Compartk 𝕏/🚀🧬🇱🇮

All my ideologies politicaltests.github.io/12axes/results…

@yunta_tsai my favorite comic

May this era be the new dawn of humanity.

- Inventory $25.8B in Q1 FY27, 4.4x the FY23 peak ($5.16B) and 5x FY22 levels - Raw materials up+75% q/q while finished goods up +5% q/q; likely hoarding upstream components (HBM, CoWoS substrates, optical) for Rubin ramp

Watching your friend make more money on the stock that you recommended but sold too early

“Until death, all defeat is psychological."

投资真正买的是被市场低估的未来现金流折现，用今天的价格，买入一组未来现金流分布，你的收益来自真实分布比市场隐含分布更好。

everybody racing to IPO before anyone shows how all the recurring revenue might end up not recurring

….

@realDonaldTrump Look who climbed out of his Israeli echo chamber….

@mandyarthur @realDonaldTrump Rep. Thomas Massie. He consistently votes to uphold limited government, individual rights, and constitutional constraints on federal power, often against both parties.

@TobyAtLarge 那太好了，就这样。路过时候叫我

@Compart42 ohh 那看什么时候约复旦里面也行🤔 光华楼或者旦苑或者大学路啥的 看我哪天路过五角场吧😂

长期有效 欢迎各位温哥华/西雅图地区的开发者 一起来约咖啡☕️ （咖啡我请😁） 无论你是sde pm ds还是单纯对开发感兴趣 都欢迎来聊聊，看看有什么能互相帮助的

@TobyAtLarge 倒是有点远，我在复旦大学本部

@Compart42 好的呀😁 我现在住漕宝路附近，周末可以走一个

@hsn8086 反社会在哪？

@TobyAtLarge 投机者承担价格判断风险，提供流动性，压缩错价，让资源配置更快反映真实稀缺。错误投机会亏损，正确投机会获利，这恰恰是市场筛选

金融化很容易让人误以为“钱生钱”是一件天经地义的事情 好像我把100块放余额宝里 我的每一块钱就开始吭哧吭哧打工自我复制 但你真的落到微观上 除了低吸高抛的零和赌场博弈以外 真正创造财富的 还是一线实地的工业党、小商小贩、和实体经济 说白了，不是钱生了钱 还是劳动人民，创造了财富🤔

大部分都是资源浪费...根本不需要多少 CPU...云厂商把面向 Agent 基建优化好后需求一下子就下来了，觉得 CPU 重要一定没写过 Agent

maximizing shareholder value

@iamai_omni 天才，笑死我了

千万不能去同一家公司，会发生反物质湮灭。

@maxinomics I didn't think this was the problem either. I was all-in Nvidia and Corewave ipo and actually made some fortune. Theoretically, labs can stop training and make money profit, but that's not a non-competitive monopolist market we see here, It's actually the most competitive area

Come on you really think I'm blind to the idea of ROIC? I just do not agree the labs will never turn net profit There are so many ways this manifests into cash flow. Even if it isn't through the labs, the models of today are good enough to generate enough economic surplus elsewhere in the economy to offset their valuations Two years ago when models weren't this capable I would have been more inclined to agree with you, but we've crossed the threshold You could stop training now, dedicate all GPUs to inference, markup each token and current + projected capex spend would be erased within 5 years

The dotcom bubble wasn't Pets.com It was the $500 billion of fiber optic cable buried between cities while every house in America was still on dial-up The AI buildout is doing the exact same thing. With one critical difference

@maxinomics What you are missing is ROIC from these constructions. Semis, clouds might seems making profit, but there are based on RPO from ai labs. And due to competition, These labs will never make net profit, So valuation will tank

@Compart42 Did you watch the whole clip?