Looking at the earnings reports of $DELL and $AVGO together, I think the signal is already very clear: The next phase of AI hardware will no longer be just GPUs. The most striking thing from Dell is: AI server revenue of $16.4 billion, +100% Quarterly AI server orders of $60.9 billion Backlog directly reaching $95 billion This indicates that enterprises and cloud providers are still frantically expanding AI servers. AVGO is even more direct: AI semiconductor revenue of $16.7 billion, +221% Future AI revenue targets are projected as: FY27 about $115 billion FY28 about $230 billion Putting these two earnings reports together actually explains the entire industry chain. More servers → More GPUs/XPUs → Greater demand for HBM/DRAM → Higher demand for network switching → More 800G/1.6T optical modules → Increasing data, so SSD/NAND will also expand → Finally, more power and liquid cooling are needed. So now when I look at AI hardware, I no longer focus solely on $NVDA. The group that benefits most directly instead includes: SK Hynix, $MU — HBM/DRAM $ANET — AI networking $LITE, $COHR — optical communications $SNDK — enterprise SSD/NAND $VRT, $ETN — power supply and liquid cooling Dell proves that: AI server demand has not stopped. AVGO proves that: Custom XPUs and AI networking are still accelerating. Both earnings reports jointly confirm one thing: AI CapEx is not over. Funds are just shifting from the earliest GPUs to: Memory → Networking → Optics → Storage → Power Spreading layer by layer. So the real next valuable target may not necessarily be "the next NVDA." Instead, it is: Which hardware will be the next to start experiencing shortages.

SOL holds firm at 100, XRP's funds are solid, Meme's rebound depends first on volume! $SOL is currently around $100.6, with the 100-dollar mark once again becoming the dividing line between bulls and bears. The latest daily spot ETF still recorded a net inflow of about $400,000, indicating that institutional funds have not completely withdrawn. The problem remains that the previous rise was too fast: if it holds above 100, we can look at 103–105; if it falls below 98, be prepared for profit-taking. $DOGE is about $0.0828, up about 1.4% intraday, but its performance over the past week remains weak. There is no new strong catalyst now, mostly just sentiment recovery. If 0.08 holds, there is a chance to test 0.085; without volume support, I won’t consider this rebound a trend restart. $XRP is about $1.36, with the latest daily spot ETF net inflow around $7.69 million, showing the funding side is stronger than the price suggests. The 1.32–1.34 range is a support zone; if it holds above 1.38, look for 1.42; as long as the ETF doesn’t turn to outflows, this looks more like a rotation rather than structural weakness. Looking at others, $HYPE is about $82.2, with whale sell signals suppressing the short term, 80 is the defense level; $BOME is about $0.00089, up over 2% intraday but down over 12% in seven days, don’t chase without volume; $TRUMP is about $2.23, with team-related wallets showing about $26.65 million BitGo transfers, rebounds still need to guard against potential selling pressure. What small coins lack now is not stories, but sustained trading volume. #Solana主网提速，节点门槛会否上升？

Earnings report exploded $DELL Paired with Trump's strict selection Indeed unbelievably strong

What the hell is going on? Suddenly, a straight shot north? Are the gambling tables in Japan and Korea this shady? The East Asian trio, huh? 🤣 $SKHY got hit while just lying there

Another AI stock exploded last night. $SNOW surged over 20% in after-hours trading following its earnings report. But what I find most interesting about this report isn’t how much it rose, but that it might contradict the hottest view of the past six months: AI will kill SaaS. Let’s look at the numbers: Revenue $1.55 billion, +35% Product revenue $1.49 billion, +37% Adjusted EPS $0.62, expected $0.45 RPO $9 billion, +30% The full-year product revenue guidance was raised directly from $5.84 billion to $6.07 billion. More importantly, management said: Recent growth acceleration, about half comes from AI products. I think this sentence is very important. Because when enterprises truly start using AI, the first question isn’t "Is the large model smart enough?" but: Where does AI get the company’s data? Customer information, financial data, orders, inventory, historical documents—all scattered inside the company. And what Snowflake does is centralize this data, enabling AI to truly access it. So AI won’t necessarily kill all software. On the contrary, it might make software that controls core enterprise data more valuable. Previously, Snowflake sold: data warehouses. Now it wants to sell: the data foundation for enterprises in the AI era. If this AI market wave starts spreading from hardware to software, I think $SNOW’s earnings report is a very noteworthy signal. The next phase of the market might no longer just ask who sells GPUs, but who controls the data AI truly needs.

Many people new to Crypto see the market for the first time: BTC, ETH, SOL, BNB, XRP—a bunch of coins that seem no different except for their prices. But the logic behind them is actually very different. A chart to understand mainstream Crypto: Asset｜You can simply understand it as bitcoin:native｜Digital gold / scarce asset $ETH｜Smart contract settlement layer $SOL｜High-performance public chain binancecoin:native｜Exchange ecosystem + BNB Chain ripple:native｜Cross-border payment network asset $DOGE｜Meme + community consensus Stablecoins｜On-chain “USD cash” The most important thing: Don’t analyze all coins using BTC’s approach. BTC’s core is liquidity, ETF funds, macro risk appetite, and long-term holder behavior. ETH focuses on on-chain activity, stablecoins, DeFi, Layer2, and fees. SOL looks more at trading volume, active applications, new projects, and Meme ecosystem. Meme coins are even more extreme: Often, studying the “fundamentals” for a long time is less effective than studying where the attention flows. So the first thing after entering Crypto shouldn’t be: "Which coin will go up?" But first understand: What asset am I actually buying now, and why does it have value?"

If a stock I particularly favor suddenly drops 20%, I won't immediately buy more. I will first check four things. First, whether the earnings forecast has been revised downward. The price has dropped, but if next year's EPS is still being revised upward, that's completely different from the price dropping along with earnings. Second, whether the company's guidance has changed. If management is still raising revenue and order expectations, I am more inclined to interpret the drop as a valuation issue. Third, whether the entire industry is falling together. If $MU and $SNDK both fall, it might be a risk-off in the storage sector; if only one company crashes, be cautious that it might be a company-specific problem. Fourth, what exactly are institutions selling. If even core assets like $NVDA and $GOOGL are being sold off together, it might be a move away from risk; if the market is stable and only your stock is falling all the way, then don't rush to blame macro factors. So when I see a -20% drop now, my first reaction is no longer: "Finally, it's on sale." Instead, it is: Is the market giving me an opportunity, or reminding me that my original logic was wrong? Cheapness is not caused by price drops. Cheapness is: Price dropping faster than fundamentals.

$AVGO This earnings report, I think the real positive impact is not just on Broadcom itself. It’s more like telling the market: The next round of AI funding is continuing to spread from GPU to Memory, Networking, and Optics. If we follow down from AVGO’s earnings report, the five I’m most focused on are: SK Hynix, $MU, $ANET, $LITE, $COHR. Let’s start with storage. AVGO clearly mentioned this time that future XPU expansion is not only about chip shortages but also about considering HBM and system memory supply. If AVGO’s AI revenue can really grow from about $58 billion in FY26 to about $115 billion in FY27, and even about $230 billion in FY28, the corresponding demand for HBM/DRAM will only increase. So the most direct read-through is: SK Hynix + $MU. The second is networking. AVGO’s AI Networking revenue is still growing rapidly, which indicates that as AI clusters get bigger, the bottleneck is no longer just computing power. Hundreds of thousands of GPUs/XPUs must communicate at high speed. This directly benefits: $ANET. Next layer down is optics. As switch ports increase and speeds upgrade from 800G to 1.6T, demand for optical modules and lasers will rise together. So: $LITE and $COHR are also very direct beneficiaries. As for $SNDK, I wouldn’t rush in immediately because of this earnings report. NAND and enterprise SSDs will definitely benefit from AI data growth in the long term, but the clearest signal AVGO released this time is still: HBM/DRAM > Networking > Optical communication. So my current understanding of this earnings report is not: "AVGO’s earnings report is great." But rather: The second phase of AI CapEx is continuing to spread. The first phase of the market only focused on $NVDA. The next phase with real elasticity might be: Who provides memory, networking, and optical connections for these AI chips. If I had to rank an order: SK Hynix > $MU > $ANET > $LITE > $COHR

I think many people have actually misunderstood $AVGO. Broadcom's biggest opportunity is not to "replace $NVDA." Precisely because companies like Google, Meta, and OpenAI don't want to always just buy NVDA. Latest financial report: Q3 revenue $29.59 billion, +86% AI semiconductor revenue $16.7 billion, +221% Q4 AI revenue directly guided to $21.7 billion, +236% This growth rate is no longer that of an ordinary AI beneficiary stock. AVGO is currently capturing two streams of revenue: First: Custom AI Chip. After super tech companies invest tens of billions annually in AI, their motivation to make their own ASICs grows stronger. It's not because GPUs are bad, but once the scale reaches a certain level, saving 10% on cost and power consumption each means tens of billions of dollars. Second: AI Networking. The more chips, the larger the cluster; the demand for switching chips, SerDes, DSP, and other connecting devices becomes even more tremendous. So the most comfortable scenario for AVGO is actually not NVDA falling. Rather: NVDA continues to sell GPUs, tech giants continue to develop their own ASICs, and the entire AI cluster keeps expanding wildly. No matter which path wins, Broadcom has a place. This is also why I increasingly feel: In the first phase of AI hardware, everyone only looked at NVDA. The next phase truly worth studying is— Who is helping tech giants reduce their dependence on NVDA? Currently, $AVGO is one of the core answers.

If I were to bet on the next sector that might take over the AI chip surge, I wouldn't choose GPUs again. I would choose: AI power grids + electrical equipment. And the one I want to focus on most right now is $ETN Eaton. Why? Because the core question for AI in the past two years has been: "Are there enough GPUs?" But I think the next phase will gradually become: "Is there enough power?" Microsoft, Meta, Google, Amazon can continue to invest hundreds of billions of dollars building data centers, NVDA can continue to deliver more GPUs, but ultimately all these servers share one common premise: You have to get the power in. And this is exactly what ETN does. Transformers, distribution equipment, switchgear, UPS, power management... From the power grid connection to the AI data center, to finally delivering power to the server racks, a large amount of equipment cannot bypass this layer. More importantly, I like it now not because the "AI power shortage" story sounds good, but because orders have already started to be fulfilled. So if I were to rank the next wave of AI diffusion: First wave: $NVDA, $AVGO — computing power Second wave: $MU, SK Hynix, $SNDK — storage Third wave: $LITE, $COHR, $ANET — optical communication/network Next wave, I will focus on power. My order of attention inside this is: $ETN > $PWR > $GEV ETN sells equipment, PWR is responsible for actually building the power grid, GEV is responsible for power generation + power grid infrastructure. And now ETN has actually dropped from its previous high. This is exactly the position I like: The logic hasn't disappeared, the performance is still being realized, but the sentiment is no longer as crazy as before. If the market trades AI CapEx diffusion again in the next round, I believe what is most likely to be rediscovered by capital is not necessarily another chip. It might be a transformer you never even looked at before.