平凡的蛙里奥

Back when I did experiments in school, I was terrified that everyone’s measured results would be different. Someone would shake their hand, someone would have a crooked ruler—by the time we totaled the data, no one dared to claim theirs was the most accurate.

On-chain authorization faces the same problem. Operators go to grab external quotes or check blacklists; but due to network latency, the numbers everyone sees might differ by a tiny amount.

If we don’t resolve even this small discrepancy, the later mathematical signatures won’t line up.

The whitepaper from <c-1/> @NewtonProtocol says it built a streaming two-phase consensus. In the first phase, called Prepare, the Gateway lets everyone view their own data first, dumps what they’ve seen out, and calculates a median as the standard answer.

In the second phase, called Evaluate, everyone then uses this standard answer to run their strategies and generate signatures. This approach acknowledges that errors exist, but uses mathematical methods to filter out the noise.

I appreciate this kind of obsessive focus on engineering details. It doesn’t just preach how sacred decentralization is. Instead, it honestly admits that in distributed environments, data can be biased—and then provides a working patch.

That pragmatism is a prerequisite for whether infrastructure can truly be used.

But we should consider both sides. In this mechanism, the Gateway’s role is quite critical. Even though it can’t fabricate signatures, it controls the aggregation and distribution of data. If the Gateway ever goes on strike—or intentionally withholds certain data—authorization could still get stuck. This is my inference based on its centralized coordination logic; I’m not saying anything has happened yet. The official stance is that this position will be rotated in the future, but whether the rotation mechanism runs smoothly will depend on how the mainnet Beta performs.

So how should we view it? Newton’s two-phase consensus aims to put a filter on distributed authorization, preventing data deviation from turning into systemic lag. But how stable that filter is depends on how it performs under high concurrency.

Don’t just listen to claims about sub-second response times—focus more on whether, during extreme market conditions and sharp data fluctuations, the authorization credentials take longer to come out.
$NEWT #Newt

This week we were originally focused on following up other projects, but GRVT’s second-quarter data is so striking that it’s hard not to take a closer look.

Looking at the core metrics, Season 2’s TVL surged from $11.3 million to $107.1 million. Open interest also skyrocketed from $11.6 million to $48.41 billion—an astonishing 42x.

Add to that a cumulative bilateral trading volume of $393 billion, and monthly active users surpassing 10,000 as early as January. When these figures are combined, they confirm that the platform is seeing real capital and trading activity flowing in continuously—not a superficial boom propped up only by points or incentives.

On token issuance, the TGE timeline has been pushed back from the originally planned Q1 2026 to after June 30, following the end of Season 2. Meanwhile, the community airdrop allocation has been increased from 22% to 28%, and Season 2’s share has been raised from 12% to 18%.

Raising the airdrop allocation right before the TGE may look like standard user-favorable adjustments on the surface, but in practice it resembles a strategic move—aimed at driving momentum and boosting activity in the final stretch.

The product side is also steadily filling in gaps: the L1 yield feature integrated with Aave is underway, and spot trading went live by the end of April.

As for trading markets, what started with perpetual futures has gradually expanded into spot and options. The overall pace suggests the project is building the complete infrastructure an exchange should have, rather than chasing a single point breakthrough.

From the standpoint of capital and compliance, the project completed a $19 million Series A this year, led by ZKsync and Further Ventures. Together with the prior $14.3 million raise, the total funding scale is substantial. In addition, it secured a Lithuania VASP license as early as 2023—this compliance groundwork is also a key dividing line between it and many purely on-chain projects.

Overall, the data-driven growth in this cycle looks genuine. But adjustments to allocations before token issuance still deserve close attention.

Instead of getting hung up on the airdrop ratio itself, it may be better to focus on how liquidity performs in reality after the token is issued @grvt_io #grvt

While reviewing资料 I found myself halfway through, and then suddenly remembered that I had promised to write about user registration last time, so I jumped straight into it.
The user system built around @grvt_io is, in my view, the most interesting part: each user can bind only one email and one wallet. It sounds simple, but it forces you to hold both Web2 and Web3 credentials at the same time. If you register via email, the system automatically generates a wallet in the background through Privy; that wallet is the one truly used to sign transactions. Users don’t perceive the private-key layer at all.
I think this design caters to people who don’t understand how wallets work, but the trade-off is that actual asset control is handed to Privy as the middle service, which makes it a bit convoluted compared to the idea of “pure self-custody.”
Registering via a wallet is completely different. Any operation involving changes in asset ownership must be signed by the already-registered wallet, with no middle layer acting as a fallback. This route is the real meaning of self-custody.
Personally I lean more toward this option, but the operational barrier is definitely much higher than the email route.
The on-chain main account ID is bound to the wallet address used to create the account. I think this is crucial: it means identity and asset control can be verified on-chain, not something decided solely by a backend table. That transparency differs from many platforms’ internal ledger model.

Going further into the API keys section, permissions can only be delegated at the transaction-account level, and they must be tied to an Ethereum address in order to take effect. In secure mode, the private key is kept by you yourself; in convenience mode, the browser generates it for you. GRVT clearly states that after the private key is generated, it is not stored—this stance is fine.

But in convenience mode, the risk is still essentially pushed onto users to manage their own browser environment. I think most people won’t realize this pitfall.

Overall, this credential system seems designed for institutions and professional users.

The two registration paths each have their own trade-offs—none of them is a free lunch. Don’t just think about saving time and choose email registration. Spend a bit more time to understand the wallet signing route. #grvt

In school, when you do experiments, the worst thing is that the results measured by each person are different. Someone’s hands shake, someone’s ruler is crooked—then when it comes time to compile the data, nobody dares to claim their numbers are the most accurate.

On-chain authorization also faces this kind of trouble. Operators go to fetch external quotes or check blacklists, but because of network latency, the numbers everyone sees might differ slightly.

If these small discrepancies aren’t resolved, the later math signatures won’t line up and can’t be combined.

I flipped through the @NewtonProtocol whitepaper and saw it has a whole scheme based on median consensus. It makes every operator run a plugin inside their own WASM sandbox to capture data. After they collect it, they don’t rush to make a conclusion. Instead, they just throw all the numbers out.

The Gateway computes a median and treats it as the standard answer, then sends it back so everyone can sign using the same result. This approach acknowledges that errors exist, but uses mathematics to filter out the noise.

I appreciate this kind of bowing to engineering reality. It doesn’t chase some nonexistent absolute synchronization. Instead, it honestly admits that the network has latency, then provides a patch that actually works end to end. This pragmatism is a prerequisite for infrastructure to become truly usable.

But the story cuts both ways.

In this mechanism, the Gateway’s role is crucial. Although it can’t fabricate signatures, it controls the aggregation and distribution of the data. If the Gateway ever goes on strike, or intentionally withholds certain data, authorization would still get stuck.

This is my inference from that centralized coordination logic—not to say anything has gone wrong already. The official line is that this position will rotate in the future, but whether the rotation process is smooth still depends on how well the mainnet Beta runs.

So how should we look at it? Newton’s median consensus is trying to put a “filter” on on-chain authorization, so that data deviation doesn’t turn into systemic sluggishness.

But how stable that filter is depends on how it performs under high concurrency. Don’t just listen to it brag about data consistency—focus on whether, during extreme market conditions and when data fluctuates violently, the speed at which authorization credentials are issued slows down.
$NEWT #Newt

7.11 U.S. Stocks｜Nasdaq four straight green days didn’t turn into a bigger rally—big banks’ earnings fail to lead, and U.S.-listed Chinese concept stocks throw a party
Brothers,

Last night (7/10), the U.S. stock market looked a bit calmer than the night before, but the internal gains were still clearly visible.
The Nasdaq ended slightly lower to pause a three-day winning streak.

The S&P 500 basically went sideways and closed with only a small move.
After financial stocks stabilized, the Dow ticked up and rebounded a bit, but it may not have fully recovered the drop from the previous trading day. cnbc+1

Last night’s tape still had a few signals worth paying attention to.

First, tech sentiment switched. The night before, Broadcom (AVGO) surged nearly 4% on extending its deal with Apple, lifting half the chip stocks. cnbc+1
After that, the chip sector shifted into a sideways range and consolidated. Money started switching from “chasing” to “stock picking.” But the main logic around AI computing power remains unchanged. cnbc+1

Second, the “heat” in Chinese concept stocks hasn’t faded. The night before, Alibaba, Baidu, and other Chinese concept names collectively jumped hard after the central bank signaled support for Hong Kong’s capital market. That enthusiasm cooled off somewhat last night.
However, fund flows are still relatively warm—southbound flows and foreign investors’ ongoing interest in discounted China assets persist.

Third, the earnings season is officially underway. Next Tuesday (7/14), JPMorgan will fire the first shot for Q2 earnings. The market expects overall S&P 500 earnings to be up about 24%, with even greater hopes for the tech sector—around 65%.

Put simply, it’s “wanting both things.”

Want tech stocks to keep high growth and support valuations, and also want banks to argue that the economy isn’t as fragile.

In short, geopolitical disturbances haven’t been fully digested yet. This real “big exam” of earnings season has already queued up.
Whether the Nasdaq’s streak of consecutive gains can continue depends on that.

The key is whether bank earnings next week can deliver a strong opening—
a great “door-opening” report.

Two things to watch closely:
1) Whether U.S. Treasury yields keep climbing and weigh on valuations.
2) Before the Fed’s 7/29 FOMC meeting, whether rate-cut/hike expectations suddenly change.
#美股超话

This week, I went through the account system documentation for @grvt_io from start to finish. I only meant to skim the architecture diagram.

But I couldn’t get past the layered design between the fund account and the trading account. This design is completely different from the way I usually encounter CEX systems.

It splits identity into two layers: the fund account is the highest-level on-chain identity marker, responsible specifically for deposits, withdrawals, and transfers. The trading account is attached under the fund account and is used specifically for derivatives trading. Once you separate them like this, permissions are separated as well—fund management permissions override all other permissions.

Even operations like adding or withdrawing an address require multi-signature thresholds before they can be triggered. I think this design is aimed at institutional compliance, not convenience for retail users.

The off-chain matching and on-chain verification I mentioned earlier—this time while reading the documentation I noticed more details. The results from both the matching engine and the risk engine are published to the blockchain as zero-knowledge proofs.

For settlement, it’s verified on-chain according to smart contract logic, not decided off-chain by people “back-of-the-envelope” guessing. Compared with many projects that claim to be decentralized but whose settlement actually relies on off-chain backends, this is definitely not on the same level.

As for the native token bridge—I actually think it’s more convincing than any marketing slogan.

On-chain data can’t lie, but it can only show that fund flows are transparent—it doesn’t necessarily mean the system runs smoothly. I haven’t tested it in a real trading environment, including latency, so I won’t draw conclusions on that.

The API key design is quite detailed: permissions can only be delegated down to the trading account level. After generating the private key, GRVT doesn’t store it—this attitude I do approve of.

However, in “convenience mode,” the step where the browser generates the private key makes me think the risk of ordinary users making mistakes is underestimated.

Overall, this architecture is more complex than I expected—institution-friendly, but not beginner-friendly. Don’t overthink the airdrop numbers; spend more time looking at the account permissions diagram #grvt

In the past, when I went to the train station in my hometown to pick up tickets, I was most afraid of getting the message: “This station supports only tickets from this rail bureau.”

Even though it’s all the same railway network, if you switch provinces, you suddenly become an outsider—you have to join a new queue and prove your identity all over again.
In today’s on-chain world, every day feels like they’re putting on a show.

You just finished a round of compliance review on Ethereum, and then when you switch to Base or Arbitrum, you have to run that same process again. The whitepaper calls this “compliance fragmentation,” and I really do find it quite torturous.

@NewtonProtocol In those chapters about the cross-chain architecture, I found that the thing it most wants to eliminate is exactly this awkward situation where you’re not recognized just because you’re in a different place. It builds a mapping model between a source chain and a target chain—pretty clever logic.
Operators only need to register once on the Ethereum mainnet, stake the ETH, and that security endorsement becomes a network-wide shared foundation. No matter which chain your transaction is sent from, Newton’s operator network can fetch your rules from the mainnet headquarters and sign you a credential remotely. It feels like you got a network-wide access pass at HQ—go to any branch and you can just have your face recognized and walk right in.
I appreciate this approach of not forcing developers to reinvent the wheel.

It doesn’t force you to move assets around, and it doesn’t make you wrestle with a separate compliance solution on every chain. It’s like a cross-chain authorization transfer station—one set of rules that the whole network recognizes. For institutions running business across multiple chains, the savings are not only in development cost, but also in avoiding that headache-inducing fragmentation problem.

This “one authorization, reusable across many places” pattern also concentrates risk. Newton’s operator set is synchronized to the target chain by relayers on a regular basis, and what gets carried over is the signed Merkle root. If the synchronization step is delayed, or if the operator status on the mainnet changes but the target chain hasn’t updated yet, there could be a security gap in the middle. This is my inference based on its asynchronous synchronization mechanism—I’m not saying anything has happened right now.

And all the crucial lifelines are concentrated in that mainnet operator table. If the mainnet gets clogged, the chains that rely on it might get sick along with it.

So regardless of how you look at Newton’s architecture, it seems to be aiming to build a unified authorization center for on-chain finance—with a fairly forward-looking approach to cross-chain connectivity.

But its ceiling isn’t about how many chains it supports; it’s about whether cross-chain synchronization can be fast enough and stable enough.
$NEWT #Newt

7.10 US Stocks | Indices Are Moving Each in Their Own World Nasdaq Rallies for Three Straight Sessions Against the Tide, Earnings Season Begins
Brothers

Last night (7/9), US stocks were extremely conflicted. Different indices each played their own game.
Nasdaq rose 0.28% to 25,870, winning its third straight day.
The S&P slipped 0.3%.
But the Dow fell more than 1%, dragged down by financials and energy, to 52,348.

The divergence among indices shows the market is still locked in a tug-of-war between geopolitics and earnings.
Last night’s trading had three hard-hitting highlights:
First, Broadcom renews big with Apple. Broadcom (AVGO) announced it has renewed its deal with Apple through 2031. The moment the news broke, Broadcom jumped nearly 4%, which helped steady the PHLX Semiconductor Index as well. Plus, Meta’s ongoing “selling compute power” story is still simmering.

Second, Chinese concept stocks collectively go “legendary.” Last night, the Nasdaq China Golden Dragon Index surged 2.05%. Alibaba jumped 11%, and Baidu rose nearly 5%.
The statement from the People’s Bank of China supporting Hong Kong’s capital markets became the strongest tailwind. This wave saw both southbound funds and foreign capital diving into bargain Chinese assets.

Third, the big boss of earnings season is coming. JPMorgan is set to officially kick off Q2 earnings next Tuesday (7/14). The market expects S&P 500 profit growth to reach 25%, and tech stocks to climb toward 65%.

With expectations at this “ceiling-level,” will it be a surprise or a shock? We’ll know soon.
In short: the shadow of geopolitics hasn’t fully lifted, but funds have already started positioning ahead of Q2 earnings. Tech stocks may have rallied high, but a hard-logic play like Broadcom-Apple’s renewal suggests AI demand hasn’t stopped.

Next week’s bank earnings reports will be a sentiment barometer. Watch U.S. Treasury yields—so long as they don’t go crazy higher again, the Nasdaq can keep this breath going.
#美股超话

In company permission management, people often use rules like “if it’s the finance department and the amount is under ten thousand, approve it,” but once that’s moved on-chain, things fall apart.
On-chain, you don’t just need to look at identity; you also have to verify your signature, review your delegation proof, and even check your asset credentials on another chain through cross-chain queries.

These hardcore cryptography tasks are simply beyond what traditional authorization languages can handle.
When I was reading @NewtonProtocol ’s white paper, I found that it had made some changes to the Rego language and added a bunch of cryptographic extensions.

It didn’t try to reinvent a hard-to-learn language; instead, it gave existing Rego a plug-in. For example, it added functions that can directly recover an ECDSA signature address, as well as logic for verifying multisignature approvals. That means when developers write policies, they can directly say, “If this transaction has signatures from three approvers and they’re all still valid, then allow it.”

I think this kind of approach, which welds compliance logic together with cryptographic evidence, is what really shows expertise.
I appreciate this kind of practical, no-showboating style.
It knows developers are already used to Rego, an industry-grade standard, so it simply follows that habit and packages up the hardest part—the cryptographic verification.

You don’t need to be a cryptography expert; as long as you can write simple logical conditions, you can define a bank-grade authorization policy on Newton.

But to be fair, let’s bring it back down to earth. Rego with cryptographic extensions is more powerful, but it also increases complexity. The deeper the policy, the greater the computational pressure on the operator node. The white paper mentions that fees are calculated based on WASM instruction count and bandwidth, which means those complex cross-chain identity verification policies could be quite expensive to run once. If costs can’t be brought down, this kind of advanced authorization may end up being reserved only for large institutions, and ordinary small projects won’t be able to afford it.

That’s my inference based on its pricing model, not a claim that it’s already outrageously expensive now.
So how should we look at this? By equipping the policy engine with cryptographic plug-ins, Newton is trying to give compliance rules real teeth, so they can actually bite into hard on-chain evidence. That’s a very deep move. But whether it can become widely adopted depends on whether it can find the right balance between performance and cost.
Pay less attention to how many complex encryption functions it supports, and more attention to how developers write complex policies.
$NEWT #Newt

7.9 US Stocks｜Fighting Flare-Up Returns Trump Announces Ceasefire Ends Dow Plunges Nearly 600 Points
Brothers

We originally thought last week’s nonfarm payrolls surprise would give US stocks a breather. But last night (7/8) was a bolt from the blue—geopolitical tensions directly detonated the market.

All three major indexes jumped into a free fall. The Dow plunged 577 points to 52,925. The S&P fell 0.28%, while the Nasdaq managed to hold on at the close with big-tech support, barely flipping green up 0.2.

The fuse for last night’s selloff was Trump’s sensational salvo. He publicly announced that the ceasefire agreement with Iran has officially ended, saying last night’s strikes were intense.

Tonight could be more of the same. Once the news hit, WTI crude jumped instantly by more than 4%, and Brent crude shot above $78. The yield on 10-year US Treasuries surged back to the high 4.56% zone. A one-two punch to both oil and rates—risk appetite was shattered in an instant.

Sector divergence was brutal. The Dow blue-chip leaders that had been setting new highs turned into the worst-hit areas. American Express and Boeing led the declines. Money was fleeing aggressively. The Nasdaq was down more than 1% at one point.

Nvidia alone surged against the trend, rallying 3.6% to hold the line. What’s interesting is that Nvidia’s recent decline has been steep. Since the start of this year, it’s only gained 6%, far behind SanDisk and Micron. Yet amid a big split in chip stocks, Nvidia has become a safe harbor for defensive capital.

In addition, the freshly released Fed minutes delivered another blow. The minutes showed that if inflation continues to run hot, officials generally agree that they would need to raise rates. With a rate-hike shadow paired with surging oil prices, the loose-policy dream for the second half of the year was smashed to pieces.

Simply put: a geopolitical oil-price shock, paired with hawkish Fed minutes, has pushed the market fully into a wartime mode. In an environment where oil prices are surging and yields are rebounding, high-valuation themes lacking earnings support will feel extremely uncomfortable.

Next Tuesday, the Q2 earnings season kicks off. If earnings can’t prove they deserve this valuation, AI hardware may also face deeper technical adjustments. #美股超话

The year I bought a house, I did a loan due diligence once. The bank hired lawyers, appraisers, auditors—everybody—and they spent more than half a month checking. They were looking to see whether I had any issues in my past.
But after the contract was signed, real trouble happened. That kind of pre-review can’t help once things go off the rails, because it only covers what’s checked before signing.
I looked through the @NewtonProtocol materials and saw it talk about enforcing things at runtime. That’s exactly the gap I thought of. On-chain risk prevention mainly relies on pre-trade audits—scrutinizing the contract code over and over.
But the whitepaper points out a painfully real fact: most DeFi hacks don’t happen because the team forgot to add checks. Instead, it’s because certain assumptions quietly fail at some moment during execution. Audits test intent; attackers exploit the edge cases during execution.

Newton’s approach is to take a rule from an audit report—a single sentence—and turn it into a hard rule that truly blocks at the moment of execution.

It guards runtime invariants, such as the value boundaries of the liquidity pool, the collateral ratio, and the oracle deviation tolerance. These bottom lines are verified one by one before a transaction is executed. If a rule is violated, execution is halted immediately—not after an incident happens, with a monitoring system raising alarms.

I agree with this shift. It acknowledges a reality: vulnerabilities are impossible to defend perfectly. Rather than expecting audits to plug every hole, it’s better to ensure that even if a hole is exploited, the worst-case outcome still has a ceiling imposed by safeguards.

But having said that…
Where exactly to draw the runtime red line, and how tight or loose it should be—ultimately comes down to human judgment. If that deviation tolerance is set too wide, attackers can still make moves within the red line without triggering the blocks (this is my inference based on the threshold mechanism; I’m not saying it has already been bypassed in practice).

So how should we look at it?
Newton moves security from verifying intent before execution to enforcing behavior at runtime—that idea is well targeted. But whether it’s actually effective doesn’t depend on how many invariants it lists; it depends on whether, in a real deployment, those thresholds are calibrated to match the scenario. If they’re set too loose, it’s basically like they were never set.
$NEWT #Newt

7.8 U.S. stocks｜Is the AI rebound another “fake fall”? Nasdaq falls another 1.2%

Last night (7/7), U.S. stocks changed course again. The Nasdaq had just risen 1.3% the day before, making everyone think AI was about to turn things around—then, all within a single trading day, it was dragged back to square one.

The Nasdaq fell another 1.2% to 25,818. The S&P 500 dropped 0.4%, and the Dow fell 0.25%​ . Internally, most stocks were actually up, but when AI chip stocks collapsed, they pulled the Nasdaq down with them.

The trigger came from Asia. Samsung Electronics plunged 6.9% during Seoul trading. Even though its Q2 guidance was actually strong, the market still reacted this way. That’s the market for you—AI stocks have surged globally so much that any small gust of news makes some people bolt for the exits.

Mizuho put it plainly: yesterday’s chip rebound was just another “head fake.”

AI deleveraging began last week.

What made matters worse were geopolitics and oil prices. The U.S. has resumed sanctions on Iranian oil, and tensions in the Middle East are flaring up again. WTI crude jumped to above $72, and the yield on the 10-year U.S. Treasury also surged to 4.55%. With both oil and interest rates rising, two heavy weights fall on overvalued tech stocks.

One thing worth noting: Nvidia rose by about 1% against the trend. What’s interesting is that it has only gained 6% this year—far behind “memory妖” stocks like SanDisk (+581%) and Micron (+228%).

But precisely because it didn’t go completely berserk, on a night when chips collectively weakened, it became a “safe haven.”

In short: this AI rally has entered a high-volatility phase where the situation changes “day by day,” with bulls and bears repeatedly strangling each other at elevated levels.

Don’t treat a one-day rebound as a signal of a reversal. The real timing marker will be this week’s Q2 earnings season. Market expectations are for S&P earnings growth of 24%, and tech’s even more explosive growth of 65%.

Delta and Pepsi will kick things off soon. Whether earnings can hold up against expectations at this ceiling level will determine whether the “momentum” in AI can continue—or not. #美股超话

People who’ve had their wallet stolen once understand that kind of helplessness. Once your private key is leaked, the other side holding the key is the legal owner. You watch your assets get transferred away right in front of you, with not even a chance to stop it.

As for passwords: trying to manage everything with just a single key is simply too fragile.

@NewtonProtocol In the whitepaper, I saw the section about anti-fraud and anti-hacking, and I thought its approach really hit the mark. It adds a second layer of authorization to non-custodial wallets. In plain terms, it’s like a blockchain version of two-factor verification: for transactions over a certain threshold, having only the private key isn’t enough—you need to pass another gate such as device binding or a session key as a second checkpoint.
This means that even if the private key is leaked on its own, the attacker still can’t move your money.

It also has a “dirty money blocking” angle. The strategy can cross-check incoming funds against a list of marked addresses—addresses that contain stolen assets or suspicious sources—and block them at the door, so you don’t unknowingly receive a hot potato and end up in trouble.

I agree with this line of thinking, because it doesn’t stubbornly try to fight down the dead-end of “preventing private keys from being stolen.” That thing can’t really be stopped. Instead, it changes the perspective: even if the key is lost, losing the key doesn’t automatically mean losing the money.

But to be fair, that second layer of protection is also a brand-new dependency. What if the second device is lost, or the session key goes wrong—could it end up locking you out the other way? How do you strike that balance? The whitepaper doesn’t spell it out (this is just my inference following that multi-factor logic; I’m not saying there’s already been an incident of lockout).

So how should we look at it?
Newton shifts wallet security from “guarding a single key” to “guarding a behavior threshold.” That’s pretty practical. But whether it truly works doesn’t depend on how many safeguards it lists—it depends on how many people are actually willing to hand over high-value wallets to this second-authorization setup.
Don’t just look at the protection checklist it provides—focus on real-world usage.

Pay more attention to how many wallets people actually use in practice
$NEWT #Newt

No trader can be arrogant; you must clearly recognize that the money you make is just the market’s gift, not a measure of how strong your personal ability is.

In a hot, bullish market, all strategies will seem to work. In a weak, bearish market, every trade is wrong. If you profit, don’t think you’re so capable—it’s the market’s gift.

Trust your own judgment. When prices rise, there are “ghost stories” for the rise; when they fall, there are reasons for the decline!

7.7 US Stocks｜Market opens to good news! Nasdaq jumps 1.3% and recovers losses—are AI hardware stocks up again?​

Brothers

At the start of the first full trading week of the second half (7/6), US stocks finally gave the bulls a face-saving rally. All three major indexes finished higher together: the Nasdaq rose 1.35% to 26,182, wiping out the losses from the previous two days; the S&P 500 gained 0.71%; the Dow set a new intraday high, but slipped slightly by 0.05% at the close to 52,874

The toughest logic on the overnight tape was still the “last-ditch comeback” in AI hardware stocks​

Last week, we were still worried that storage price hikes would crush downstream mega-tech. But overnight, the wind direction changed

JPMorgan stepped in and shouted that AI infrastructure demand has three layers of positive drivers. As a result, the SOXX index surged more than 3% directly. Dell (DELL) became the standout stock of the whole session, jumping over 8%!

​Trump personally went on stage for Dell yesterday, calling out and affirming Dell’s CEO. With political tailwinds plus Dell’s real strength in AI servers, capital rushed back into the AI infrastructure track.

Meta also kept the momentum, finishing up nearly 9%！​ Zuckerberg’s “Meta Compute” story of selling computing power convinced the market. People believe Meta can shift from just “burning money to do AI” to “charging rent with AI.”

But Tesla and Microsoft turned into “the unfortunate ones”​

Tesla (TSLA):​ Last night it released its Q2 delivery data: 480,000 vehicles! Up 25% year over year, far above Wall Street expectations of 406,000

So what happened? The stock crashed by about 7.5%, falling below $400. A classic case of “good news already priced in.” The market is worried that those 480,000 deliveries were achieved via price cuts—“killing one thousand with losses of eight hundred.” Gross margin is the next big test

Microsoft (MSFT):​ Last night it announced layoffs of 4,800 people, with the Xbox segment hit hardest. One of the worst-performing tech giants this year, and it fell another 1.2% overnight.

In short: last night, “selling compute power” won, while “selling endpoints” lost. The AI infrastructure logic is still holding strong, but the pressure on consumer electronics and high-valuation leaders hasn’t gone away yet

Starting next week, second-quarter earnings reports will start coming in thick and fast. Whether it’s real results or just PPT talk—we’ll see then. Keep an eye on pullback opportunities in AI hardware, but don’t go catching knives like Tesla that’s “good news already priced in” #美股超话