熊什么猫

Nvidia’s Q2 earnings report essentially sets the tone for the recent AI market run.

Quarterly revenue came in at $96.2 billion, up 106% year over year. Data center revenue was $89.0 billion, up 117%, and remains the absolute growth driver.

Its Q3 revenue outlook is $108 billion, above market expectations. More importantly, guidance for roughly 70% revenue growth in the next fiscal year directly eases prior concerns in the market about AI capital expenditures peaking.

Nvidia also clearly said the outlook is constrained by supply rather than insufficient demand. I think JPMorgan’s view that the guidance is somewhat conservative has a point.

After the earnings release, Wall Street collectively raised its price targets. Goldman Sachs lifted its target to $300, Citigroup to $315, JPMorgan to $320, and Bernstein significantly to $400.
Multiple firms followed with upgrades. Divergence among sell-side analysts about the durability of AI chip demand is rapidly narrowing.

On Thursday, Nvidia’s stock surged 8.7%, marking its biggest single-day gain since April 2025. Its market value increased by about $442 billion in one day—an all-time second-highest record for a global single stock’s one-day growth, only behind Microsoft’s $450 billion increase a little under a month ago.

The company’s total market capitalization is currently about $5.5 trillion, still the largest among all publicly traded companies worldwide.

A few months ago, the market was debating whether AI trading had gotten overheated and whether cloud providers’ capital expenditures would deliver questionable returns. One earnings report has flipped overall sentiment.

Previously, the market was pricing expectations. Now it’s gradually moving into the phase of performance validation, and the logic behind valuation has changed.

Going forward, the market’s focus will shift. The production ramp of the Rubin platform, whether high gross margins can be sustained, whether China data center revenue will recover, and whether potential commitments will rise on the balance sheet under the partnership financing model.

These variables will all influence the stock’s subsequent trajectory.

The anchor has been set for the AI infrastructure chain—but that doesn’t mean things will be smooth from here onward.

After BTC surged past 80,000, market risk appetite was clearly boosted, and the Meme sector has once again become the direction of incoming capital.

What’s more, it’s not just a single chain rallying. Solana, BSC, and Robinhood Chain are all producing their own dark-horse moves, and the narrative is switching very quickly.

As of the time of writing, the speed of Robinhood Chain’s breakout has been more intense than I expected.

CASHCAT has already touched near its all-time high. Its current price is about $0.209, market cap is $205 million, up 46% in the past 24 hours, and also up 13% over the last 6 hours—the follow-through isn’t weak.

PONS is moving in sync as well, up 46%, with a current price of $0.091 and a market cap of $64.8 million. It’s the platform token of the on-chain mainstream Meme launchpad ecosystem. In essence, it’s benefiting from ecosystem traffic, and it’s not quite the same logic as Memes that are purely consensus-driven hype.

Earlier on, Meme’s capital focus was still on the Base chain. When Basecat was hot, it was even treated as a representative of the Base Builder culture, and it even made it onto Coinbase—so it was a notable “breakout into the mainstream” candidate at the time.

The market has always preferred the new and disliked the old. Once Robinhood Chain’s “platform-native” narrative came out, capital attention shifted quickly. Those older hot Memes on Base still have community foundations, but new capital and discussion have clearly been diverted; the hype has already slipped from the main storyline to the margins.

On-chain data also supports the heat. The stats show that Robinhood Chain’s DEX trading volume was $645 million over the past 24 hours, $3.68 billion over the last 7 days, and the 7-day month-over-month increase was 23.9%.

If you compare horizontally, daily trading volume is roughly 22% of Solana’s and 40% of Ethereum’s. For a new chain, this scale is already extremely huge.

Also, on the BSC side there’s a new Meme called SUE, whose 24-hour gain directly hit 5,910%. Its current market cap is only $2.09 million—an archetypal early-stage, high-volatility product. Solana’s BATON is also gradually spreading in the community. Its current market cap is $1.6 million; its all-time high touched $2.5 million. In just the last 6 hours, trading volume reached $9.5 million, with very active turnover.

In short, the Meme market is very high-risk underneath. Take it in at your own discretion; if you really want to play it, position sizing definitely needs to be controlled.

NVIDIA's Q2 13F filing is out. Total holdings rose from over 18 billion in Q1 to over 63 billion, an increase of roughly 245%.

Don't assume that this huge jump means they added more shares in Q2. Besides the first time they entered the S&P application list with $SPCX , the share counts of the other seven stocks didn't change at all from Q1.

The increase in market value mainly comes from two factors.
First, position #SpaceX was included in the reporting for the first time, bringing in more than 20 billion directly.
Second, the existing holdings' stock prices went up, so the book value increased passively.

Concentration is very high. Intel remains the top holding, making up nearly half of the portfolio. SpaceX is newly added and ranks second, accounting for about one third. Together, the two companies make up about 80% of the entire portfolio.

One more detail that's easy to get wrong. These SpaceX shares were not bought on the secondary market. They came from equity previously invested in xAI, and later transferred when SpaceX exchanged shares to acquire xAI. The remaining holdings are smaller in size. They are all upstream and downstream players in the compute supply chain, including CoreWeave, Coherent, Synopsys, and Nokia.

NVIDIA $NVDA is already an industry leader. There's no need for it to make money just by buying and selling stocks for price differences. These holdings look more like building an ecosystem around its core business, with the key links of the industry value chain firmly in hand.

There are plenty of variables in the compute race. Once the industry runs a bit longer, just keep watching as the story unfolds.

Saylor said this really gets to the point. With the BIP-110 fork, the results came out very quickly. 99.85% of the hashrate stayed on the original chain, while the fork side only mined two blocks—already more than 80 blocks behind.

At the current pace, the forked chain would have to wait for the system to automatically adjust the difficulty, and would still need to mine 2015 more blocks. That works out to about 25 years—roughly half a lifetime.

“Consensus is something you fight for.” Saylor’s line is full of substance.

Many people think that a fork is just copying a piece of code, giving it a new name, posting a tweet, and then you have new coins. But whether a chain is worth anything is never about whether you fork—it’s about whether people are willing to follow you.

If the hashrate isn’t enough, there’s no security. No one would dare put big money on it. If nobody uses it, nobody has applications for it, then what’s the point of holding the coins? Without funds coming in, the price can’t be supported. Mining won’t be profitable, people will leave, the hashrate drops further, and you get a vicious cycle.

A chain nobody uses—no matter what you say about it, it’s empty. If you’re missing any of those things, the fork is just people playing separately.

#BTC How many times has this been forked over the years? Not many people can still think of it now. Most are already long gone.

Because forking is too easy, while consensus is too hard. Consensus isn’t something written on paper. It’s built vote by vote: miners vote with their hashrate, users vote with their money, and developers vote with their time. When you say you forked, it only shows you existed—it doesn’t mean anyone else recognizes you.

In today’s bear market, it’s especially easy to see this kind of thing. When the market is hot, any random fork can be hyped for a while—everyone is looking for new stories. When the tide goes out, what remains is still the stuff that genuinely has real users and people who truly believe in it.

For BIP-110, this probably won’t be mentioned much after a while. But I remember that line: you can fork freely, and the network can also freely choose not to follow you.

That’s probably the most awesome part of $BTC .

Have you tried investing in $USD1 ? I put a little in from the very beginning and haven’t really managed it much.

After a few months, the returns aren’t that high. But being able to have a steady inflow during a bear market—I’m already quite satisfied.

Just saw that Big Sis Jia is continuing, and then with USD1 × #WLFI she went to Binance Square to throw out money.

From August 4 to August 8—5 days. 20000 USD1 plus 600,000 coins of $WLFI , until it runs out.

Live stream tips: If you go live on Binance Square and chat about USD1 or WLFI, trading shares and market analysis all count. Big Sis Jia will randomly drop tips, and viewers in the room also have a chance to win red packets.

Chatroom red packets: The Chinese chatroom is set up. Red packet codes are dropped from time to time—just wait and watch.

Future surprises: Big Sis Jia said she’ll be updating on X over the next few days. Keep an eye on it—don’t miss out.

👉 Entry: 聊天室

Bear markets are tough, but #USD1 always manages to surprise us a bit.

CPO is pulling pretty hard today.

Mingpu Optoelectronics hit the daily limit, LianTe Technology surged 15%, and Tianfu Communications jumped 9%. The whole optical communications sector is moving up.

The reason is that NVIDIA has confirmed mass production. Gilad Shainer said it on a technical forum: the switches have started delivering to closely cooperating customers, and they are also rolling them out internally.

Citigroup previously raised its 2026 and 2027 capital expenditure forecasts for four major cloud providers to about $754 billion and $1.112 trillion, respectively. They think the period around the first-half performance releases could be a window to position.

After the sell-off in July, the optical module index dropped 30%. Earlier, the positioning was too concentrated. With this move, short-term risks have mostly been released.

The benefiting directions are very clear. Optical modules and optical components: JINJI XuChuang, Neway Technologies, Tianfu Communications, and Source Photonics. Server ODM/OEM: Foxconn Industrial Internet and Inspur Information. Liquid cooling: Infinik. Switching networks: Ruijie Network and Tsinghua Unigroup.

Also, some background. In June, SemiAnalysis published a report saying that large-scale CPO mass production would be delayed to 2028 or even 2029. Back then, it directly triggered a wave of panic selling. Today’s NVIDIA update is essentially a positive response to that.

That said, this time the “mass production” being discussed corresponds to small-batch onboarding by top-tier supercomputing customers in the second half of 2026, not a broad ramp-up. The real large-scale rollout will come after 2027. The 39 billion figure mentioned by TrendForce is for 2030. What’s being traded right now is still expectations. If performance is truly going to materialize, we’ll need to see how well Vera Rubin gets deployed.

My personal view is that the direction is correct. Today is more about sentiment repair plus news-driven catalyst.

#bnb Futu has gone live.

Hong Kong’s first licensed securities broker has opened the $BNB order book trading. Currently, it is only available to eligible professional investors in Hong Kong.

I think the key point here is that there is now another legitimate entry route.

That means people who usually trade Hong Kong and US stocks don’t need to go through registering with a crypto exchange anymore. They can simply buy $BNB directly in their Futu account. Market depth and data are real-time and transparent, and the experience is basically the same as buying stocks.

Plus, the fundamentals of #BNB are right there—public chain ecosystem, exchange endorsements, and so on. If you know, you know. Now that it’s been connected to traditional brokerage channels, it effectively expands the audience to the group of stock traders.

So for professional investors in Hong Kong, there’s one more allocation option.

With more licensed channels, in the long run it’s not a bad thing. BNB is awesome!

#币安

⚽️ This morning, both quarterfinal matches went to extra time, and it’s getting more and more exciting.

England faced Norway, and Bellingham stepped up again. In stoppage time of the first half, he equalized. Just three minutes into extra time, he finished a comeback brace to make it 2-1—sending Ha Bao straight home. On the other side, Argentina vs. Switzerland was just as tense: after surviving until the 112th minute, Alvarez struck a world-class long-range goal. Then Lautaro added another to seal the win, and Argentina took it 3-1.

The semifinals are set for the four teams: France, Spain, England, and Argentina. The semifinal schedule is out too—early on the 15th, France will face Spain, and early on the 16th, England will take on Argentina. Both matchups have plenty of past data: France vs. Spain met twice in the past two years’ major tournaments, and Spain won both. And you don’t need to talk about England vs. Argentina either—they’re longtime World Cup rivals.

I also checked the latest title probabilities on @predictdotfun. France is first at 39%, with Argentina and England tied at 20% each, and Spain at 19%. Right now, France is in a class of its own, but the other three aren’t far behind—anyone still has a chance.

Back to how the prediction market works. predict fun works by routing users’ collateral into the Venus protocol to earn yield automatically. While the prediction is waiting for results, the funds can generate additional returns. This differs from traditional betting models, where capital is passively locked.

The platform itself has no KYC requirement and follows a self-custody route. On top of that, BNB Chain’s transaction fee costs are relatively low. Trust Wallet and Binance Wallet are both natively integrated.

Now there’s less and less left of the World Cup—one match at a time. With every game, the probabilities change. What the market reflects is the current consensus, and honestly, it’s still hard to say who will ultimately lift the trophy. Who do you think has the better shot?

#世界杯 #预测市场

🦅 The use cases of on-chain stablecoins have long remained stuck at the transaction intermediary layer, and the actual user entry point for everyday users has never quite taken off.

When I saw WLFI Markets launch the Zebec SuperApp, though, it offered a new idea.

This is jointly supported by #WLFI and @Dolomite . In Zebec, users can directly access risk-control lending pools and high-liquidity vaults via #USD1 . Earning interest on deposits and taking out loans can be handled all within a single app, without needing to jump to a separate DeFi page. The maximum annual percentage yield is 5.2% APY—keep in mind that in this bear market, that number is fairly competitive in the stablecoin sector.

I think the real highlight of this update isn’t that it adds another yield channel, but that it brings financial services to where users already are.

The users of #Zebec focus on payment scenarios, with high daily usage frequency. These users are not the same group as pure DeFi players. They’re not as familiar with complex on-chain operations, but they hold stablecoins long-term. By embedding yield and lending features directly into the apps they already use, it effectively lowers an entire tier of the usage barrier.

But behind the scenes, this is ultimately the ecosystem strategy of $USD1 gradually coming to fruition.

In the end, stablecoins win by the depth of application scenarios. Previously, market understanding of USD1 stayed mostly at the asset level—reserve mechanisms, issuance logic, and such. Now, the pieces of the puzzle at the application layer are being filled in one by one: payments have an entry point, interest-bearing opportunities have channels, and lending has pools. The more complete the scenarios, the stronger the willingness to leave funds in place—only then can the ecosystem flywheel truly start turning.

I think it’s quite a smart timing to push the application rollout at this point. In a bull market, everyone’s attention is on price appreciation; product experience and usage scenarios don’t get much attention. The bear market is different—when the bubble deflates, users’ demand for convenience/safety/all-in-one solutions rises instead.

Let's chat about a project I've been keeping an eye on, the second round of SBT for #NeoSoul is now live and will end on July 1.

I checked on-chain, and over 24,000 have already been minted, and there are quite a few participants.

This round has three tiers: Orbit with 15,000, Vector with 3,000, and Zenith with 90. The higher the tier, the scarcer the supply; especially if you get into Zenith, it basically symbolizes your status as an early core participant.

According to the project team, this SBT serves as the foundational reputation layer of the Agent economy. My understanding is that in the NeoSoul ecosystem, the level of credential you hold corresponds to your permissions and the depth of your participation. In simpler terms, the level you currently hold is somewhat of an initial credit anchor within the NeoSoul ecosystem.

This tiered design naturally creates community stratification through scarcity; the higher you go, the tougher the entry requirements, which will definitely lead to more benefits down the line.

If you want to check if you qualify, you can take a look at their X official Twitter.

So if you, like me, are bullish on the AI Agent direction for the long term, I suggest you grab one for protection.

The stronger the AI model, the more reluctant high-value data becomes to be used.

Financial strategies, corporate records, and medical information, once inputted, could be permanently stored by centralized platforms, which has become the most tangible bottleneck for privacy AI.

Private chats are just the starting point. But when AI needs to handle high-value scenarios like Agent execution permissions, how sensitive data can be utilized during computation without direct exposure fundamentally determines how far this space can go.

And #Arcium is building a crypto execution network at a deeper level, enabling AI to complete computations and validations without exposing raw data. Essentially, it's adding a layer of execution environment, meeting the computational demands of AI applications while isolating the risk of sensitive data exposure.

Venice $VVV has streamlined the demand side, showing the market that users genuinely care about how their inputs and preferences are handled. But you need to understand that user needs and enterprise needs are two distinct lines, with the latter's data isolation requirements being on a different scale.

VVV proves that demand exists, while Arcium determines whether that demand can be amplified.

Additionally, the route has been confirmed that #Arcium will be listed on a well-known major exchange.

So this $ARX TGE might be a key moment for the market to start re-evaluating the foundational layer of privacy AI. Stay tuned.

The competition among stablecoins used to focus on liquidity depth, but when institutional capital really steps in, the reliability of the safety infrastructure becomes the true game-changer.

#WLFI The initial phase will deploy $100 million USD1 on Mova, backed by the post-quantum security layer provided by Naoris Protocol, effectively betting on this judgment.

Because $USD1 can stand up to scrutiny, fully backed by short-term U.S. Treasury bonds and cash equivalents, its reserve structure is clean and transparent, making it particularly solid among a sea of stablecoins.

Being selected by WLFI as a peer for the underlying settlement network, MovaChain is certainly no ordinary player. It integrates quantum-resistant encryption and compliance frameworks right from the protocol layer into the settlement layer, avoiding the need for patching later on. This intrinsic security awareness is precisely what institutional funds care about most.

The activity initiated by MovaChain is essentially a global consensus binding, defined as the first checkpoint towards the next generation of account security standards. I feel it allows you to participate personally in a rehearsal for post-quantum security validation as a future financial signer. In an environment where computational threats are constantly evolving, there is no turning back, which is why that line in the tweet, "compromise is failure," feels particularly concrete.

The 100% reserve of USD1 will be proven by Naoris' real-time validation network, rather than relying on periodic audit reports. This means the safety anchor of stablecoins is shifting from post-facto accountability to on-chain real-time verifiability, a design too crucial for long-term asset settlement.

The stablecoin game has an inevitable contradiction: the supply keeps growing, but the real on-chain use cases remain limited.

Most stablecoins end up just chilling on exchanges, earning interest, acting only as standby funds, rarely engaging in on-chain activities. This issue is even more pronounced in a bear market, where risk appetite for funds decreases, but sensitivity to returns increases, yet purely high-yield pools often come with unsustainable subsidy models.

@币安Binance华语 wallet's recent focus on the DeFi activity of #USD1 , in this context, feels like it's not just about luring in funds with high yields. Instead, it seems more like an attempt to shift idle stablecoins from passive holding into genuine interactions with on-chain protocols.

Looking at the specific mechanics, the design of the Lorenzo USD1 main pool shows some thought. The returns are split into two parts: a stable base annual yield from the protocol's own interest rate, and an additional boost from BANK token airdrops, which serve as ecological incentives. The benefit of this dual-layer structure is that even after the incentive period ends, the underlying yield still exists, preventing it from dropping to zero the moment activities cease, unlike pure subsidy pools.

The value of the airdrop part hinges on the future market performance and ecological positioning of the $BANK token. There's uncertainty here, but it's not entirely a vaporware concept; it has real demand backing from the protocol's operations.

From an ecological perspective, this move is more significant than a one-time event. Binance wallet needs a sufficiently attractive stablecoin asset to serve as an entry point for DeFi, while USD1 requires more on-chain circulation scenarios to prove it's not just another pegged coin.

The convergence of these dual demands led to the concentrated resource push for a pool. Allocating most of the rewards to the main pool rather than distributing them evenly indicates that the goal isn't just to cast a wide net for new users, but to let funds settle within a specific protocol, creating a usage habit.

Finally, for $USD1 itself, the spot trading area on the exchange side has covered direct conversions of mainstream assets, the contract side has also received the highest collateral rating, and the wallet side fills in the on-chain financial management scenarios, essentially capturing all three major financial flow nodes of trading, collateral, and wealth management.

The competitiveness of stablecoins ultimately hinges not just on supply and reserve transparency, but on how many real scenarios they can be used as tools, rather than just being stored.

#WLFI

Last night, Trump touched down in Beijing, bringing along a bunch of business folks, but the spotlight first hit three guys climbing into the same ride: Musk, Huang Renxun, and #WLFI co-founder Eric Trump.

Sitting up front in that lineup is pretty slick. We’ve got power, energy, and on-chain finance all packed into one car. Eric stepping into this spot is a bigger deal for WLFI than any statement could ever be.

Looking back a week gives more clarity. Trump Jr. just debunked some rumors at the Miami Consensus conference, saying the family pulling out of WLFI was just some hype driven by the bot matrix, stemming from the team page being down for a few hours. He laid it out plainly in a CoinDesk report: just because someone says something doesn’t make it true. The narrative is manufactured.

While he made solid points, Eric popping up in the front row of his dad's China visit is like backing up that debunking with actions.

What I'm really focused on is $USD1 .

A lot of people are still looking at WLFI through the lens of short-term sentiment. But the true potential of the USD1 stablecoin isn’t about the buzz; it’s whether it can be integrated, used, and settled in the real world. This is a completely different game from just trading concepts.

Currently, #USD1 has already completed its native deployment on Tempo, and the token unlock rules are transparently out there, with no inflation for two years and linear release starting in 2028—these foundational moves deserve to be called solid. Now, stacking on top of that Eric's level of public endorsement brings it a step closer to real-world usage than most stablecoin projects out there.

There’s a long road between signals and results. But at least this Beijing trip shows one thing: the intersection of politics, tech, and on-chain finance is becoming more real, and Eric with WLFI is right at that crossroads.

The hype is temporary; products are what lasts. Going forward, I’ll be keeping an eye on whether USD1 is actually getting real use.

#USD1 #WLFI

I just stumbled upon an announcement from WorldClaw, and it turns out there's more to it than meets the eye.

They launched a product called WorldRouter, in collaboration with WLFI. One account can tap into over 300 AI models, including ChatGPT, Claude, Gemini... all the big names are in there. The price is about 30% cheaper than the official rates.

If you’re often hopping between different models, this definitely saves time and cash. But I think the real focus is on the latter part of the announcement.

WorldClaw positions itself as AgentOS, which you can think of as a dedicated operating system for AI Agents. Their idea is that AI shouldn't just have conversations; it needs to get things done—remember stuff, schedule tasks, compare prices, pay bills. And all these operations require a unified payment and settlement method, which is why they used #USD1 .

This $USD1 runs on the BNB Chain and Solana. In other words, your AI Agent can autonomously make payments, receive funds, and keep accounts on these two chains, with every step being crystal clear.

Additionally, they’ve set up a points and benefits system. For every $1 spent, you earn WorldClaw Points. There are four tiers: $9.9, $99, $999, and $9999, corresponding to different amounts of AI points and rewards. If you stake $WLFI , you can unlock even more features. High-tier users will receive a dedicated hardware device and have the chance to attend private events at Mar-a-Lago and meet Donald Trump Jr.

The announcement is worded cautiously, with plenty of terms and regional restrictions. WorldClaw isn't just setting up a simple transfer station; they’re building a foundational system that enables AI Agents to work, pay, and build credibility. My understanding is that $WLFI in this ecosystem is more like a key— the more you lock, the more permissions and functions you can unlock.

The purpose of #USD1 is to turn every thought and action of the AI into a traceable on-chain transaction.