Post

挖矿的小羊
挖矿的小羊
Show original
AMD发了2026年Q2财报—— 营收115.4亿美元,同比增长50%,创历史新高。 净利润23亿美元,同比增长163%。 数据中心营收67亿美元,同比增长107%,占总营收58%。 每股收益1.66美元,同比增长246%。 营收超预期、利润超预期、毛利率56%。 全是利好,对吧? 然后盘后股价直接跌了8.8%,一度跌超9%。 一份“高分卷”,被市场扔进了垃圾桶。 原因就两个:花钱太狠,预期太高。 第一,资本支出8.08亿美元,市场预期只有2.986亿美元。接近三倍的差距。AMD在AI基础设施上烧钱的速度,超出了所有人的想象。 第二,三季度营收指引130亿美元,虽然高于分析师预期的125亿,但部分买方机构此前押注的是140亿。 “超预期”不够用了,市场要的是“超级超预期”。 这不就是我们币圈每天都在经历的事吗? 利好落地就是利空。财报越好,跌得越狠。预期才是这个市场唯一的神。 那这对我们加密市场意味着什么?两个传导链条,直接关系到你的持仓。 链条一:GPU被AI巨头抢光,矿工正在变成AI算力商 AMD数据中心营收翻倍增长,Instinct AI GPU需求是核心驱动力。GPU产能正在被Meta、OpenAI、Oracle这样的AI巨头包圆。 加密矿工获取新显卡的难度和成本只会越来越高。 但更有意思的是另一件事—— 比特币矿工正在加速转型成AI算力租赁商。 公开矿企已经签了价值超过700亿美元的AI托管合约。分析师预计到2026年底,上市矿企来自AI的收入占比可能从现在的30%升到70%。 Core Scientific刚和AMD签了15年、529兆瓦的AI基础设施协议,预计带来140亿美元收入。Riot Platforms的数据中心业务已经首次入账3320万美元。 矿工不挖矿了,改行给AI当房东了。 这对去中心化GPU网络——Akash、Render——意味着什么? 传统矿工带着海量算力涌入AI租赁市场,去中心化算力网络的定价权正在被稀释。算力硬件复用正在冲击这些项目的估值逻辑。 链条二:AI概念币的“AMD时刻” AMD盘后跌了9%,Hyperliquid上的AMD永续合约直接触发了206万美元清算,多单占了95%。 半导体抛售曾多次拖累比特币走低。 今天BTC在6.4万美元附近盘整。如果今晚美股正式开盘后AMD继续深跌,AI相关加密资产——FET、TAO、RNDR——大概率面临联动抛压。 AI是这轮唯一有业绩预期的赛道。但AI芯片股的每一次动荡,都会传导到AI叙事币上。 反过来,如果市场消化利空后V型反弹,那就是一次抄底AI赛道的机会。 关键催化剂还在后面: AMD首款机架级AI系统Helios,本季度开始向Meta、OpenAI、Oracle出货。这是AMD首个集成自研CPU、GPU和网络芯片的完整系统,直接对标英伟达的整机方案。 Helios Q3的出货数据,才是真正的胜负手。 不要死在黎明前的“指引不及预期”里。 $BTC $AMD $ETH #AMD财报超预期,增长已被透支?
挖矿的小羊
挖矿的小羊
Why would Morgan Stanley dare to set a target price of 300 for a stock that dropped from 225 to 108? SpaceX, the largest IPO in human history, once approached $3 trillion in market value on its first day of listing. And then? It plummeted all the way. The 52-week high was $225.64, now down to just $115—nearly halved. The financial report showed revenue of $7.814 billion, up 92% year-on-year, far exceeding the expected $6.9 billion. And then? It continued to fall 6.5% in after-hours trading. Unexpected earnings results have been met with unexpected sell-offs. Doesn't it sound familiar? The script of all the positive news has been played out in the chip sector, and then moved to space. So here's the question—why does Morgan Stanley, this "old fox," still dare to set a target price of $300? Not because they were foolish. It's because the valuation logic is completely different. Morgan Stanley uses a method called SOTP—the segment valuation method. Simply put, it's like splitting a company into four parts, settling accounts separately, and then adding them up to the end. The results after tearing it down were terrifying: Space business (rocket launches): $8 per share It's the company that sent humans to Mars, one of only two in the world capable of recycling rockets—Morgan Stanley thinks it's only worth 8 yuan. Network and Starlink: $128 per share A satellite internet empire with 12 million users and annual revenue exceeding $10 billion—worth 128 yuan. X platform and Grok AI: $12 per share Musk spent 44 billion to buy the social platform, plus a self-developed large model—priced at 12 yuan per package. Enterprise AI business: $152 per share — Accounting for more than half of the total valuation. Do you understand? Morgan Stanley thinks this company is worth 300 yuan. Among them, the rocket is worth only 8 yuan, and the AI value is 152 yuan. And what is the current stock price? $115. It can't even cover Starlink alone. In other words, the current market pricing is—Starlink giving away freebies, Rocket giving away freebies, AI being a negative asset. Deutsche Bank puts it even more bluntly: based on current market value, the market's pricing for SpaceX's AI business is close to zero. A company with AI business quarterly revenue of $2.561 billion, a 247% year-on-year increase, and has just signed multi-billion dollar monthly computing power contracts with Anthropic and Google— The market said: This thing isn't worth much. Do you think this makes sense? But the market has its own logic. On August 6, SpaceX experienced its first large-scale lock-up unlock. 911.5 million internal shares will be granted free trading rights. Currently, only about 5% of tradable shares are tradable, but after the lock-up is lifted, this will double to 12%. At the current stock price, its value exceeds $100 billion. Meanwhile, short positions have already bet $24.6 billion, with $7.3 billion in unrealized profits still increasing their positions. Twenty-seven institutions unanimously called for a "buy" position, with an average target price of $236. Bears are both floating and increasing their holdings. Both sides are proving themselves with real money. I've rarely seen this kind of split between bulls and bears in my career. What inspiration does that offer us? AI valuation is undergoing a shift from "concept" to "cash flow." SpaceX's AI business has already signed cloud service contracts worth tens of billions of dollars annually. Grok 4.6 will be released next week, feeding all of SpaceX's more than twenty years of engineering data into model training. AI is not a story. It is a reality that is happening. And the market now rates this reality as—zero. Either the market is right, and AI is worthless. Either the market is wrong, and someone is selling gold at ankle prices. SpaceX's financial report is not just a financial report, but a capital verification. The lifting on August 6th is not a restriction, but a demon-revealing mirror. Bears are betting that early investors will panic sell. The bulls are betting that the market will finally see the true value of AI. I was watching the show, and also doing the math. $BTC $ETH $xSPCX #SpaceX首份财报超预期, unlocking remains a key variable

Disclaimer: OKX Orbit content is provided for informational purposes only. Learn more

Replies

No comments yet. Be the first to reply!