Benjamin retweeted
I think $AVGO is fucked too. Triple kills ! Genius Deal!
$LITE shareholders should be annoyed...
The 23.9% premium in the $MRVL deal is real on paper, but in dollars it's gone.
1. 4.5 $MRVL per $LITE = $1,132/share on 28 Oct = 23.9% above Lumentum's undisturbed $914.
This is the 23.9% premium reported today.
2. Marvell has since fallen ~22% to $196.40 (Friday close).
So 4.5 shares = ~$884 = 3.3% below Lumentum’s undisturbed price of $914.
3. The deal's $102.5B equity value is now ~$80B.
4. Lumentum fell ~8% in the same selloff to the ~$840s.
5. So even vs where Lumentum actually trades, the premium has shrunk from 23.9% to ~5%.
I think Marvell needs to get back above ~$203 for Lumentum shareholders to see any premiums over the undisturbed price.
It's essentially a premium paid in a currency (stock) that’s fallen 22%.
Welcome any pushback on my napkin math calcs. By no means a comprehensive model.
Benjamin retweeted
请大家注意,炼油股 $VLO $MPC $DINO $CVI $CLMT 和油轮股 $ECO,其实对于大家而言并不比科技股更难研究。
能源市场有一套非常成熟的价格发现体系。原油, 成品油, 裂解价差, 运费都有市场公开定价。你能够直接观察市场如何定价一年后的供需预期,并不断用库存、贸易流向和实际产能验证这些预期。
这些都是公开信息,这个市场更接近一个有效市场。
能源市场的优势在于其真实的、涉及主权国家的供需关系,以及大量可以交叉验证的公开数据。
彭博社、路透社、标普能源的公开信息,结合期货曲线,已经足够让普通投资者建立相当完整的研究框架。
这个市场是最不需要阴谋论和内幕消息的。
原油,成品油市场期货曲线观察
并不利空炼油股,油轮股,和一切。 $Vlo $mpc $dino $clmt $cvi $eco
按 10 月 2 日收盘,Brent Dec/Jan 从上次约 $1.06/bbl 扩到 $3.54/bbl($102.25 / $98.71),是非常明显的重新变陡;
WTI Nov/Dec 从 $1.50 → $1.68/bbl,变化不大。
ULSD Nov/Dec 从 13.58¢ → 14.38¢/gal,同样没有因为昨天柴油大跌而明显松动。
更值得注意的是远端。
Dec-27 Brent 从约 $80.92 → $81.93(+1.25%)
WTI 从 $74.93 → $75.56(+0.84%)
Dec-27 ULSD 从 $3.3046 → $3.3669/gal(+1.9%)。
因此我们的 Dec-27 ULSD-WTI crack proxy 从约 $63.9 → $65.85/bbl,重新扩大约 3%。
这发生在一个很有意思的背景下:G7 刚宣布协调释放约 1 亿桶柴油和原油储备,其中柴油释放还会前置进行。消息出来后美国 ULSD outright 当天下跌约 3.7%,欧洲 gasoil 跌幅更大。
但曲线告诉我们:
政策可以把眼前价格砸下来,但市场仍然没有相信柴油结构性短缺已经解决。
如果这次储备释放真的让市场开始相信“只是 temporary shock”,我们应该看到 ULSD Nov/Dec 明显压平 + Dec-27 ULSD 下跌 + 2027 crack 收缩。
实际发生的是:近月 spread 没松,Dec-27 ULSD 反而上涨得比 WTI 更快,2027 crack 再次往 $66/bbl 靠近。
所以我会把这次变化理解成:
Prompt price relief, but persistence premium is rebuilding.
还不能称为完整的 regime shift,因为 2027 crack 仍在我们最近的 $64–67 区间内,没有突破 $70–75;但方向已经和上一轮不同。
尤其如果未来几次检查在储备释放真正进入市场后,2027 crack 仍然不跌,甚至继续扩大,那会是很强的信号:市场认为战略库存只能填短期缺口,无法解决 2027 年的 refining/product capacity 问题。
这是最后的疯狂。“两个阶段、三步走”马上将会跨入第二个阶段:市场彻底看破嘴炮式预期管理,全面交易债务货币化+滞胀。现在正如我们几个月来一直斩钉截铁地预判,国债小阳春终究昙花一现,这几天美国长期国债崩盘式下跌。但是我要提醒大家:在全面进入第二个阶段之前,川普/贝森特/沃什马上会放出最后的终极大招,但大招的边际效应在急剧递减。从地缘TACO,到嘴炮放鹰,再到托举日元,有效的力度和持续的时间越来越短。接下来不管再放什么大招,都无法再让市场长期买账,但是几天到几周的“回光返照”可能会有。
国债收益率暴涨一定会挤压风险资产的估值,但市场可能会先走一段“Cash is trash”,疯狂购买资产的狂热交易。在七十年代美元与黄金脱钩的时候发生过。这时候不要被诓骗。行情一定分为“三步走”:第二步的“动荡”是绕不过去的,国债收益率必将压制估值。之后才是第三步的“分化”行情。现在还在第一步的“超宽幅震荡、上下游跷跷板”行情。
在中期选举之前,美政府必定还会释放超级大招,这时要沉得住气,需知这不过是回光返照。美政府终究回天乏术,完全无法挽救美债和美元信用。另外,国债收益率暴涨可能会对黄金这个无息资产形成最后一波压制,美国政府为了挽回美元和美债的信用,也难免还会使出一些盘外招。但任何回调都是千载难逢的上车机会。随着美债与美元信用不可逆转地下滑,黄金这类核心资产与现金牛股票,必将在债务货币化与滞胀交易中一飞冲天!
Benjamin retweeted
行了,也确实是第一次为了一个男人熬夜、买token、做片。
五分钟,花了我不少钱,大家别吃文字版的瓜了,视频给你们准备好了~
孙割,你这新概念作文一等奖真不是盖的,你这故事后面拍电影必火,一定要找我合作🤝
🤖 Made with AI
Benjamin retweeted
Roughly: Photonics > Memory > Storage
As you can see YMTC is going to be the biggest producer of NAND in the world by 2027. NAND is Storage aka the market that $SNDK is in and much less complex than HBM.
HBM made by $MU and $SKHY have alot more technical moat and the Chinese company CXMT will have a harder time catch up to $MU and $SKHY compared to YMTC catching up to $SNDK.
Furthermore, looking at Nvidia's roadmap of 192GB HBM per Rubin Ultra, photonics looks to be more in focused than HBM.
I'm still studying details but I'm roughly thinking that $LITE > $SKHY, $MU > $SNDK.
Benjamin retweeted
Strategy increased USD Reserve to $5.10B, established additional USD Cash of $1.59B, and repurchased $136M of $STRC. As of 8/23/26: Strategy holds ~4% of Total BTC Supply and has ~0% Net Leverage. $MSTR strategy.com/press/strategy-…
Benjamin retweeted
Replying to @anzihfhhf @bboczeng
大部分好资产是需要时间来演绎的
@bboczeng 说的有一点没错,快速上涨都是资金推动而不是基本面推动的。赚大钱需要长坡厚雪而不是资金推动
Benjamin retweeted
X 平台开源了整个平台的算法权重,这在历史上是少有的。
Grok 通常不会去阅读帖子内的内容,但是会预测每个人看到这条推文后会做什么。
比如会不会点赞、回复、转发、复制链接或关注作者,这能提高不少权重分数。
而举报、屏蔽和点击不感兴趣,每种行为都会带来负面性的权重,系统把这些预测结果,综合叠加起来,得出这篇推文对该用户的最终分数。
我们在算完它的所有权重分数后(如下图所示),得到了几个关键性的结果,即分数也并不能决定一切。
1. 对多次刷到的惩罚措施
如果同一名读者的信息流里,已经出现过你的推文,那么你的第二篇推文,只保留 62.5% 的分数,第三篇只剩约 44%,最低不会低于 25%。
也就是说,算法明确限制同一位用户,多次刷到你的推文,并且多次这个行为,会降低权重和分数。
2. 原创推文不会有任何折扣
X 明显更偏爱原创推文,不太愿意推荐回复和转发。如果读者没有关注你,你的回复或转发,基本不会进入到对方的时间线中。
而即使读者已经关注你,回复和转发类内容,也同样要乘以 0.75,只有来自已关注账号的原创推文,可以保留全部权重。算法有意把回复和转发视为相对较弱的内容。
3. 相似内容筛选
在完成所有打分 + 权重后,系统 Grok 还会通过一个重排序模型,将相似帖子分散开来。
即使你的分数不错,但如果内容与前一篇推文太像,排名仍可能向后掉几位。
哪些事情不要做
不要刷点赞,点赞权重只有 0.5,在整套 Grok 算法里是几乎最差的,同样的还有个人主页访问,几乎没用。
之所以长推文,变得没以前那么重要,主要停留时长的分数,也不再像以前那么高,是否停留的权重是 0,实际停留时间,权重分数也很低。
不要把到处回复和转发,当成触达陌生人的增长方法。陌生账号的回复和转发,会在评分前被过滤,即使对方已经关注你,你们是互相关注状态。
不要故意发布容易引发举报、屏蔽或被点击不感兴趣的内容,系统只要预测到一次可观的举报,账号就会存在极为剧烈的降级,X 不鼓励撕逼、不健康的内容。
这可能也让色情、擦边、垃圾内容或相互骂人的推文,大大退出时间线的原因,系统也可能在阻止这些内容,进入到大家的时间线中。
最终的结果,X 真正奖励的不是看起来很热闹,而是用户愿不愿意主动参与讨论、分享并关注作者。而真正的惩罚,也不是没人点赞,而是刷到的读者,明确表示不想再看到这种内容。
最后,感谢马斯克愿意将 X 平台算法开源,使我们看到了 X 背后的运作方式。
AB KUAI DONG
Benjamin retweeted
If you look at the Google corporate bond spread, at such a record-high 100bps level it’s still raising debt like free money on the table. Why? Because it believes the cloud business is a “buy more, save more” game and doesn’t give a shit about capital cost, as every dollar borrowed can yield 20-30¢ profits a year immediately.
Is it reasonable for CSPs to be euphoric? Yes—it means they see demand growing more crazily than we outsiders can imagine.
Is it a risk? Yes, because no one knows how long the demand growth will be sustained. When the music stops, it’ll be the doomsday.
Is it a good time to short? NO! As a retail trader, respect the market and follow the trend. I’d never bet money on just the narrative and fuel bulls to squeeze higher. The charts will tell us when the tide turns and we can join that side.
But one thing he mentioned is right: the core checkpoint for $NBIS next quarter is the DC building progress at Vineland, NJ, delayed by power solutions opposed by community residents.
This quoted post is unavailable.
感觉 SpaceX 现在这个阶段的定位,已经成为了一个大号的 NeoCloud。
但我不太理解,如果要投资 NeoCloud 的话,为什么不选择 Nebius 或者 Coreweave,估值更合理更便宜,潜力也更大。
Nebius 和 Coreweave 不只是单纯的出租裸金属算力,还会覆盖 AI 托管推理、任务编排、模型自动路由这些上层服务,覆盖完整的 AI 生命周期。
在 AI 从训练进入到推理时代之后,未来很大一部分长期计算需求都会由这些公司承担。
而对于 SpaceX,我一直认为它最大的看涨期权就是 Space 以及 Starlink 太空相关的业务,投资的是火星殖民和多行星物种叙事。
但如果现阶段 SpaceX 最大的目标是全面 All in 算力,那真的不如投资 Oracle、Nebius 或者 Coreweave。
Benjamin retweeted
🦔Nvidia announced agreements yesterday with the six biggest names in private capital, Apollo, Blackstone, BlackRock, Brookfield, Goldman Sachs, and KKR, to raise over $500 billion so its own customers can buy Nvidia chips. These are memorandums of understanding, with final terms still to come. Jensen Huang is calling GPUs a new investable asset class, like real estate or toll roads. The financing lets OpenAI and others buy hardware without putting it on their balance sheets.
My Take
I flagged Nvidia's customer financing back in July. This is that same idea, now at 60 times the size. Lucent and Nortel ran this exact play in the late 1990s, when they lent customers the money to buy equipment they couldn't otherwise afford. Both ended in the largest corporate bankruptcies of their day. The vendor lends to keep its own revenue up, the sales look strong for a while, and the risk stacks up on the buyer's side of the ledger. We have seen how the story ends.
I don't buy the idea that a GPU is an asset like a toll road. A toll road still earns money in 30 years. A GPU is close to worthless in five, and Nvidia's own next chip crushes the resale price of the last one. So they want pension funds and insurers to lend half a trillion dollars against hardware that ages like fruit, and the six firms who put the deal together collect their fees up front no matter how it turns out. Huang keeps his sales up and hands the downside to whoever holds the paper. I have watched this movie twice now, and the vendor never ends up holding the bag. Everyone downstream does.
Hedgie🤗




























