绿箭牌2.5元

For almost two years, the entire AI story was told through two characters: the GPU and the memory chip. Nvidia got the headlines, Micron got the memory trade, and almost nobody outside the industry could name the company that actually connects thousands of those GPUs into one working brain. Then at COMPUTEX, Jensen Huang said the name out loud. Marvell. And the stock jumped 32% in a single day. Here is how I would explain it to a friend who does not follow chips. Imagine a city obsessed with building taller and taller skyscrapers, the GPUs. Marvell is the company quietly laying the high-speed expressways between every building. You do not notice the roads until someone points out that without them, the city is just a row of beautiful, disconnected towers. What does this mean for us ordinary investors? Two things. First, the AI buildout is wider than the few names everyone already owns. The plumbing has value too. Second, by the time a CEO says your name on stage and the stock is already at an all-time high, the easy part of the story is behind you. The real question is no longer what Marvell does. It is whether the orders keep coming after the applause fades.

N1X specs: 20-core ARM CPU, 6,144 CUDA cores (RTX 5070 class), Blackwell architecture, 3nm, 45-80W. Partners: Dell XPS, Lenovo Legion 7, Asus ProArt, MSI, Microsoft Surface. Holiday 2026 launch. If benchmarks match the specs, Qualcomm's ARM Windows monopoly is over.

Compared to BIDU's AI hype, BABA feels undervalued, anyone thinks so?🤔

$Coinbase(COIN.US) Moves with Bitcoin, not the stock market. A different beast on its own schedule

$Taiwan Semiconductor(TSM.US) talking about power as a bottleneck is a reminder of the immense challenges Intel faces in its foundry build-out

This is ABSOLUTELY MASSIVE. Micron Technology just crossed USD 1 trillion market cap today, surging +19%. Biggest single-day gain since 2011. And before you think this is just hype — UBS just raised their price target from USD 535 to USD 1,625. That's a 204% increase in a single price target revision, the highest target on Wall Street, from one of the most respected semiconductor analysts alive. Their argument: "AI has permanently changed how Micron deserves to be valued." 🚨 Here's the data you need to understand why this is different. Micron's Q2 FY2026 revenue was USD 23.9 billion — up 196% year-over-year and 28% ABOVE their own guidance. EPS was USD 12.07, up 756% year-over-year. And CEO Sanjay Mehrotra confirmed that their entire 2026 HBM4 supply is sold out under long-term fixed-price contracts. They are only fulfilling 50 to 65% of customer demand. There is no more product to sell. Every AI data centre buildout in the world is queuing for memory that Micron literally cannot make fast enough. Q3 guidance is USD 33.5 billion in revenue at 81% gross margins. A year ago these margins were in the mid-30s. When margins go from 35% to 81% in four quarters, that's called pricing power. AMD caught the wave too today, up +7%. The whole AI chip supply chain is moving. If this is the signal you've been waiting for that the memory supercycle is real — this is it. 📈

$ASML(ASML.US) The stock price had already started to fall before the earnings report was released. Waiting for the results.

Costco's Q3 2026: A Master Class in Why Business Quality Matters More Than Valuation TimingI'll be upfront: Costco has never been cheap by traditional valuation metrics. It trades at a significant premium to most retailers. And yet, every time I've looked at selling on valuation grounds, the business has continued compounding in a way that made the timing call feel irrelevant. Q3 FY2026 reinforced this again. Net sales up 11.6% to USD 69.15 billion. Comparable sales up 9.8%. E-commerce comparable sales up 21.5%. EPS $4.93 versus $4.28 a year ago. Membership renewal rate 92.2% in the US and Canada. The membership model is the thesis. Costco collects USD 1.37 billion in membership fee income that it essentially treats as pure margin contribution before selling a single product. The retail operation runs at deliberately thin margins; the membership fee is where the business earns its profitability. This means Costco's incentive is always aligned with the customer: keep prices low to retain memberships, not to maximise product margin. With 41.2 million paid executive members growing at 9.6% YoY, the recurring revenue base keeps expanding independently of any individual quarter's sales performance. The long-term investor mistake with Costco is spending too much time on whether the P/E ratio looks elevated today. The right question is whether Costco will have more members, higher renewal rates, and more executive tier subscribers in five to ten years than it does today. Every quarter's evidence points to yes. That's why I hold it and don't trade around it.

CoreWeave’s Nvidia backing shows AI infrastructure race, MU memory tightness plays into it.

$Trump Media & Tech(DJT.US) Could it go higher? Sure. Could it go to zero? Also possible😅

$Oracle(ORCL.US) One of the worst performers yesterday, down over 5%. The lawsuit overhang + market panic = brutal combo

Net income down 11%. Revenue growth at 12.5%, well below the 20% consensus. On paper, this was a weak print.The stock moved higher. EPS came in at USD 2.36 against a consensus of roughly USD 2.13. Options had priced in a larger downside move. Short interest was elevated going into the report. When the number didn't crater, covering did the work.The setup now: this is a relief bounce, not a fundamental re-rating. Revenue deceleration at PDD is a multi-quarter trend. One EPS beat doesn't change the trajectory. Watch whether the stock holds above the pre-earnings level across the next three sessions. If it fades back below that level, the bounce was a fade opportunity, not a buy signal.