华尔街萌喵

这里面其实有俩问题：长鑫在Dram层面的技术，没有长江存储在Nand领域的技术领先，如果放开设备管控，长江在NAND领域很快就能干翻三星之流，长鑫还差点。 1）大概2024年前后，长鑫的Dram差不多才到18.5nm，目前最多也就到15nm，同等时期海力士/三星差不多已经到10nm深水区了，接近物理极限。更不用谈以HBM这样3D堆叠为代表的高性能存储。 2）Nand领域，3D堆叠技术，长江完全不输海力士/三星。技术路线相当牛逼，只不过苦于没有EUV和好的刻蚀设备，现有的设备即便能做到300层往上，良率也很低。 EUV的重要性就在这里体现，因为随着芯片的物理极限缩小，基本在光刻阶段要多重曝光。 目前的替代方案是用浸润式DUV进行多重曝光，即便如此，DUV的光源最多单次也只能打到40nm左右，EUV单次最低能干到13nm。这就是差距。如果High-NA的EUV，甚至低于10nm。 其次比如说高深宽比的刻蚀机。国内独苗就是中微，大体能极限做到90层，但应该还没有量产，同期的Lam（泛林半导体）和Tel（东京电子）至少能做到300层往上，目前差不多都在做400层以上的稳定量产。 单台EUV（LoW-NA）1.5亿美金起步，单台极高深宽比刻蚀机最低也在5000万美金起步。 只能说，中低端或者部分高端，在长江和长鑫这里优势会越发明显，客观来说Nand的价格战也是长江最先打起来的，三星和海力士也不得不转向高端，并寄希望美国的封锁能持续下去。 否则只要设备和材料端，国内一突破，价格战将是海啸级别的。

$AMD EPYC CPUs Reach Record Server Revenue Share of 46.2% According to the latest data from Mercury Research, AMD achieved this with only a 27.4% unit share, indicating that while AMD’s CPUs do not represent nearly half of total units sold, they command much higher average selling prices than the competition.¡ That competition is primarily $INTC, still the largest CPU vendor Intel currently holds the majority of units sold at 54.9%, down 3.4% compared to Q4 2025. We lack exact revenue share data for Intel CPUs, but given Intel’s majority in unit shipments and AMD’s near-half share of total server CPU spending, it can be concluded that the average selling price of Intel Xeon server CPUs is lower than that of AMD EPYC In the data center, ARM-based designs accounted for about 17.7% of unit shipments in Q1 2026, meaning nearly one in five CPUs was Arm-based. Whether these came from third-party integrations by Ampere or other $ARM CPU makers, or from in-house CPU designs by companies like Google, AWS, or Microsoft, remains unclear. However, Mercury Research collects extensive market data, lending confidence to these figures Strong CPU demand and AMD’s success are being driven by the recent surge in agentic AI, which is increasing the number of CPUs required in new deployments to almost match the number of GPUs. The traditional setup, where one CPU was paired with four or even eight GPUs, is shifting toward a one-to-one ratio of CPUs to GPUs in agentic AI deployments AMD is selling every CPU it produces. Intel is also seeing strong demand, even selling dies located on the very edge of the silicon wafer, which would normally become scrap, to eager customers. However, AMD currently appears to be achieving higher ASPs for its products

What an insane day for photonics. $SIVE up 31.3% $TSEM up 23.1% $AAOI 20.01%. It feels like a lot… but this just means you’re early to the next supercycle and there’s a lot of room to go. Lot of people on X ask what’s next after $SNDK? Here they are.

$AMD EPYC CPUs Reach Record Server Revenue Share of 46.2% According to the latest data from Mercury Research, AMD achieved this with only a 27.4% unit share, indicating that while AMD’s CPUs do not represent nearly half of total units sold, they command much higher average selling prices than the competition.¡ That competition is primarily $INTC, still the largest CPU vendor Intel currently holds the majority of units sold at 54.9%, down 3.4% compared to Q4 2025. We lack exact revenue share data for Intel CPUs, but given Intel’s majority in unit shipments and AMD’s near-half share of total server CPU spending, it can be concluded that the average selling price of Intel Xeon server CPUs is lower than that of AMD EPYC In the data center, ARM-based designs accounted for about 17.7% of unit shipments in Q1 2026, meaning nearly one in five CPUs was Arm-based. Whether these came from third-party integrations by Ampere or other $ARM CPU makers, or from in-house CPU designs by companies like Google, AWS, or Microsoft, remains unclear. However, Mercury Research collects extensive market data, lending confidence to these figures Strong CPU demand and AMD’s success are being driven by the recent surge in agentic AI, which is increasing the number of CPUs required in new deployments to almost match the number of GPUs. The traditional setup, where one CPU was paired with four or even eight GPUs, is shifting toward a one-to-one ratio of CPUs to GPUs in agentic AI deployments AMD is selling every CPU it produces. Intel is also seeing strong demand, even selling dies located on the very edge of the silicon wafer, which would normally become scrap, to eager customers. However, AMD currently appears to be achieving higher ASPs for its products

What's happening right now in our capital markets is going to DESTROY the retirement savings of millions of Americans. Anyone of good conscience needs to rise up and say enough. This must be stopped. I don't say that lightly. I've been doing this for 45 years, and what's happening right now to the integrity of our capital markets is unlike anything I have ever seen. This is not about Elon Musk or Donald Trump. This is not about whether you like rockets or hate rockets. This is about the systematic CORRUPTION of the financial system that every American depends on for their retirement. In the entirety of its existence, Tesla has generated approximately $36 billion in cumulative profit. That includes over $20 billion in government emission credits and tax subsidies. The company is valued at $1.7 trillion and its CEO is the richest man on the planet. I'm not talking about the stock price. I know the stock has made people money. That's the popularity contest. I'm talking about whether this company creates enough economic value to JUSTIFY the capital invested in it. And it doesn't. The returns on invested capital have been chronically below what any serious investor would demand. That's not wealth creation. So the product here isn't the car. The product is the STOCK PRICE. Elon Musk is selling hopium and an entire generation of investors is buying it without even knowing what a PE ratio is. I posted two pieces recently on Tesla and SpaceX. Each got over 1.5 million impressions. Thousands of hate replies but NOT ONE response with an actual argument. Not one. It was all "Libtard" and "Elon derangement syndrome." You would not get past a first-round interview at Fidelity thinking this way. But Tesla is just the opening act... SpaceX just filed for a $1.75 TRILLION IPO. $15 billion in revenue but no profit in sight. The private valuation was walked up from $200 billion to $400 billion to $800 billion to $1.75 trillion in two years. And Reuters has confirmed that SpaceX made early inclusion in the Nasdaq-100 a necessary condition for listing on the exchange. Nasdaq obliged by adopting a "Fast Entry" rule in March that lets mega-cap IPOs join the index after just 15 trading days, completely exempt from the normal seasoning and liquidity requirements every other company had to meet. And this matters because over $600 billion in passive funds track the Nasdaq-100. Unlike the S&P 500, which still requires months of seasoning and stricter float thresholds, the Nasdaq-100 is now a 15-day on-ramp for trillion-dollar IPOs. Every ETF and mutual fund benchmarked to that index will be FORCED to buy SpaceX within weeks of it going public regardless of whether the valuation makes any sense. Your 401(k) is literally the exit liquidity. You don't even get a choice. The structure of the market makes you a participant whether you want to be or not. That's what makes this different from every other bubble in history... You can't opt out. And the agencies that were supposed to protect you from exactly this? They're doing NOTHING. Peter Lynch would always say the product is not the stock and the stock is not the product. Show me one Hall of Fame investor who ever made his fortune chasing hype. Lynch, Druckenmiller, Soros, Buffett, Griffin, Cohen. Not one of them managed money this way. It's only the cult on X who thinks momentum and greater fool is an investment strategy. As Buffett said, in the short run the market is a popularity contest. In the long run it's a weighing machine. This popularity contest has gone on longer than any I've witnessed in my career. But gravity always wins. And when it does, the people who forced your pension fund into a money-losing rocket company at 120x revenue will have a lot of explaining to do. This must stop. And it WILL stop. The only question is how much damage gets done first. Are you listening?

关于今年的降息，4月份美国银行和JPM财报后我表达过出现了一些担忧，主要是3月份消费完全不见放缓，加上4月份美股的反弹，财富效应，高收入人群更不可能放缓了，然后结合今天的PPI数据，企业就可以把价格转嫁到消费者手里。我过去几天其实一直在思考“大科技撒钱”对于通胀的影响，在生活科技产品上已经看到了，有没有外溢到其他地方需要观察。但从宏观视角看：消费强劲+大科技撒钱 = 通胀风险。关于就业，我年初的观点并不改变：仍然有下行风险。下半年美联储可能会面临很难的抉择，结合下个月美联储的转向（降息倾向到中性），美联储这艘船已经在动了，降息周期的利好已经在尾声了。现在市场反应不过来，因为还和硬件在热恋中，估值继续上升后，下半年风险陡增。所以不要怕什么踏空的，机会是等出来的。