海盗鸭

A New Mexico judge has just ordered Meta to pay $567 million to the governance fund after a jury found the company violated the state's unfair conduct law. For $METAB , the amount may not necessarily cripple its cash flow, but the risk in this case is not just a fine: the funds are explicitly earmarked for addressing child harm, meaning regulatory focus is shifting from content disputes toward measurable remedies. Other states could easily follow this playbook. In the past, when platforms faced lawsuits, the market often treated legal costs as one-off expenses; if similar rulings keep recurring, the issue becomes a long-term compliance cost—covering age verification, adjustments to recommendation algorithms, data retention, and limitations on product design. Even more troublesome is that protecting minors often faces less political resistance, making bipartisan action more likely than with general tech regulation. In the short term, the stock price may not change direction because of $567 million, but investors should watch whether Meta appeals and whether the ruling could trigger class-action lawsuits or additional state-level cases—that's what will determine whether the risk can be capped.

#科技监管 #child online security

After Sandisk and Western Digital released their earnings reports and the stock prices plunged, Micron also fell, but clearly managed to dodge the worst part of this round of storage-chip selloff. This difference contains more information than “the whole sector is down.” $SNDKB and $WDCB are more likely to be priced by the market based on NAND prices, consumer-electronics demand, and inventory-cycle dynamics. Once the earnings report fails to provide stronger guidance, the earlier rally gains are easier to cash in. $MUB , while still unable to escape the storage cycle, has an extra layer of DRAM and AI server high-bandwidth memory expectations, so investors are temporarily more willing to tolerate a higher valuation. However, holding up better doesn’t mean the fundamentals fully decouple. If NAND weakness continues to spread into enterprise storage, or if the growth rate of AI capex slows, Micron’s valuation will be reassessed. Next, I will compare the three companies’ inventory levels, gross margin guidance, and expansion plans. Whichever one proactively increases supply first may be able to press down the newly repaired price cycle again.
#存储芯片 #美股财报 #Semiconductors

Market maker Wintermute has just obtained a U.S. brokerage dealer license and publicly said it wants to use this to position itself for the growth of tokenized securities in the U.S. On the surface, the news is about licensing progress; behind the scenes, however, crypto liquidity is beginning to actively work within traditional securities regulations. $ETH is very likely to become an important object of observation for this on-chain settlement route. In the past, market makers mainly profited from price spreads in crypto spot and derivatives. Now that it has become a licensed broker, it can access a wider range of issuers, institutional clients, and compliant products. This also indicates that industry competition is shifting from “who has crypto” to “who can legally connect on-chain activity with Wall Street.”

That said, a license is not a guarantee of revenue. Tokenized securities are also constrained by custody, investor eligibility, on-chain privacy, and secondary-market liquidity. It’s worth noting that the true beneficiaries may not be a specific token first, but rather an integrated institution that can simultaneously provide issuance, clearing, and market making. Next, watch whether Wintermute has actual products and trading volume, and which chain the underlying assets ultimately land on—this is more valuable in terms of trading signals than merely focusing on the completion of registration.

#代币化证券 #Ethereum

Micron’s stock price once plunged 31%. What’s unusual is that Wall Street didn’t follow suit by turning bearish; instead, it continued to back $MUB . This split is more worth watching than a simple drop: the market is trading storage cyclicality that has already peaked, while analysts bet that AI-server–driven high-bandwidth memory demand is still expanding. Both sides are essentially right—just on different time horizons. After years of trading cyclical stocks, what I fear most is “very good earnings, but even higher expectations.” The moment guidance on the quote, gross margin, or capex loosens even slightly, valuation will be the first to get cut. Meanwhile, a 31% decline also means many pessimistic expectations have been priced in quickly. Going forward, don’t just look at institutions’ target prices—focus on whether HBM supply is being delivered as promised, and on changes in traditional DRAM pricing and inventory. If demand is strong but prices still weaken, it suggests supply expansion is sprinting ahead, and the near term will likely remain under pressure.

#AI芯片 #US stock earnings reports

At a SpaceX earnings call, Musk specifically mentioned Nvidia, and the market immediately linked three trading entry points—$SPCXB , $NVDAB , and $TSLAB . What is truly worth separating is not which company he praised, but the fact that SpaceX and Tesla are both competing for the same scarce resources: high-end GPUs, power, data center capacity, and engineering talent. What Nvidia gains is not just a single customer order, but a demand overlap created by simultaneous expansion across aerospace, autonomous driving, and generative AI. However, the larger the customer, the stronger the motivation to develop in-house chips and push prices down—so the long-term relationship is not a one-way benefit. Musk’s remarks often boost short-term sentiment, but the financial report ultimately comes down to procurement scale, delivery timelines, and capital expenditures. If SpaceX continues to ramp up its AI efforts, the pressure from losses may show up first; Tesla, meanwhile, needs to prove that its compute investment can translate into product revenue. For Nvidia, the hardest proof is still order visibility and gross margin—not a single endorsement on a call.

#SpaceXAI支出拖累首份财报 #AI芯片

SpaceX’s first financial report disclosure shows a $540 million loss in its Bitcoin holdings, bringing $SPCXB and $BTC directly together in one balance sheet entry. Many people will read this as “bearish on Bitcoin,” but I don’t think it’s that simple: the book loss reflects the purchase cost, the valuation timing, and the accounting treatment—it doesn’t mean the company has already cut its losses, and it doesn’t indicate problems with the rocket business. The real question is: for a space company with heavy cash burn, why allow a highly volatile asset to tie up liquidity? SpaceX still needs to keep investing in launches, Starlink, and AI infrastructure; when the Bitcoin price falls, the holdings can’t reliably cover expenses and may even amplify financial-report volatility. On the other hand, if the company hasn’t sold, that $540 million loss could also reverse as the price recovers. In the short term, it’s easy to panic at the headline; over the long run, what matters more is the size of the holdings, whether impairments occur or assets are sold, and whether operating cash flow can support expansion.

#SpaceXAI支出拖累首份财报 #Bitcoin regains the $64,000 level

ElizaOS’s token plunged 19% in a week to a historic low. The reason: founder Shaw Walters directly announced that this coin was "dead"—the project moved all remaining treasury funds to settle lawsuits with token holders, and then Eliza continued, but no longer with its own token. Frankly, this is pretty clear-headed. But for people holding the coin, it’s a disaster. Last year, the AI agent track was hyped into the sky; now one by one the charade is exposed. Many so-called AI coins have no real cash flow backing their valuations—they’re propped up purely by narrative and emotion. Concept coins like the one at $SOL are seeing a sell-off faster than the pace at which they once surged. The lesson is clear: in projects where a founder’s single statement can reduce a token to zero, what you hold is not an asset—it’s faith. There will be more on-chain "death announcements" like this. Don’t catch the thrown knives.

Lenovo has recently followed AI server orders from $AMZNB and its orders surged 73%. Now ahead of the earnings report, it’s also laying out plans. For this kind of “Amazon partner” concept stock, the logic is actually quite simple: cloud giants expand capacity, and the supply chain profits. But there’s risk too—once Amazon cuts orders or switches suppliers, the stock price will drop immediately. Last year’s Q3, Lenovo already fell more than 15% because Dell won the orders. At this position ahead of the earnings report, it’s recommended to wait for the data before taking action—don’t jump on the train early and get trapped by a mismatch between expectations and reality.

Analysts have set a target implying a 44% upside for Amazon stock ($AMZNB ). It sounds tempting. But these target prices are calculated based on a bunch of assumptions—whether cloud computing growth can be sustained, whether e-commerce profit margins can improve, and when AI spending will start paying off. If any one of these variables deviates from expectations, that 44% comes down. This isn’t to say the stock can’t rise—just don’t treat analysts’ optimistic expectations as guaranteed returns. You still need to do your own homework.

Google Cloud experiences explosive growth, with the price being the loss of core AI talent that was built to power it—commercialization takes priority, and cutting-edge research takes a back seat. The signal is subtle: back then, the DeepMind teams were there to “build the strongest model.” Now the company wants quarterly revenue and cloud orders. The philosophies don’t align, so people leaving is only a matter of time. In the short term, cloud revenue is up and the stock has support. But in the AI race, what matters is long-term talent density. The loss of key people is like chronic blood loss—it won’t show up clearly in the financial statements for now. $GOOGLB I’m not pessimistic in the short term; the cloud cash flow is real. But if OpenAI and Anthropic continue to siphon Google’s people, then in two or three years the moat may have to be called into question. Big players’ worries are never written into the numbers for a given quarter.

#AI芯片 #U.S. stock earnings report

Marvell’s call options order book shows abnormal large orders, with trading volume clearly deviating from the norm. Institutions are betting that MRVL will move upward in the short term. This kind of unusual options activity is not something I would normally treat as a buy signal, but it at least indicates that big money is positioning early. The issue is that options are a leverage game—you're betting on both direction and timing. If the expiration dates don’t line up, everything can still end up worthless. $MRVLB itself sits on the AI network chip trend, so the logic isn’t flawed. But unusual options moves are often the shadow of something material hitting the tape early—either expectations ahead of earnings, or rumors of a merger/acquisition/orders. The thing you must not do when following orders is just see “large bullish calls” and rush in. First, figure out which catalyst they’re betting on and for what time window. This signal is worth noting and then comparing against how the stock moves over the following days.

Oracle’s AI bet this round is a bit too aggressive. To expand data centers, it has taken on massive debt to the point of putting its own credit rating on the line—gambling that it will pay off. This is a typical high-risk strategy of either “one winner takes all” or having cash flow crushed by interest. Oracle’s logic is to seize capacity while AI compute demand is booming, lock in large orders with customer commitments. The problem is that these capital expenditures must be repaid with future cloud revenue; in the meantime, if AI demand growth falls short of expectations, credit downgrades and rising financing costs would trigger a chain reaction. A strategy like this—using credit rating to buy growth—does have a chance of winning, but it leaves the company with very little room for error. It’s completely different from the old Oracle image of being steady and emphasizing stability. If you want to get on board, think it through first: are you buying a transformation story, or a financial safety cushion. $ORCLB

The Dow and the S&P 500 hit new highs again, and behind it is a broad rebound in tech stocks that’s propping up the market. Before SpaceX reports earnings, it surged 9.8%, adding fuel as well. This kind of broad rally looks exciting, but I care more about its sustainability—new highs in the overall index often concentrate in just a handful of heavyweights, and if the breadth isn’t there, the move can be “puffed up.” Right now, capital is clearly flocking into the AI narrative; if a leading company’s earnings report turns sour, sentiment can flip quickly. With the index already at high levels, the cost-effectiveness of chasing is falling. If you want to follow the broader market, use something like $SPYB —it’s more worry-free and you don’t have to gamble on a single stock’s earnings-report “blind box.” $SPYB $SPCXB
#SpaceX surged 9.8% before earnings

A bunch of institutions suddenly and collectively upgraded SK Hynix to a “buy” rating, and their reasons all point to a shortage of HBM high-bandwidth memory. This signal is more interesting than a single firm calling it out—this storage-chip cycle is different from the past. AI servers’ demand for HBM is structural, not a game of price swings like ordinary memory. SK Hynix bet on HBM earlier than its peers, so it’s now reaping the first-mover dividend. As the storage sector climbs out of last year’s trough, it’s customary for the leading players to benefit first. Keep an eye on how Micron follows suit and the timing of that momentum—the whole sector has strong linkage effects. $SKHYB $MUB
#AI chips

The analyst has just lowered the $DELLB target price. The reason cited is that signs of a slowdown in enterprise IT spending are becoming increasingly clear. Dell has been riding high on the AI server concept over the past two years, but actual order conversion has not been as fast; customers are still waiting and watching. The PC-side business is also a traditional cyclical market, and right now it is precisely at the low point of the replacement cycle. In the short term, the stock price may continue to be revised downward along with earnings expectations.

Ripple This move is pretty interesting—XRP holders can now directly stake their XRP to borrow stablecoins. On the surface it adds another use case, but in reality it pushes XRP from being a “pure speculative token” toward “collateralized assets.” For long-term holders, it’s a signal: you don’t have to sell your coins to get liquidity, which effectively reduces selling pressure. But on the other hand, lending is a double-edged sword: if the price drops and the collateral ratio collapses, you’ll still see cascading liquidations—the DeFi script is something we’ve seen way too many times. The $XRP wave isn’t so much a price catalyst as it is narrative reinforcement; the key things to watch are on-chain locked-in amounts and the borrowing scale. It’s good to see use cases landing, but don’t directly translate “you can borrow money” into “it will go up”—they’re two different things.

Korean bitcoin holders have almost not been affected by the Coldcard hardware wallet vulnerability; analysts say it was "cultural habits" that saved them. What exactly are these habits? Multi-signature, keeping hot and cold separation, and not putting all your eggs in one basket—these basic security principles are fairly well practiced by Korean old-timers. In contrast, some people want to save time and put their entire life savings into a single hardware wallet; when something goes wrong, it’s all gone. $BTC Safety always comes first. No matter how advanced the technology is, it can’t stand up to human oversight.

The figure that Amazon uses to make AI spending look “less dangerous,” refers to the profit margin of its cloud service AWS. In its earnings reports, AWS gross margin has stayed steady at above 30%—even if it’s spending heavily on GPUs and building data centers, its cash flow is still positive. This is different from Meta and Microsoft—whose AI investments are basically just pure costs. This $AMZNB model essentially turns AI spending directly into cloud service revenue, meaning customers end up helping it share the risk. However, AWS growth is slowing, and competitors like $MSFTB ’s Azure are also grabbing market share. In the long run, it’ll still come down to who can get AI applications deployed first.

SoftBank’s Q1 earnings beat expectations, and the core reason is that its investments in $INTCB have an unrealized gain of $8.2 billion. At this point it’s worth keeping an eye on—Intel’s share price is still struggling at low levels, yet SoftBank can make so much off its books. That suggests its original acquisition cost was low enough. Looking back at SoftBank’s positioning in tech stocks over the past few years, it definitely timed its moves better than chasing after hype. However, an unrealized gain is not the same as cashing out. If SoftBank is really going to realize it, it will depend on whether Intel can maintain the momentum in its process technology catch-up.

Bezos plans to sell $4 billion of stock with the ID $AMZNB , but there’s a “catch”—the money isn’t being withdrawn directly; it will be used to fund Blue Origin and other projects. When the market hears “a $4 billion reduction in holdings,” the first reaction is to sell, no matter where the money ultimately goes. For retail investors, a founder’s continued selling is a bearish signal. Even if the company’s fundamentals are fine, the share price may face downward pressure in the short term. If you’re thinking about buying the dip, you can wait until this wave of selling pressure has been digested.