合约数量相对实际成交量高得离谱，说明一堆人挂着杠杆仓位，但真正愿意接盘的真金白银没多少。订单簿就像一层窗户纸，一旦清算单砸下来，下面根本没多少买单垫着，跌幅会比想象中狠得多。
这种薄盘口在熊市里特别危险。2022 年 LUNA 和 FTX 那两轮，大家也是看着订单簿觉得还行，结果一根大阴线下来，流动性瞬间抽干，价格半小时内跌去一两成。现在大饼虽然没到那种极端，但活跃合约和成交量严重错配，意味着一旦有人主动砸盘，下方承接会比平时稀疏得多。杠杆多头越集中，踩踏时的负反馈越凶。
回头看更冷。大饼从去年 10 月 126000 美元那个高点一路腰斩，现在晃在 6 万出头。历史上每次走熊，跌幅往往要打到 76% 到 84% 这个区间才算见底。按这套算法，眼下的位置离真正的大底可能还远，所谓的支撑更多是心理层面的。
短线上，57000 这条线就是多头的生死线。跌不破，大家还能装作没事继续震荡磨时间；真跌破，那批高倍多单先被收拾，市场情绪会瞬间从平静切到恐慌。做波段的可以把这条线当成观察锚点，价格挨近它的时候，成交量和大单方向比平时更值得盯。
长线逻辑另说，比特币的稀缺叙事和机构 ETF 通道还在，但那是另一个时间维度的事，救不了眼前这根针。眼下最需要想清楚的是，你自己开的那张多单，到底扛不扛得住这一下。
所以问题抛给还在车上的人：你的多单杠杆，抗不抗得住这一根针？还是说你早就把止损线画在了别人都看不见的地方。港银突然追着大陆老客户交代钱到底从哪来
那些前两年在香港开了投资账户的大陆人，最近可能都收到了一条有点扎心的银行推送。汇丰香港开始给一批存量的大陆投资客户发通知，要求他们在9月12日之前，通过手机银行App提交一份开户和维持账户的声明，把投资资金的来源一笔一笔交代清楚。
重点不在新开户，而在存量老客户。过去开新户已经被要求确认资金来自合法境外渠道，这次是把刀口转向了已经在户里待了许久的人。声明里要你确认投资活动用的钱来自境外合法来源，而且白纸黑字写着，银行有权在执法或监管机构请求下披露你的个人信息。
时间卡得相当死。8月20日之前没交，投资相关的服务就可能被暂停；拖到9月12日还没交，服务可能直接终止。也就是说，钱还安安静静躺在户里，但你大概率先用不了了。
汇丰的回应很官方，说这是遵循相关监管要求，邀请相关大陆投资者主动提供自证，把KYC和尽职调查里的信息更新到当前有效状态，这样才能持续不断地给客户提供服务。不过他们也强调，这轮申报只针对投资服务客户，不是所有人。
这事为什么值得咱们这圈人留意。不少大陆用户开港户，本就是冲着更自由的资产配置去的，出入金、买币、做境外投资，很多动作都走这条通道。现在银行反过来要你证明每一笔钱都干净，逻辑就整个反过来了。以前是通道敞开给你用，现在是你要向通道自证清白。
更微妙的是，新开户的口子其实早就收过一轮，这次是回头去清理老户。监管把网一点点往里收，先用一份声明试水，后面会不会有更多银行跟进，现在谁也说不准。
所以如果你或者身边人正用着港户做投资，这两条deadline值得在日历上圈一下。不是要你慌，但别等到服务被暂停了，才想起来翻那条被你一键划走、从没点开的银行推送。钱到底从哪来，现在得你自己说清楚了。融了一亿八千万的社交明星七个月内被转手两次
今天早上，Neynar 的联合创始人 Rishav Mukherji 发了一段话，说公司正在为 Farcaster、Clanker 以及 Neynar 自己的产品寻找新的归宿和运营团队，目前已经在跟好几个团队接触。
如果你对时间线没概念，我提醒一下：他们是今年 1 月 21 日才刚刚接手 Farcaster 的。到今天，正好七个月不到。
上一任是谁？是两位前 Coinbase 高管 Dan Romero 和 Varun Srinivasan，他们做了五年，公司 Merkle 累计融了大约 1.8 亿美元，光 2024 年那一笔就是 1.5 亿。当时他们退出的说法是，Farcaster 需要新的产品路线和新的领导团队，剩余的钱退还给投资人，收购价格双方都没公开。
七个月后，接盘的这一位也开始找下一个接盘的了。而且措辞很坦诚，Mukherji 直接承认没能实现年初接手时定下的目标，现有团队也不适合下一阶段。他把 Farcaster 面临的处境概括成三样东西：一个很紧密的社区、一个运营成本不低的旗舰应用、一个增长缓慢的市场。
这话听着客气，但拆开看挺扎心的。
技术上，Neynar 其实干得不差。他们把基础设施的运行成本压低了 80%，增加了分布在不同地区的验证者，还把原来只有核心团队能碰的协议代码库开放给了这些运营方，下一步甚至打算把新增验证者的决定权交给现有验证者链上投票，也就是让核心团队不再单独说了算。
成本降了 80%，路子也走对了，然后呢？然后还是撑不住。因为真正烧钱的不是协议，是那个面对普通用户的 App。
Farcaster 是混合架构，账户身份和密钥管理放在 OP Mainnet 的合约上，发帖、关注、互动这些高频数据由 Snapchain 验证者网络保存。协议这部分谁都能读、谁都能拿去做客户端，理论上很去中心化。但客户端的产品迭代、内容分发、客服、钱包、开发者接口，全都得有一家公司真金白银地养着。
最要命的数字在这儿：Romero 去年说过，每个月大概有 20 万到 30 万人打开 Farcaster 客户端，而这个数要放大 10 到 100 倍，网络才能算跑到可持续规模。Merkle 后期已经很努力了，塞进内置钱包、上代币交易、推一年 120 美元的 Pro 订阅，想用交易和订阅养住开发者和创作者。Neynar 接手后保留了钱包和交易，重心转向服务开发者。七个月，还是没找到那条能撑起整个体系的增长路径。
再往前翻一天更有意思。Farcaster 第一次易主的前一天，Lens 也把日常产品运营交给了 Mask Network，原团队转做技术顾问，理由几乎一模一样，说协议开放这步做完了，接下来的增长需要统一的应用、产品设计和分发渠道。
两个去中心化社交里最响的名字，一年之内相继把运营交出去，一个已经交了第二次。
现在还有一堆没落地的事：接手方到底是谁，协议、客户端、Clanker 和开发者业务是整体转让还是拆着卖，账面上那笔要返还的资金给谁、给多少、什么时候给，验证者链上投票哪天真正启用。Mukherji 只说了应用和开发者产品暂时照常跑，对用户没有直接影响，团队成员后面会转向新项目。
我一直在想一个问题。我们过去几年讲开放协议的故事，讲得最动人的部分是把身份和社交关系从平台手里拿回来。这件事 Farcaster 确实做到了，账户还在链上，数据还在验证者网络里，谁都能读。
但谁来付客服的工资，谁来做那个让普通人愿意每天点开的 App，谁来承担增长本身的成本，这个问题从来没有被协议解决过。协议可以去中心化，运营似乎永远需要一家公司。
所以你觉得，Farcaster 的问题是团队不行，还是这套分工本身就有个绕不过去的窟窿？Your frequently used Ethereum wallet may quietly become invalid next week
The Ethereum Foundation itself issued a warning, clearly saying: the upcoming Glamsterdam upgrade may cause some wallets, block explorers, and gas estimation tools to stop functioning properly.
The subtle part is that the Ethereum Foundation itself is issuing the warning. A team building their own house warns in advance that certain doors might not open. This is rare elsewhere, but in the Ethereum ecosystem, it's an old tradition.
The problem lies in the Gas model. This upgrade introduces a new dimension called State Gas, which specifically charges separately for the operation of creating new blockchain states. The result is straightforward: transferring ETH to an account with an existing balance still costs 21,000 Gas; but once you transfer to a new address that has never received money, you have to pay an extra State Gas fee.
The foundation made it clear: any software that writes "all ETH transfers only 21,000 Gas" in its code will fall victim this time. Wallet fee estimates will be miscalculated, indexers may miss transactions, and outdated transfer logic will directly display error messages.
What's even more troublesome is that ordinary users have no idea what logic is behind the wallet they use. You think that just clicking send is enough, but if the estimation code behind it hasn't been updated, the fees you see might be wrong, and the transaction could get stuck halfway.
This isn't the first time Ethereum has quietly changed its rules in an upgrade. Every time a major version is pushed forward, a batch of gadgets that can't keep up with the pace are left behind. For developers, it's just a few lines of code adjustments; for ordinary users, it could be a baffling transfer failure.
The foundation's solution is to run the Plataberget testnet ahead of time. But the testnet only goes live on August 13, and Glamsterdam will be activated on mainnet this Thursday. The developer window is actually very short, and the Sepolia and Hoodi testnets will have to wait a bit longer.
Ultimately, this isn't a shocking vulnerability, but rather the cost of having to change the underlying layer for expansion. EIP-8037 bills state creation operations separately. In the long run, this can alleviate state bloat, but in the short term, wallets and users who don't keep up with the pace will have to suffer the bitter fruit of incompatibility.
Has the wallet you recently used been updated?债市在尖叫美联储9月还敢降息吗
钱包里的币还没回血，债市先炸了。30年期美债收益率冲到2007年以来最高，法国借贷成本创2008年新高，德国逼近2011年，英国国债收益率都快摸到6%。全世界的长债一起被甩卖，这场景上一次见还是金融危机前后，区别是这回没有明显的危机引爆点，更像钱自己用脚投票。
财联社这边的标题很直白：美联储9月加息风险并未解除。市场本来赌着9月要降息，可债券交易员用脚投了相反票。收益率飙成这样，等于在说通胀没死透、政府还在乱花钱，降息空间被一点点挤没了。债市这把火，烧的是对财政纪律和通胀反复的担心。
对加密圈意味着什么。BTC今年跌了快27%，一直趴在6.4万附近窄幅晃。低波动背后，是流动性没进来。美债收益率这么高，钱宁愿躺在无风险利率里吃票息，也不进高风险资产。每次收益率跳一跳，BTC的向上突破就远一步，这就是资金用脚投票的结果，跟币圈自己讲的故事没关系。
还有个细节。日本30年期国债收益率也冲到4%以上，全球长债同步走熊，说明这不是美国一家的事，是市场在对各国政府的债务规模重新定价。这种环境下，所有风险资产都被同一个逻辑压着：无风险收益太高，谁还来接高风险的盘。
短看，债市恐慌只要不缓和，风险资产就难有像样行情，BTC大概率继续磨。长看，真要降息预期重新点火，被压久了的加密资产弹性会比股票大。
把话说直白点，现在币圈最需要的不是利好，是钱。钱被美债吸走了，BTC就只能原地踏步。什么时候看到30年美债收益率从高位回落、ETF重新净流入，那才是行情真正要动的前兆。币圈常说自己是避险资产，可真到了全球流动性收紧的时候，它跌得比谁都快，这层遮羞布早该撕了。但眼下，盯美债收益率比盯K线更管用。你觉得这轮债灾，是恐慌过头还是刚刚开始。北京时间8月18日，现货黄金突破4430美元/盎司，日内涨0.4%，纽约期金同步站上4490美元。布油站上91美元，WTI站上84美元，金银油集体上涨。 金价涨，期权市场的信号更值得看。 量化交易公司Susquehanna指出，黄金期权市场正从“下行保护”转向“押注进一步上涨”。期权偏度已发生“实质性转变”——看跌期权需求下降，看涨期权需求上升，逆转了夏季早些时候的格局。一笔典型交易：投资者以约5.55美元的价格买入8000张SPDR Gold Trust 11月到期、行权价460美元的看涨期权，该ETF周一收于405.49美元。 同时，黄金基金录得今年1月以来最强资金流入。 为什么涨？三重驱动。 第一，美元信用持续受损。 东吴证券首席经济学家芦哲指出，特朗普对美联储的干预、美债突破40万亿、去美元化交易持续——前期压制黄金的因素正在出尽，新的美元信用风险溢价重新显现。 第二，地缘风险没消停。 特朗普刚警告“不排除轰炸”，油价站上91美元。 第三，技术面突破触发空头回补。 黄金重新站上50日均线，突破6月中旬以来前高阻力，触发程序化交易空头回补。 机构怎么看？ 伦敦金银市场协会调查显示稳定币终归要进主权笼子里
你的USDT躺在钱包里，真以为它只归代码管吗。跨境支付公司Airwallex的创始人Jack Zhang最近泼了盆冷水，说稳定币会成为全球资金流动的重要轨道，但主权金融体系才决定怎么把它装进笼子。技术能跑遍全世界，可规矩还是各国自己定，这话说得一点不客气。
他举了一串例子。巴西央行已经把虚拟资产服务塞进外汇和资本监管框架。越南一边禁加密资产当支付手段，一边在搭受监管的加密市场。菲律宾直接无限期停了虚拟资产服务商牌照。三个国家三种姿态，但内核一样：稳定币可以跑，但得在我的地盘、按我的法跑，别想绕过监管自己飞。
这对咱们持币的人意味什么。USDT、USDC再大，发行方终究要跟各国监管谈条件。最稳的那批基础设施，不是最自由的，而是能接住主权金融体系、还能在法币和稳定币之间稳稳换汇的。说白了，合规通道才是护城河，野路子随时可能被一刀切，今天能用明天可能就进不去。
Airwallex自己也在下注这条路，它在巴西拿了央行支付牌照，在墨西哥收了本地支付牌照，把企业账户和拉美金融体系接起来。这动作说明大玩家押的不是去中心化神话，而是接住监管后的真实生意。稳定币越被主流用，越得按主流的规矩办。对USDT这种靠离岸撑起来的，压力只会越来越大，因为主权体系要的是看得见的牌照和备付金，而不是一句代码即正义。
短期看，各国监管落地越清晰，稳定币越不像野孩子，越能进机构钱包，流动性也会更踏实。长期看，这玩意儿会变成受监管的金融管道，而不是什么去中心化乌托邦。
对咱们小散来说，这反而是好事，起码跑路和暴雷的概率被压低了。但代价是，稳定币会越来越像银行体系的小弟，哪天某个主权说不准用，它就真的不好使。持仓稳定币的朋友，别光看收益，也看看它背后哪个主权在兜底。你觉得哪家的稳定币，最后能笑到最后。美银砍掉七成MSTR猛买ETHA
这批机构的操作越来越看不懂。美国银行刚披露的二季度持仓里，MSTR被砍掉七成，直接减到只剩零头。可另一边，ETHA的持仓暴增29倍，从几乎可以忽略变成重仓。一手卖一手买，动作干脆得像换赛道，跟咱们散户追涨杀跌的纠结完全两个画风。
数据摆在这。BoA一季度还捏着不少MSTR，二季度直接砍了70%，摆明不看好Strategy那套借钱买币的游戏。可ETHA是贝莱德的以太坊现货ETF，29倍增持说明它真金白银往ETH里冲。卖比特币财库股、买以太坊ETF，这转弯比散户还利索，也顺手给市场扔了个信号：大行的偏好在挪窝。
为啥这么割裂。MSTR今年跌了快四成，靠发优先股和债撑着买币，市场开始怀疑这杠杆还能玩多久。ETHA不一样，它是现货ETF，底层是实打实的ETH，没有公司杠杆的叙事包袱。大行用脚投票，把赌注从财库公司挪到原生资产，本质上是在给两种资产定不同的风险溢价，财库股的信仰明显打折了。
这背后还有条暗线。贝莱德自家ETH产品上线以来，机构申赎一直比散户积极，说明传统钱袋子对ETH的兴趣在抬升，只是平时不声不响。美银这29倍加仓，更像把一条本来小众的线索公开化了。不是一天两天的冲动，是季度级别的仓位调整。
盘面参考很清楚。机构资金在比特币和以太坊之间做再平衡，偏向ETH现货敞口。对咱们来说，ETH的相对强弱值得盯，尤其是ETF持续净流入的那几天，往往就是资金态度最诚实的时候。
也有人解读成美银在押注ETH现货ETF会比财库股更抗跌，毕竟底层资产透明、没有公司杠杆的暗雷。但换个角度，砍MSTR也可能是单纯止损，不见得是长期看多ETH。大行调仓的理由永远比表面复杂，咱们看方向就好，别当成无脑跟单的信号。短看ETH对BTC的汇率能不能稳住，长看机构把ETH当配置资产这条线会不会越走越宽。你觉得美银这波换仓，是嗅到了什么咱们还没看见的信号。Nearly 1.8 billion yuan in short positions were just one point away from collapse
Did your account turn red this week? BTC is hovering around $64,076, but there's a group of people more anxious than anyone. Two addresses, 0x8c96 and 0x431f, have been shorting simultaneously in the past, now holding 2,800 BTC short positions totaling about $179 million, making them the largest BTC positions on Hyperliquid. These two are not ordinary retail investors, but top accounts copied by a bunch of people, following the trend with every move.
These two were playing with a 40x cross-margin strategy. 0x8c96 short 1,800 coins, liquidation price $64,855. 0x431f short 1,000 coins, liquidation price $65,097. At the current $64,076 price, the two short positions are only 1.2% and 1.6% away from liquidation—a thin layer of paper. If BTC pushes up a bit more, the $200 million bet would instantly evaporate. 40x leverage means that for every 1% price move, the position bears 40% profit and loss. Normally it's fun, but near liquidation, it's really dangerous.
Interestingly, these two were originally copy-based traders—one placed an order and a bunch of people copied it. But now the team has broken up. 0x8c96 is still increasing their position, pushing their short positions to 1,800 coins. But 0x431f chickened out first, stopping out 206 shares and losing $80,000. The main account desperately increased risk, while the copy account secretly reduced positions. The scene was so realistic, it looked just like a group of people shouting to rush in, only to be led by the big boss to exit first. Even more impressive, they had already withdrawn their stop-loss orders near $63,971, leaving only two take-profit lines at 58,123 and 58,423, clearly betting on further declines.
The market impact is very direct. The 179 million short positions hover in the range of 64,076 to 65,097, less than 2%, essentially holding a cache of dynamite that could ignite at any moment. Once BTC breaks above 65,100, these short positions are forced to close, and buyers push themselves higher—this is what we commonly call a short squeeze. Conversely, if it falls below 58,000, their take-profit orders will be realized, indicating the bears have temporarily won. The space for both sides is small, and a market change could happen at any time.
This is where the logic of short-term short and long-term price is laid out. In a low-volatility environment, such high-leverage concentrated positions are the market's most vulnerable fuse. Ordinary players should definitely avoid such multiple-sized orders—one insertion and it goes to zero. Don't compete with the trend; focusing on the liquidation-dense zone of 65,100 is more effective than blindly guessing the direction.
This kind of copying squad isn't the first to cause trouble. When two addresses simultaneously increased their positions to 169 million, people were already watching. Now, increasing to 179 million is causing internal discord, indicating that the high-leverage camp itself lacks consensus. When long and short positions are locked in a stalemate, these concentrated positions often become the powder keg of the market. What do you think—will these 180 million short positions be blown up first or run out first?The person who called you to chase the meme opened his own shop
Ansem did something quite interesting these past couple of days. This KOL, who built up a huge loyal fan base by calling out meme coins during the last bear market, suddenly announced the launch of his own token issuance platform, Ansem.io. To put it simply, he used to be the one who stood in front and called everyone to charge, but now he's set up the stage. In the future, anyone wanting to issue tokens has to register with him first. Back then, he became a legend by calling out for orders with just a few meme coins, and fans trusted him more than he did.
This platform is built on Pump.fun; the project team creates tokens on the Ansem.io and can trade directly on the Solana mainnet on the first day, with the process no different from a regular pump gameplay. It sounds like they're giving the community a more convenient tool, but there's a rather intriguing detail inside. For each new coin, a portion of the funds used for the initial purchase is automatically allocated for a community airdrop, and for better exposure, the project team must permanently burn a certain amount of ANSEM tokens to unlock Gold and Diamond levels. People usually complain that the pump is being exploited, but now there's a new KOL fee station in front.
In other words, if a project wants to be seen, it must first burn money into Ansem's own coin. The platform also introduced the Z500 index and a boost mechanism to rank community tokens in real time—whoever wants to climb the rankings spends money to boost them. It's a bit like turning attention into a business with clear pricing, with price tags clearly displayed, and whoever pays rises.
We're actually quite familiar with KOL posting platforms. They often talk about empowering communities and lowering barriers, but the underlying logic is a closed-loop traffic monetization system. Ansem himself started out with meme storytelling; he knows better than anyone how much attention is worth and knows that people are willing to pay for stories that think they can get rich quickly. Now he has productized this mindset: the platform is charging for revenue and charging for exposure, not missing out on both. Ultimately, the traffic is in his hands, and he sets the rules, which is much more stable than just shouting loudly.
Interestingly, the coin ultimately runs on the Pump.fun; the underlying zero-sum game hasn't changed at all; the only change is who collects this exposure fee. Those who followed him through the meme scene will continue to play on the platform he set up or simply watch from a different perspective. What do you think: does this approach of a KOL personally running a platform give retail investors an extra path, or does it sharpen the sickle sharper? Some say it's the most elegant way to monetize influence, while others think it's just a different way to collect rent.After the big bing is sluggish to the extreme, who will be washed out first in this wave of fluctuations?
Over the past few weeks, people who opened market software to watch Bitcoin probably felt sleepy. The price kept fluctuating around $64,000, with daily fluctuations of less than one point. There were fewer and fewer discussions on forums, and even open interest in the futures market was quietly shrinking.
This quiet is not without reason. Sean Farrell, Head of Digital Asset Strategy at Fundstrat, recently reviewed eight similar low-volatility phases from the past, and the conclusion is somewhat alarming. In the following sixty days, the absolute median price fluctuation of Bitcoin reached 30.2%, with four rises and four falls—a completely evenly split direction. The model can tell you the wind will be strong, but it itself doesn't know which direction it will blow.
He also mentioned that the recent rally mainly drives short closing rather than real new buying interest. Since last Friday night, Bitcoin-denominated futures open interest has dropped by about 8%, indicating that many short sellers are actively withdrawing their platforms. Without new buying support, the foundation for this rebound is actually quite weak.
Farrell is focusing on the 30-day volatility this time, which measures short-term price movements. When this value hits a historic low, it often means the market has entered a stalemate where no one wants to act first. But stalemates never last forever; in the past eight times, not once has it stayed this low.
Looking at the state of on-chain and community levels, low volatility is the easiest time to let your guard down. Many people have withdrawn their stop-losses and added leverage, thinking that a long-term sideways move is inevitable, so they should just place a position and wait. But historically, those thirty-point swings often suddenly fall or rally when everyone thinks nothing is wrong, leaving only a few hours to react.
What's even more worth pondering is the external variable he pointed out: real bond yields. Farrell believes that if real yields keep climbing, they could very well be the needle that breaks the current low-volatility slump. Currently, US Treasury yields are peaking, rate cut expectations are repeatedly bouncing around, and it's only getting harder for Bitcoin to keep playing asleep. So far this year, Bitcoin has dropped nearly 27%. The market is waiting for a direction, but no one can give an answer.
What we fear most is never a drop, but this annoying silence. When everyone is guessing which side to kick, the real risk might be that you weren't prepared at all. Who do you think this thirty-point fluctuation will wash out first?Even Ethereum, which claims to be permanently booked, will be saving less
On August 18, Ethereum developer kevaundray posted a proposal EIP-12188 on GitHub, suggesting shortening the retention window for consensus layer blocks. Simply put, it means nodes no longer have to hold onto historical blocks for so many years, saving a lot of storage.
When this idea came up, there wasn't much arguing; instead, it gained support from several client developers. Lighthouse contributor Michael Elsproul said that after the changes, it didn't significantly affect his own operations; Dapplion also nodded, feeling the direction was right. It seems like a low-key burden reduction by the tech geeks.
But if you think about it carefully, it gets interesting. The old saying we usually hear is that the hardest advantage of blockchain is that the ledger is permanently stored, no one can change it, and history is always there. But now, Ethereum's core developers are seriously discussing preserving history for a while, with a simple reason: nodes are about to collapse.
The contradiction raised by the proposal is very realistic. As the chain grows longer, the storage costs for full nodes keep rising. Ordinary people and small teams are struggling more and more to run nodes. At this rate, the network will gradually concentrate in the hands of a few players with money to maintain it. Shortening the retention window is like loosening node restrictions, allowing more people to run.
In fact, discussions about clipping historical data have been ongoing on the execution side; this time it is just extending similar ideas to the consensus layer. Looking at both sides together, the direction of Ethereum's burden reduction is already very clear.
Some might think, 'I don't run nodes, so it has nothing to do with me.' But in reality, block explorers, wallets, and the RPC services you use all rely on someone to run nodes. If node running becomes increasingly concentrated in the hands of a few big institutions, then who can see what data and get old records, the power will be in the hands of a few.
There are costs involved. Some in the discussion mentioned that the execution layer's historical data clipping might affect long-running nodes from providing old blocks externally. In other words, if you want to dig up an old transaction from a few years ago, you might not always be able to get it from a node on the spot.
Even more subtle is trust. Many people put their coins on-chain for something that no one can change and can always be traced. Once old data is no longer accessible to everyone, that sense of security quietly diminishes. Technically, it's fine, but psychologically, that hurdle isn't so easy to cross.
Of course, this proposal currently only moves the consensus layer; the historical trimming of the execution layer is a longer, separate discussion. The developers have also made it clear that changes do not threaten network security nor affect node normal operation. It may not sound that scary, but the direction it points is worth watching: when Ethereum starts calculating the economic ledger, the word 'permanent' is no longer as absolute as before.
Do you think the cheaper node running or the ledger always exists should come first?去杠杆进行中链上借贷缩两成
链上的钱正在悄悄退潮。Galaxy Research 刚出的二季度报告显示，加密抵押借贷总额环比下降了 16.8%，掉到 561.6 亿美元，比 2025 年三季度的高点 786.9 亿整整少了四成。DeFi 借贷这块跌得更狠，环比收缩 27.61% 到 204.3 亿，这是 2023 年三季度以来 CeFi 规模第一次反超 DeFi。
最值得玩味的是节奏。这轮去杠杆不是 2022 年那种一个季度崩掉五成多的断崖，而是连续三个季度分别只降 10%、5% 和 17%，用报告里的话说叫有序而温和。Tether 还是 CeFi 老大，占了 58.54% 的份额，稳得有点吓人。贷方这边没出现恐慌式抽贷，借款方也大多在主动降敞口，市场比上一轮冷静太多。
期货市场那边，二季度未平仓合约微降 3.08% 到 1032 亿，但七月已经弹回约 1140 亿。说明杠杆不是没了，只是从借贷市场挪到了合约，热度还在换地方烧。机构企业债这块，Strategy 五月刚完成 15 亿美元债务回购，把整个 DAT 行业的未偿债务拉低到 161 亿，也在有意压杠杆。
这份报告还有个细节容易被忽略。DeFi 借贷跌得比整体狠，说明去杠杆主要集中在链上原生资金，传统机构的 CeFi 反而稳住了。换句话说，撤的是最敏感的热钱，留下的是相对长期的钱。这种结构比全面收缩要健康，但也意味着小币的链上弹药确实在变少，拉盘越来越靠真需求而不是借钱堆。
短期看，借贷收缩意味着链上可加的杠杆变少，小币想拉盘缺了资金燃料，盘面更容易被现货主导。长期看，这种阶梯式缓降反而是健康信号，没出现恐慌清算，说明持币结构比上一轮扎实。真正的雷是突然的强平潮，现在还没看到，但利率一往上顶就得盯紧。
说到底去杠杆不可怕，可怕的是假装没发生。链上杠杆在退，意味着靠资金堆出来的上涨越来越难，接下来拼的是真需求和真流动性。谁还抱着上一轮猛加杠杆的打法，这轮容易第一个被洗出去。
你们觉得这轮去杠杆，是会缓缓落地，还是哪天冷不丁来一记狠的。赚过百万的巨鲸回头加仓GPS了
有个钱包最近又出现在大家视野里。地址 0x8b6e 这位巨鲸，之前靠做多 AKE 一把赚了超过 112 万美元，算是这轮里少有的赢家。结果他没歇着，转头又在 Aster DEX 上开了 GPS 的多单，两倍的杠杆，买进去 2723 万枚 GPS，仓位价值大概 45 万美元。
这笔单现在浮盈约 8 万美元，涨幅 35.66%。更夸张的是 GPS 这币近期已经涨了大约 70%，巨鲸进场的位置不算低，却还在加。一边是已经落袋的百万利润，一边是继续追高的新仓位，这人的打法就是赢家通吃、不赢不撤，和那些亏了就死扛的散户完全两路人。
盘面怎么看。一个能精准逃顶 AKE 的地址，回头押注 GPS，本身就是一个信号。山寨里这种局部热点一直在轮动，MEME 和中小市值币的弹性比大饼猛得多，但也更容易一波流。跟着巨鲸屁股走风险极高，他浮盈八个达不溜，你追进去可能正好接在情绪顶，毕竟他建仓成本比你低一大截。
短期看 GPS 这种强势山寨还有惯性，量价没走坏前可能继续冲。长期看中小市值币的行情最怕流动性一断就塌，一旦大饼回调，最先被扔的就是这种高贝塔小币。做趋势的别被单笔暴利带节奏，人家是提前埋伏加严格止损，散户往往是看到涨了才冲，节奏从头就错了。
这单还有个细节值得玩味。0x8b6e 用的是两倍杠杆而不是他惯常的高倍，说明这人也在控风险，不是无脑上头。能在 AKE 上赚百万又懂得降杠杆，这种选手比那些满仓梭哈的散户可怕得多。山寨的暴利故事里，活下来的从来都是既敢冲又懂收的人，光看浮盈数字容易忽略背后的纪律。
说到底巨鲸也不是永远对。他能逃顶 AKE 是本事，但 GPS 这波追高能不能复制之前的运气，没人打包票。咱们普通玩家没有他的信息差和止损纪律，硬抄作业大概率抄在半山腰。看戏可以，上手要算清自己扛不扛得住。
你们觉得这位巨鲸这波 GPS，是又押对了还是赶着去接最后一棒。The Bank of Japan is planning to raise interest rates every two months
A piece of news that's quite far removed from the crypto world could be even more disruptive than you think. DBS Group just moved forward its expectations for the Bank of Japan's rate hikes, saying the pace might accelerate from the current rate every six months to every three to four months, and the next rate hike could even be moved up to September, with the overnight rate reaching 1.75% by mid-2027.
What does this have to do with crypto? It's a big deal. In the past two years, many institutions have been borrowing cheap yen to buy BTC and US stocks—in other words, carry trades. When the yen raises rates, borrowing yen goes up, so you have to calculate whether these positions are worthwhile. If you really want to concentrate your positions, you'll be the first to dump on risk assets. BTC is the most sensitive to global liquidity; when the yen tightens, it tends to shake along with it. Historically, the 2024 round of yen carry unwinding triggered massive global asset volatility.
How should we interpret the market? Short-term negative news is that rising yen rate hike expectations have further dampened already fragile risk appetite. But Japan's economic data is actually improving, and DBS has revised Japan's GDP forecasts upward for this year and next, indicating that rate hikes mean the economy can hold up, not force it. The market now gives about a 75% chance of a rate hike in September, with the October hike almost entirely priced in.
Many people think that yen rate hikes are far from them, but in reality, there is a barrier. Behind every BTC in your account, there may be cheap yen borrowed by institutions. Once carry trades are closed out in concentrated trading, the first to be sold are these highly volatile assets, which have nothing to do with the quality of the crypto world itself. This transmission chain is usually invisible; once triggered, it happens quickly, and when news comes out, it becomes slower.
In the long run, before the global interest rate center shifts downward, it will be quite difficult for crypto to enter an independent bull market. What really needs to be watched is the September rate meeting; whether or not to raise and how much will directly affect the retention of carry funds. The line between exchange rate and coin price is tighter than most people think. Don't wait until the yen drops to realize your position is still hanging there.
Ultimately, the exchange rate line is more important to watch than the candlestick. Many people only look at BTC charts when trading coins, unaware that there is a layer of yen leverage behind their positions. If the USD/JPY suddenly plunges, risk assets will be hammered first, and by then, it will be too late to find the cause.
Have you considered the impact of yen rate hikes in your positions?IREN’s first 50MW Horizon delivery is more than a construction milestone. It gives investors an initial proof point for the company’s shift from bitcoin mining toward AI cloud infrastructure under its five-year Microsoft contract, valued at about $9.7B.
The key question now is repeatability: three Horizon sites remain, within a program totaling roughly 200MW in Childress, Texas. My read is that valuation should increasingly reflect execution across the full rollout, not simply the scarcity value of power and land. Not advice, just analysis.
#IRENDeliversForMSFTThe fresh and enthusiastic financial report has been released
Two words: weak
Xiaomi sells more expensive, cars sell more, but profits from its core business are thinning.
The market-anticipated turning point in automotive profitability has not appeared; instead, the simultaneous decline in smartphone and automotive gross margins has stolen the spotlight.
Detailed data:
· Revenue was 108.9 billion yuan, down 6.1% year-on-year, below the lower end of the previously publicly expected range.
· Adjusted net profit was 6.2 billion yuan, down 42.6% year-on-year; Paper net profit was 9.5 billion yuan, mainly supported by 6.5 billion yuan in fair value investment income and subsidy income.
· Mobile phone shipments plunged 26.5%. Although ASP rose 25.9%, gross margin still fell from 11.5% to 8.5%, indicating that price increases have not yet addressed storage cost pressures.
· Automobile deliveries grew 28.2% to 104,200 units, but gross margin fell from 26.4% to 19.2%, with an operating loss of 2.6 billion yuan. The loss narrowed from 3.1 billion yuan in the first quarter, but still fell far short of 300 million yuan in the same period last year.
· IoT revenue fell by 19.2%, with internet services being one of the few stable sectors.
#财报观察员: Xiaomi is about to release its financial report. Which business line do you think is more optimistic? 加密ETF周抛九千万月却吸了九亿
账户这周绿了吗。别急着盯盘，先看一组让人犯晕的数字。加密 ETF 上周净流出约 9055 万美元，听着像退潮，可把时间拉长到一个月，竟然又净流入 9.77 亿美元，再拉到一整个季度，净流出直接干到 54.06 亿美元。同一个池子，周抛月吸季逃，三种口径三个方向，这水到底在进还是在出。
把三档数据摆一起看更清楚。周度净流出 9055 万，月度净流入 9.77 亿，季度净流出 54.06 亿。正负交替说明机构内部也分裂，有人借反弹减仓，有人逢低接。回看前几个月，加密 ETF 其实经历过连续多周净流入，这波周度转负更像获利了结，不是趋势反转，但季度层面 54 亿的净流出摆在那，长线资金的态度没那么坚定。
钱为什么走得这么别扭。根子不在币圈，在美国国债。美国 30 年期国债收益率这两天冲到 5.31%，是 2007 年以来的最高点。长端利率一抬头，机构那边的资金成本就跟着涨，投加密 ETF 的机会成本也变高了。加上美伊那档子事又闹起来，霍尔木兹随时可能再出幺蛾子，长钱就更不敢往高风险资产里塞。
更麻烦的是，日本英国中国三国六月一起减了美债，海外需求也在撤，长债供给又停不下来，收益率想掉下来不容易。BTC 这种高波动资产，对美元流动性和长端收益率最敏感。ETF 周度净流出说明短线资金在撤，但月度还能转正，说明中长期钱并没彻底走。
这种分歧期，价格最容易上不去也下不深，横着磨。咱们做趋势的，别被单周数据吓去砍仓，也别看月线红了就上头追。短期利空是利率还在顶，长期逻辑没坏，机构配置加密的口子撕开了就不会轻易合上。真正要盯的是 30 年美债能不能压回 5% 以下，压不回去 BTC 想突破就缺一口气。
说到底 ETF 是币圈和机构流动性之间的水管。水管一边连着真金白银，一边连着 BTC 的价格。现在水管在抖，不是完全关死也没全开。最该做的就是把仓位管理好，别在分歧期赌方向，等水管重新稳定再说。
你们觉得这波 ETF 月线净流入，是真金白银在进场，还是只是跌多了的技术性回补。你的下单权正被券商交给AI代理
Robinhood这周在X上发了一条不太起眼的公告，说自家的Agentic Trading加密交易功能已经开始向部分用户开放了。简单说，就是你可以把一个AI代理接到账户上，让它自己决定买什么卖什么，股票、期权、加密资产都能碰，而且只能用你存进去的那笔钱。它走的是MCP服务器那套标准接口，目前只向符合条件的用户逐步放开，不是人人能开。
Robinhood给这个代理开了独立的子账户，交易标的限定在加密、股票和期权，还写死只能用账户里的本金，不能借钱加杠杆。等于给AI套了缰绳，可缰绳的长度由平台说了算。这事放在半年前还像科幻片，现在老牌券商都在往同一条路上冲，把交易的决定权从人手里挪给一段代码。
Coinbase那边更早就在推AiFi，让AI代理自己研究、决策、下单、互付。两边打的是一个主意，以后操盘的未必是人。这些代理的决策全靠模型吃进去的新闻、行情和社群情绪，信息源干不干净，直接决定它下一秒下什么单。可模型自己不会告诉你它为什么这么想，黑箱这头永远开着。
Robinhood这次留了口子，听起来像是怕AI把人亏穿底。可真让一段程序替你下单，最吊诡的地方就在这，你以为自己在控制风险，其实把方向盘交出去那刻，亏和赚都不再由你说了算。
留口子归留口子，真正的风险不在亏多少，而在你渐渐看不懂它在干嘛。程序下单以毫秒计，一天能转几十个回合，等你想拦的时候单子早成交了。一边是券商拼命把AI代理推到台前，另一边监管还没跟上，代理下了错单责任算谁的，现在没人说得清。
更值得琢磨的是节奏。大饼最近在六万四附近晃，ETF上周还在净流出，市场情绪并不算热，偏偏这些券商挑在这个时候把AI代理推向用户。是真觉得技术成熟了，还是想在平淡行情里找新故事讲，咱们外人看不准。
我比较好奇的是另一头。你愿意把自己账户的下单权交给一个AI代理吗。它不会恐慌也不会贪，但万一它读错信号、或者被人喂了坏数据，亏的可是你的真金白银。等哪天身边朋友说他的账户是AI在管，你先别羡慕，问问他自己还看得懂那些单子吗。Most people feel the current Bitcoin market is: bored to the point of sleep. The fluctuations were like an ECG stopping, lingering around $64,000 every day. But if you lift the rug and look at the data, you'll find this is not dormant at all, but an extremely brutal chip plunder.
Long-term holders (LTH) hold 16.35 million Bitcoins. Many people might not have a clear idea of this number, so let's put it this way: the total supply of Bitcoin is only 21 million. After subtracting the millions lost and the unmined coins, the remaining cash in the market has almost been drained by these veterans.
Even more impressive, 1.38 million coins were purchased in the past 90 days. This shows that while everyone was shouting that the bull market was over and about to end, these old foxes were driving bulldozers to buy up goods inside the market. In the past 15 days, only 2 days have been reduced; this is not trading at all, it's about hoarding gold nuggets. This extremely low selling pressure means that as long as even a tiny bit of incremental capital comes in, prices will skyrocket like a skyrocket due to lack of stock on the market.
The current price is $64,200, and the average holding cost for veterans is $49,400.
On paper, I made 30%. If it were an ordinary retail investor, they might have already cashed out and gone for a hotpot meal, but for these LTH that has weathered countless winters, 30% profit isn't even enough to fill the gaps.
This 1.30 multiplier is very interesting: it serves as a "safety cushion," but also as a "psychological game point." For the big players, it's not enoughShentong's 10-billion-yuan position is being sold off at a 20% discount
The young man who wrote the AI prophecy article that went viral online has recently been the first to give up.
According to the Science and Technology Innovation Board Daily, the hedge fund Situational Awareness, founded by former OpenAI researcher and AI prodigy Leopold Aschenbrenner, is selling its approximately $5 billion worth of Anthropic shares at a 20% discount. This fund recently faced a liquidation crisis due to high leverage and was forced to sell at low prices in a hurry to recover cash. The fund's name is Situational Awareness, which shares the same name as Leopold's breakout work, showing his strong belief in the AI trend.
Looking at this incident in detail, the contrast is striking. Leopold is only in his twenties. Last year, his long article went viral in Silicon Valley, with the core argument that the AI wave would reshape everything, and anyone who didn't bet enough would be left behind. But his own fund ended up falling precisely because of the act of using leveraged bets themselves. Those who once wrote articles urging the world to heavily invest in AI now want to sell their core AI equity at a discount.
What's even more subtle is the background of the acquisition. Anthropic itself is not short of stories; its annualized revenue is reportedly over $65 billion, with valuation rumors approaching $2 trillion, and it could ring the alarm on Wall Street as early as this fall. On one side, the target company is being hyped to the sky; on the other, its early institutional shareholders are fleeing at discounts. This misalignment itself is worth pondering. A 20% discount is no small amount; cutting 20% of the $5 billion chip would mean 1 billion evaporated into thin air.
Actually, the signal isn't just this. The Bank for International Settlements just named the AI bubble in its annual report, saying core companies' valuations are high and risk premiums are being suppressed to a minimum. It also highlights the cyclical financing between chip manufacturers and cloud providers through equity investment, long-term procurement, and computing power leasing, which hides the risk of repeated pledges. In other words, the more lively the books, the more fragile the real leverage may be.
Interestingly, even the veteran quantitative giant Jane Street lost about $15 billion in July, partly attributed to the AI fund's holdings. A young person's leverage can affect the bills of a Wall Street veteran, indicating that money on the AI chain has long been tangled.
So here's the question. When those who understand AI best are being forced to reduce their positions, are we witnessing a normal liquidity squeeze, or is it the first crack in the AI narrative? Do you think this wave of price clearance is the starting point of panic, or just a few players losing control?$XIAOMI 小米刚放出的上半年数据看得出来压力不小。整个上半年总营收2080.63亿元，同比跌了8.4%；经调整净利润122.91亿，直接同比下滑42.8%，利润下滑幅度远大于营收的下滑。
拆分看二季度单季，营收1089亿，去年同期是1159.6亿，同比下降6.1%。能看出来，营收收缩幅度相比上半年整体有所收窄，但利润端掉得很厉害。
简单大白话讲：卖出去的整体收入只是小幅缩水，但赚钱能力被明显挤压。
大概率几方面原因：手机行业竞争内卷严重，硬件利润被压缩；汽车业务还处在持续投入烧钱阶段，研发、建厂各项开支持续吃掉利润；海外市场环境波动，也拖累整体收益。
二季度营收降幅比上半年整体要好，说明业务并没有在持续加速恶化，有一定企稳迹象，但利润是最大痛点。
往后接下来要看两点：一是小米汽车的销量能不能跑起来，能不能慢慢从烧钱转向贡献收益；二是手机端能不能稳住出货，改善硬件毛利率。如果汽车放量不及预期，持续大额投入，后续利润依旧会承压。
#财报观察员：小米即将发布财报，你更看好哪条业务线？
#30年期美债收益率创2007年以来新高
#闪迪收涨逾8%，长期协议受关注 基本面研报 $ASML / ASML Holding（NASDAQ·光刻机） $1.9K（24h +2.12%）
概括一下：ASML Holding（$ASML）综合评分 57/100，评级 叙事重于落地。 业务基本面以 外部付费 为主， 市值收入倍数 尚在合理区间。
先看公司：ASML Holding（$ASML）在 NASDAQ 上市，光刻机 赛道。 白话讲：EUV光刻机垄断。 对标 TSM、NVDA。 业务增长靠订单交付和市场份额扩张，核心看营收增速和毛利率是否匹配资本开支强度。 宏观利率和行业景气度决定估值中枢。 不涉及代币经济和链上结算逻辑。 产品落地：已正式运行且有付费使用，营收可查 SEC 10-Q/10-K，财报数据是法定披露。 最新版本 未查到，近 90 天有效提交 未查到 次。
用户层面，MAU 和客户数以 10-Q/10-K 为准。股票 24h 成交额 $1.22M， 流通股和市值结构待确认。核心看营收增长率和毛利率是否匹配股价预期。 收入端，营业收入 $35.33B（最新财报/一致预期）， 毛利润按行业平均估算待补，净利润待 10-K/10-Q 确认， 股东收入看回购和股息。 美股公司赚钱不等于代币持有人赚钱，MSTR/COIN 等 BTC 关联标的要单独剥离 BTC 浮盈。 代码侧，90 天有效提交 未查到，活跃贡献者 未查到， 最新版本 未查到。GitHub 是 A 级证据可以直接核验。 投资背景，ASML Holding（$ASML）为上市主体， 股东结构以 13F/10-K 披露为准。一级合作以 IR 公告为 A 级证据， 媒体提及和行业会议属 C/D 级，不单独作为商业落地依据。
估值锚，流通市值 $723.31B， 用 P/E、P/S、EV/Revenue 估值，不适用代币解锁。 BTC 关联股（MSTR/COIN/MARA）要拆分 BTC 敞口和主营业务再估值。 和同行放一起看（统一口径，不跨赛道乱比）： 流通市值方面，ASML Holding $723.31B，TSM $2.24T，NVDA $5.45T。 FDV 方面，ASML Holding 未披露，TSM $2.24T，NVDA $5.45T。 年化收入方面，ASML Holding $35.33B，TSM $4.44T，NVDA $253.49B。 月活地址或用户方面，ASML Holding 未披露，TSM 未披露，NVDA 未披露。 数字以公开数据快照为准，部分缺失由官方自报或行业口径补。 估值，当前市值 $723.31B，P/S（共识收入）20.5x。 周期股（矿企/GPU）用周期调整 P/E。 悲观看 $723.31B 砍半，中性维持区间，乐观看 P/S 扩张 20-50%。 综合来看：基本面扎实（评分 57/100）。股权价值锚看营收、净利润、回购分红。 流通市值相对基本面中性，FDV 接近 MC，无大解锁，抛压可控。 潜在雷点：宏观利率上行压估值、AI capex 投入不及预期、监管诉讼（SEC/DoL）。 持续关注：营收增速、毛利率、回购金额、订单积压、机构持仓变化（13F）。 数据来自公开渠道，仅供参考，不构成投资建议。指标偏差超 30% 需重新评估。
就聊这么多，下期见。
#基本面研报 #美股 #研究 #OKXOrbit链上RWA存款三年涨了三百倍却还不到四十亿
DefiLlama 的最新数挺反直觉。存进 DeFi 协议的 RWA，也就是真实世界资产，总价值从三年前的1200万美元，涨到了现在的接近40亿美元，翻了三百多倍。听起来很猛对吧，但换个角度看，40亿放在整个加密市场里还是个小不点，连很多单条公链的锁仓都不到。
RWA 的逻辑其实很直白：把美债、国债、信贷、房地产这些传统资产搬到链上，变成能自由转让的代币，然后在 DeFi 里当抵押、当收益、当流动性。机构要的是合规和效率，链上要的是真实收益和稳定现金流，两边一拍即合。像 Centrifuge 这类协议就在干这件事，把机构级资产代币化丢进 DeFi 生态，让链上用户能间接吃到国债利息。
为什么值得咱们盯。因为 RWA 给链上带来了和投机币完全不同的钱。它挂钩的是美债收益率这种真实利息，熊市里反而更抗跌，也给了 DeFi 协议一条不那么靠发币续命的活路。ETH、SOL 这些原生资产波动大，RWA 像是给整个系统加了块压舱石，让 TVL 不再全靠代币价格撑着。
从链上看，目前 RWA 的锁仓主要集中在以太坊等少数几条公链上，底层资产又以美债和货币市场基金为主，结构还很简单。和传统债市几百万亿的规模比，40亿只是刚起步，可增长曲线已经说明机构在用脚投票。对咱们而言，RWA 不只是个概念板块，它代表着链上收益开始和真实利率挂钩，这会让 DeFi 的 TVL 不再只看代币价格脸色。
短期看 RWA 还在早期，40亿对比传统债市是九牛一毛，增量空间大但落地慢，托管和合规哪一步卡住都难放量。长线逻辑清晰，机构化、合规化是确定的方向。对咱们做趋势的，RWA 赛道相关的代币和协议，回调时分批布局比追高舒服得多，别在利好出来的当天追着买。
不过也别把它神化，链上 RWA 的托管和赎回仍绕不开线下机构，出了事可不是代码能自动解决的，底层资产违约照样连累链上持有人。你们觉得 RWA 会是下一轮牛的真主线，还是又一个被讲大的故事？Bitcoin is standing at the crossroads of a “fair coin toss”
In the past few weeks, Bitcoin has seemed like it was paused. It’s been hovering around $64,000, unable to rise or fall, with the market so quiet it makes you sleepy.
But the quieter it is, the more alert you need to be.
Sean Farrell, Head of Digital Asset Strategy at Fundstrat, reviewed eight historical instances when Bitcoin’s 30-day volatility dropped to historically low levels.
The result: in the following 60 days, the median absolute price change of Bitcoin was 30.2%.
Out of those eight times, four were up, four were down.
This is not some mystical indicator. It’s a market rule that has stood up to backtesting—
Low volatility is always followed by high volatility.
The question now is: are you betting on a 30% rise or a 30% fall?
Putting the numbers into real money—
At the current $64,000 level:
Up 30% → $83,200
Down 30% → $44,800
That’s nearly a $40,000 difference up or down.
This is not a small move; this is volatility at the level of “wealth redistribution.”
Farrell himself said: Monday’s 2% rebound was mainly driven by short covering, not new buying.
Since last Friday, Bitcoin futures open interest priced in Bitcoin has dropped about 8%. Shorts are retreating, but longs are not aggressively entering.
This is a stalemate where “no one wants to be the counterparty.”
What’s even more painful: the global 10-year real yield surged to 2.41% on August 14, the highest since Bitcoin’s inception.
When government bonds can give you nearly 5% risk-free returns, why would an asset like Bitcoin, which pays no interest, attract incremental capital?
Farrell’s exact words: the sustained rise in real yields is Bitcoin’s biggest downside risk right now.
On one side is the historical rule of “low volatility must be followed by big moves,” on the other is the macro headwind of “risk-free yields hitting new highs.”
This is not a simple multiple-choice question. It’s a “coin toss”—heads $83,000, tails $45,000.
Bitcoin has dropped nearly 27% since 2026.
At this point, the fearful are cutting losses, the greedy are bottom fishing.
But the truly smart are waiting—waiting for the sound of the “coin landing.”
In all eight historical samples, none failed.
This time won’t be an exception either.
In the next 60 days, it will be either $83,000 or $45,000.BIS戳破AI泡沫说风险溢价已被压到极限
国际清算银行（BIS）在最新年报里罕见地敲了警钟。他们说 AI 的乐观情绪过去一年撑起了全球增长和风险资产，可泡沫的苗头已经出来了。核心公司的估值站在高位，市场隐含的长期盈利增速明显高于历史基准，而最该警惕的是风险溢价被压得越来越薄，投资者拿到的风险补偿在变少，等于在用更少的缓冲去赌更贵的资产。
BIS 还点了一个很多人没注意的雷：循环融资。一些芯片商和云厂商通过股权投资、长期采购承诺、算力租赁加数据中心回租，搞出一套自我循环的账。钱从这个口袋进那个口袋，部分资产还可能被重复质押。一旦股权市场大调整，这些信用链条会被重新定价，利差走阔，融资环境瞬间收紧，最先崩的就是杠杆最高的那批。
对咱们玩 crypto 的人来说，这不是隔岸观火。AI 叙事和加密一样，都是这轮风险偏好的发动机。美股科技和 BTC 经常同涨同跌，靠的是同一个流动性故事。BIS 担心的长端利率上行和信用利差扩大，恰恰是压估值的两只手。真要到那一天，高估值资产包括大饼都得跟着折现率往上抬，没人能独善其身。
把这层关系说直白点，加密市场里很多人是加了杠杆在赌上涨，一旦信用环境收紧、股市承压，率先被杀的就是高β资产。BIS 担心的不是明天就崩，而是当所有人都在同一个方向下注时，拐点的杀伤力会被放大。咱们做趋势的，越是市场安静越要把止损和现金流想清楚，留着子弹才能在下一次错杀里捡便宜，而不是被 Margin Call 抬出去。
短期看市场还在狂欢，VIX 趴在年内低位，没人觉得会出事。但历史的回撤往往始于最安静的时候，去年高盛就说过 AI 基建贡献了近半的 EPS 增长，说明利润高度集中。长线逻辑没变，AI 和加密的机构化还在走，可短线别把杠杆加在别人的乐观上。你们觉得这轮 AI 泡沫，会比2000年的互联网泡沫温柔吗？A chess piece was quietly snatched from the enemy's barracks and landed in his palm—this was never a simple retreat, but a heavy cannon fire that had yet to land.
On the board, a true player does not rush to cheer just because a piece leaves the enemy's formation. Withdrawing coins is like moving a horse from the enemy's pawn line back to your own camp, meaning control of that piece has changed hands. Self-custody means moving the king out of the opponent's chariot range and taking fate into your own hands. But this step alone does not reveal the general's path. It may be to establish a defensive line on the rear flank, to lure away the next abandoned piece, or simply as a prelude before upgrading the pawn.
57,000 HYPE, $3.36 million. In terms of board value, that's equivalent to a castle plus a bishop. But the numbers themselves are meaningless; the key is which square the piece is placed. Taking it from Coinbase means every subsequent move will leave the public beatmap and enter a hidden game we can't observe.
What I want to ask is: why now? Why this quantity of 57,000 coins? If it corresponds to a precise position, it might be a rehearsal for a mid-game choking. If it corresponds to a long-term locked accumulation, then it's like the king chariot after its flip-side, waiting for the endgame to release pressure.
Don't forget there's another chessboard. The market interactions of US stock tokens $xAAPL are like another card unfolding simultaneously. A true expert calculates the movements of multiple boards at once, mapping the potential value of each piece to different situations.
Now, this piece has been stripped from the hot wallet's inventory and moved to an unknown address. It's like on a chessboard, a piece that was originally observed by an opponent suddenly enters the shadow square. We cannot see its next move, but we can infer its strategic intentions based on the rhythm and number of moves it accumulates withdrawals.
Sometimes, large withdrawals are just one person placing an important pawn into a safe under their control. Other times, it's preparing to withdraw forces for a large-scale attack. It's like in chess: what seems like a retreat is often meant to be a more aggressive right flank.
I only look at this 57,000 amount and how it accumulates withdrawals. If it is accumulated in batches and small steps, like slowly advancing a border pawn, then it is most likely a long-term self-custody will. If it is a large one-time withdrawal, then it is more like a lure before abandoning the player. But regardless of the case, one thing is certain: the moment this piece leaves the hot wallet, its risk map completely changes. It no longer depends on any opponent's goodwill, no longer affected by any centralized server outages. It is like a queen hanging alone in the center of the chessboard, with huge potential but must face all attacks alone.
The grandmaster never rushes to pronounce judgments. The real strategy is to calculate the endgame after twenty moves, and this withdrawal is only the first step. It changes the ownership of the pieces, but not the outcome of the game.
I stared at this address, as if staring at a chessboard piece just touched by a finger. It tilted slightly, as if it would fall into a square I hadn't yet seen clearly—and the coordinates of that square were the only mystery at this moment.
#影响周期 · Daily #链上事件 · Exchange withdrawals #57,000 HYPE · $3.36 million #coinmovealert特朗普家族的加密公司居然拿到了银行牌照
财联社这条消息看得人一愣。特朗普家族持股的一家加密公司，最近拿到了银行牌照。标题里那句金融史上最明目张胆，不是空穴来风，毕竟这家人一边在白宫搅动政策，一边在加密市场拿牌照开银行，角色重叠到让人分不清谁是裁判谁是运动员。
拿到银行牌照意味着什么。以前加密公司想碰传统 banking 业务，得绕一大圈找合作银行，现在自己就是银行。托管、支付、稳定币清算这些最肥的活儿，都能名正言顺往自家碗里装。对加密行业来说，这是机构化的又一步，巨头入场的速度比很多人想的快，连政治家族都下场把生意做成持牌机构了。
反差就在这儿。嘴上把加密当竞选筹码的人，手底下却把加密公司做成了持牌银行。政策利好和家族生意之间的边界越来越模糊。市场短期可能解读为利好，毕竟合规通道更顺了，大钱进场更方便。可一旦利益冲突被摆上台面，监管反噬的风险也跟着翻倍，今天给的牌照明天也可能被重新审查。
把这件事放回更大的画面里看，加密公司拿银行牌照过去几乎是不可能的任务，监管卡了多年。现在政治家族的入场把门槛踩出了一条缝，后面大概率会有更多机构跟进复制这条路。对稳定币和托管生意是实打实的利好，钱进出的管道更顺了。但硬币的另一面是，持牌就意味着被盯得更死，以后的一举一动都在监管眼里，草莽时代的自由红利会越来越少。
对咱们做趋势的，这条线长线偏多，机构合法化的故事还在讲。但短线别上头，这种消息往往冲高回落，真金白银的流入要等牌照落地后看存款和托管数据。仓位卡在关键位的，别被一个 headline 洗出去，更别因为情绪加杠杆去赌政策的延续性。
说到底，加密从草莽走向持牌，是趋势也是风险。你们觉得特朗普家族这步棋，是在给行业铺路，还是给自己挖护城河？亲手松绑AI监管的a16z遭司法部查
那个在加密圈和AI圈都呼风唤雨的风投巨头a16z，这回自己被查了。据彭博报道，美国司法部已经对安德森·霍洛维茨基金发起反垄断调查，矛头指向一个让人玩味的问题：同一家基金的人，怎么能在互相竞争的公司董事会里同时坐着。
事情要从去年那笔收购说起。调查最早是冲着Fivetran收购dbt labs去的，收购本身已经无条件通过，但监管对董事会席位的追问一直没停。a16z联合创始人本·霍洛维茨是Databricks的董事，合伙人马丁·卡萨多则是Fivetran的董事，而这两家公司在数据处理领域正面对打。卡萨多此前还待过被Fivetran收掉的dbt labs董事会。换句话说，同一笔钱投出来的两家对手，董事会里都站着a16z的人。
这种调查通常的结局很轻描淡写：董事辞职就完事。但背后的信号一点都不轻。a16z手里管着900亿美元资产，刚募完一支150亿美元的新基金，投资组合里塞满了SpaceX、OpenAI、Cursor这些名字。它能撬动的远不止钱，还有行业的游戏规则。
最讽刺的恰好在这里。过去两年，a16z两位创始人给特朗普竞选捐了数百万美元，还成功推动了AI政策里安全护栏的弱化。一边喊着少管、放开、让市场自己跑，一边自己的董事会布局却被司法部盯上，说你可能在用交叉任职卡住竞争。这画面怎么看都像个玩笑。据彭博披露，这轮调查其实已经悄悄进行快一年了，只是现在才浮出水面。
更值得玩味的是时机。加密和AI这两年越贴越紧，a16z既是crypto里最活跃的声音之一，也是AI牌桌上筹码最大的玩家。当监管开始顺着董事会这条线摸进VC内部，紧张的不只a16z一家。那些在多家竞品公司同时占座的一线基金，恐怕都要回头数数自己的董事席位了。
所以问题来了：当最会讲去中心化、最会喊减少干预的人，自己坐到了反垄断的被告席边上，这股风最终会吹向谁。加密借贷缩水四成这次去杠杆竟没崩盘
市场一直有个担心，就是这轮上涨全靠杠杆顶着，一旦借贷盘撑不住，就是2022年那种连环爆雷。Galaxy Research刚出的二季度报告，把这个担心以一种最不吓人的方式摆到了台面上。
二季度加密抵押借贷总规模环比掉了16.78%，降到561.6亿美元。跟2025年三季度的峰值789亿美元比，已经缩水超过四成。借钱炒币这件事，确实在悄悄退潮。
有意思的地方在这。过去大家印象里，DeFi借贷才是大头，链上协议随借随还，那才是加密的特色。这回反过来了。二季度DeFi借贷环比暴跌27.61%，掉到204.3亿美元，反而被CeFi中心化借贷反超，后者只降了9.62%到229.8亿美元。这是2023年三季度以来，CeFi规模第一次压过DeFi。
Tether依然是这个市场的隐形房东，在CeFi里占了58.54%的份额。换句话说，链上借钱的人少了，找中心化机构借的人却相对稳定。这也解释了为什么最近链上借贷协议的锁仓量普遍涨不动，钱更愿意待在熟人机构手里。
最让分析师松口气的是节奏。这次去杠杆跟2022年完全不是一回事。那会儿一个季度能直接塌掉55%以上，是悬崖式崩盘。这回是连续几个季度分别只掉10%、5%和17%，一步步往下走，没有哪一步是断裂的。Galaxy的判断是，只要不出现极端清算或者对手方违约，这种温和的台阶式下降还会继续。
期货那边也是类似味道。二季度未平仓合约微降3.08%到1032亿美元，但七月已经弹回约1140亿。机构公司债这边，Strategy在五月回购了15亿美元债务，把整个DAT行业的未偿债务压到161亿美元。
我倒觉得值得琢磨的是另一面。借贷退潮通常意味着热度在降，钱没那么敢加杠杆了。但这次没有伴随价格崩盘，说明现在的持仓结构比2022年健康不少。问题是，这种慢刀子割的方式，反而更难让人警觉。等大家发现借贷盘已经缩了四成，市场可能早换了种脾气。
接下来真正要盯的，不是借贷会不会继续降，而是这种温和去杠杆能不能撑到下一轮扩张。你们觉得这算好事还是隐患。Proving that Arbitrum needs to change its engine to accelerate 4000 times
Arbitrum plans to upgrade its engine to a bigger one. The Offchain Labs research team just released progress in ZK technology, including a vector commitment scheme that reduces the time to generate proof for 64,000 data entries from about 2 minutes to 32 milliseconds, a speed increase of about 4,000 times. The new proof system Zaratan, for the first time, achieves integer-native fully concise proofs, cutting the overhead of computations like RSA by about 5,000 times.
Just looking at the numbers might not make a difference, but in plain terms: in the future, verifying a transaction or a segment of computation on Arbitrum will be fast and cheap, without needing to trust any centralized node. But the validation AI segment has also moved forward. Lightweight verification protocols can reduce the validation of large model inference from minutes to milliseconds, lowering the threshold for running AI inference on-chain.
The key is not speed, but the architecture must change. Arbitrum is exploring combining ZK proofs with the original fraud proofs and TEE trusted execution environment, creating a multi-proof architecture. This means no longer betting on a single verification method; multiple mechanisms run in parallel, single points of failure are eliminated, and L1 settlement cycles can be shortened. For a Layer2, the faster the settlement, the higher the capital efficiency, and the shorter the time users wait to confirm.
The market impact extends to the coins. This kind of underlying upgrade won't directly drive AR's price in the short term; the market is speculating on expectations. But in the long run, with settlement security and speed rising, Arbitrum will have confidence to handle more real transactions and assets, and TVL and fee income are the foundation of the slow bull market. ETH itself will also benefit; if L2s run smoothly, the mainnet's value capture will be stable.
Looking at the bigger picture, the arms race between ZK Rollups and OP Rollups has never stopped, and competitors like Base and OP are also aggressively improving performance. If Arbitrum gets mainnet running this step, it will be a first-mover in the new track of multi-provers. But from publishing the paper to mainnet deployment, there are still auditing, testnets, and countless pitfalls in between.
What really matters is when the multi-proof architecture launches on mainnet—that's the watershed. Don't treat technological progress as a buy order. Do you think this 4,000-fold acceleration is a turning point in the Layer2 arms race, or just another paper worth listening to?
Multi-prover architectures don't strongly affect ordinary users, but developers and big capital care a lot. When settlement certainty is high, institutions dare to put large assets and high-frequency strategies on-chain, which is also the key chip for Arbitrum and Base to compete for institutional RWA business. Whoever sets security and speed as default values first will get the biggest share in the next round of real capital inflow.