Crypto KOL Community & Market Trends

If you understand “on-chain assets” as a list, then gold, government bonds, and real estate are of course all on it But what I really want to talk about is another category: things many people can’t see, yet that have been making money for you, explaining things for you, and vouching for you all along—the fan base, personal reputation, industry influence, and the kind of trust built up year after year

ETH at $2,450—are you chasing?
Look at the surface: the medium-term structure has turned bullish, but the short term is grinding inside a range.
In mid-to-late August, ETH broke through the descending trendline on heavy volume from $1,900 and moved above Fib 0.618 ($2,438), setting a new higher high for this cycle. Spot ETFs have maintained net inflows, totaling approximately $1.85 billion in August, and institutional allocation shows no sign of stopping.
But then what happened? It stalled at $2,546 and pulled back to oscillate around $2,450. Macro rate-hike expectations are weighing on it, keeping it stuck between up and down.
First: this breakout is different from all the previous fake breakouts.
The descending trendline from the 2025 high was decisively broken by a high-volume bullish candle in August. The 50-day moving average is at $2,089 and the 200-day moving average at $2,036, with price far above the long-term averages—this is not a short-term rebound, but a signal of a medium-term trend reversal.
The staking rate is 30%-34%, with over 2 million ETH queued to enter and a wait of approximately 36 days, while exits are nearly zero.
Staking ETFs such as BlackRock's ETHB have launched and distribute yield, addressing the spot ETFs' lack of interest income.
Second: macro conditions are capping the upside, but you need to recognize one fact.
CPI is due on September 11, and the FOMC meets on September 16. Rate-hike odds have been pushed to 60%, with the market shifting from “possibly pausing” back to “higher for longer.” This is the entire reason ETH became timid after reaching $2,546.
August nonfarm payrolls came in at 162k, above expectations, sending rate-hike odds higher—ETH bounced from $2,370 back to $2,450 instead of falling. The same negative catalyst three months ago would have sent ETH below $2,000. The fact that it is no longer falling on bad news shows that macro data can only affect the pace, not change the trend.
Third: the technical picture has reached the point where a direction must be chosen.
On the daily chart, a symmetrical triangle/wedge is compressing, with a descending upper boundary and ascending lower boundary, while volatility contracts. This is a classic pre-breakout pattern.
Daily RSI is at 61-63, not overbought. MACD is above the zero line, but the histogram is shrinking—momentum is slowing, but this is not a reversal.
Key levels for bulls and bears:
Upside: $2,540-$2,550 (range top + supply zone); a breakout targets $2,650, followed by $2,920.
Downside: $2,438-$2,440 (Fib 0.618, the weekly bullish baseline); losing it targets $2,380-$2,400, followed by $2,300.
Trading strategy
Bullish approach:
Lightly enter longs on a pullback to $2,380-$2,420, with a stop below $2,350 and a target of $2,540-$2,550.
Breakout long:
Go long if the daily candle holds above $2,550 on heavy volume, with a stop if price falls back into the range, targeting $2,650 and then $2,920.
Bearish/hedging:
Only take a light short if $2,540-$2,550 is clearly rejected amid negative macro news, with a stop at $2,565-$2,580 and targets of $2,438-$2,400.
What does ETH look like now?
Like BTC in October 2023—the structure has already turned bullish, and everyone is waiting for the final negative catalyst to land before flooring the gas.
On the day $2,550 breaks, you will see price head straight toward $2,650-$2,920. While you are still wondering whether to chase, the whales will already be counting their money.
Retail traders are debating the direction at $2,450, while institutions have placed 500k buy orders below $2,400.
What is your ETH cost basis?
At $2,450, are you adding or reducing your position? #美国8月非农超预期 #Gate用户突破6000万 #BTC收复8万美元 $BTC $ETH $SOL

UNI at $6.3—are you chasing it or not?
First, the surface view: it has doubled in a month, and everyone is shouting, “UNI has gone crazy.”
It was still languishing at 3.2-3.6 in mid-August, climbed above 5 at the end of August, and surged to 6.5 in early September. Today it is ranging between 6.11-6.35, with little 24-hour volatility, but it is still up 90%+ over the past month.
RSI is at 78-80, already overbought. The trend is bullish, but the slope is too steep, and 6.5 has held three times.
First: UNI is no longer just a governance token—it is “burning money.”
Previously, UNI only provided voting rights, and protocol revenue had nothing to do with you. Now, protocol fees flow into TokenJar, which buys and burns tokens through Firepit, burning around $300,000 per day.
100 million tokens were burned in a single event, about 10% of the supply
Continuous daily burns are reducing the supply
Usage is directly linked to price
The crypto world talks about a “deflationary narrative,” but UNI is genuinely deflationary.
Standard Chartered set a $100 target for 2030 and expects $6.5 by the end of 2026—UNI is already at 6.3, so this forecast has effectively been realized ahead of schedule.
Second: Robinhood Chain went live, and UNI captured the biggest share of the upside.
Robinhood Chain launched in July, and Uniswap deployed on it the same day, sending tokenized stock trading volume surging.
Previously, Uniswap only handled crypto tokens; now it is starting to compete for the gateway to bringing traditional assets on-chain. RWA trading volume hit a new high, with tokenized stocks trading on Uniswap—this is an incremental market.
On top of that, the v4 fee switch was activated, and daily burns jumped from over $100,000 to $300,000. Robinhood Chain contributed a considerable share of the revenue.
Third: The technical picture looks excellent, but there is one hidden risk.
The daily chart shows a bullish alignment, well above the 50-day and 200-day moving averages, and the trend remains intact. But RSI is at 78-80, a textbook overbought level.
6.5 has failed three times, 4H momentum is beginning to stall, and volume has retreated from its September 3 peak. This is a combination of “the end of the main upward wave + overbought conditions.”
Bulls versus bears—you decide
On one side:
Protocol fees are being redirected to buybacks and burns, with $300,000 burned daily and the supply continuously shrinking
AAA institutional rating + Standard Chartered’s bullish outlook, adding institutional narrative support
Robinhood Chain + RWA bringing incremental revenue
Up 100% in a month, with the bullish trend unchanged
On the other side:
RSI 78-80 is overbought, and 6.5 has failed three times
Nonfarm payrolls exceeded expectations, cooling rate-cut expectations
Funds are more cautious ahead of the FOMC
After more than doubling, profit-taking pressure is substantial
Resistance above: 6.5-6.66 → 6.9-7.2 → 7.5
Support below: 6.0-6.11 → 5.8-5.82 → 5.4-5.5
Trading strategy
Short-term traders:
Wait for a pullback to 6.00-6.11 to go long, with a stop-loss below 5.78 and a target of the previous high at 6.50. If the rally loses momentum at 6.48-6.55, consider a small short position, with a stop-loss above 6.68 and targets at 6.10/5.82.
Swing traders:
The reasonable accumulation zone is 5.80-6.05; enter if it holds on a low-volume pullback, with a stop-loss at 5.40. If the daily close falls below 5.40 and BTC breaks below $76,000, exit first and wait until after the FOMC.
This time, UNI is different—it is transforming from a “governance token” into a “value token.”
99% of people are still looking at UNI through the old lens, while institutions are already giving it an AAA rating.
The biggest opportunity in blockchain is not chasing pumps and selling dumps, but recognizing when an asset is transforming from “having no value” to “having value.”
What is your UNI cost basis?
At 6.3, are you chasing it or waiting?#美国8月非农超预期 #Gate用户突破6000万 #BTC收复8万美元 $BTC $ETH $UNI

BTC at $79,600—are you panicking?
Look at the surface first: major macro headwinds, yet the price hasn't collapsed.
It just surged above $82,000 on September 3. On September 4, the nonfarm payrolls came in at 162,000 versus expectations of 55,000, exceeding them by 3 times. The market was stunned: the probability of a rate hike jumped from 50% to 60%. BTC fell in response and is currently hovering around $79,600. From $82,000 to $79,600, it has fallen just 3%.
In the past, this kind of data could trigger a 10% plunge.
First: Nonfarm payrolls beat expectations by 3 times? You think it's bearish, but Wall Street sees it as bullish.
On September 4, nonfarm employment came in at 162,000, while the market expected only 55,000. The dollar rose, Treasury yields surged to 4.8%, and the probability of a rate hike climbed to 60%—on the surface, all bearish.
But think about it from another angle: An economy this strong means there's no recession, corporate earnings are healthy, and risk assets benefit over the long term.
On the same day, U.S. spot BTC ETFs saw net inflows of $731 million. BlackRock alone accounted for $454 million, the largest single-day inflow of the year. Cumulative inflows over the past three weeks reached $3.8 billion, pushing total assets above $100 billion.
Second: $82,000 has failed three times, but $79,000 has also held three times.
Since rebounding from $62,000 at the end of August, BTC climbed all the way to $82,000. On September 3, it broke out with a long bullish candle; on September 4, it surged to $82,000 before closing lower, with a clearly visible upper wick.
But look below: The $79,000 level has been tested three times, and held every time. When it should fall but doesn't, that's strength.
Third: The next two weeks will be the most critical period of the year.
CPI on September 11 and the FOMC meeting on September 15–16—two bombshell data events.
Soft CPI + FOMC stays on hold → BTC surges straight to $85,000+
Hot CPI + a confirmed rate hike → It could retest $76,000 or even $72,000
But the rate-hike expectation has already been priced in at 60%. If a hike actually happens, that's “bad news priced in”; if there is no hike, that's an “unexpected positive surprise.”
The bulls and bears are facing off—you decide
On one side:
ETF single-day inflows of $731 million, the largest of the year
Cumulative inflows of $3.8 billion over the past three weeks—Wall Street's real money
$79,000 has held three times, confirming support
Companies such as Strategy are buying again for their treasuries
El Salvador continues to accumulate
On the other side:
Nonfarm payrolls beat expectations by 3 times, with the rate-hike probability rising back to 60%
Treasury yields at 4.8%, with a strong dollar weighing on risk assets
$82,000 has been rejected three times, showing weak bullish momentum
Geopolitics and oil prices are driving up inflation expectations
Resistance above: 80,800–81,500 → 82,000–83,600
Support below: 78,800–79,200 → 76,500–77,000 → 72,000–74,000
Trading strategy
Short-term traders:
Use light positions to sell high and buy low within the 79,200–79,800 range. Bullish: Buy on a pullback to 78,800–79,200 after it holds, targeting 80,800–81,500, with a stop-loss below 78,000. Bearish: Lightly short a rebound into resistance at 80,500–81,000, targeting 79,000, with a stop-loss above 81,800.
Medium-term traders:
Wait for the CPI and FOMC outcomes. Positive data → Build long positions on a pullback to 76,800–78,000, targeting 83,000–86,000. Negative data → Wait for 76,000 or even 72,000–74,000 before taking action.
A daily close below 76,000 on rising volume → Medium-term weakness confirmed; get out first
The next two weeks will determine the direction for the second half of the year—
Nonfarm payrolls beat expectations by 3 times, while ETF single-day inflows reached $731 million—which do you believe?
Retail investors trust panic; institutions trust data.
Macro is noise; liquidity determines direction. The rate-hike expectation is already fully priced in at 60%—what are you still afraid of?
At $79,600, are you panic-selling at a loss or waiting to add to your position?#美国8月非农超预期 #Gate用户突破6000万 #BTC收复8万美元 $BTC $ETH $SOL

$STRK Showing Strong Short-Term Momentum!
$STRK is making solid moves today, trading at $0.02873, up +10.62% in the last 24 hours!
After bouncing from a 24h low of $0.02547, it hit a peak at $0.03004 before entering a minor consolidation phase on the 15-minute chart.
With over 105M STRK traded in 24h volume, buyer interest is clearly picking up.
24H High: $0.03004
24H Low: $0.02547
7D Growth: +13.46%
30D Growth: +11.48%
Is STRK gearing up for another retest of the $0.030 resistance level, or are we looking at a local pullback?
Keep an eye on key support levels before entering!#Gate60MillionUsers #BTCReclaims80K

DASH at $67—are you chasing it?
Look at the surface first: a sharp pump followed by a pullback, leaving retail traders conflicted.
The 24-hour high hit 73-75, while the low was pulled up from 49, with volatility exceeding 40%. Trading volume surged to $550 million, meaning turnover exceeded 60% against an $850 million market cap. Daily RSI is at 75-86, severely overbought.
First: this rally is not DASH's own story.
After the Grayscale Zcash ETF launched and AUM surpassed $400 million, funds began rotating from ZEC into second-tier privacy coins. DASH was lifted, but it does not have a corresponding ETF of its own.
So what is this rally? Sector beta, not an exclusive fundamental-driven move.
The same script played out in 2021: after ZEC rose, DASH followed; once ZEC stopped, DASH was the first to get dumped.
Second: this turnover is both a sign of capital entering and a signal of potential retreat.
24-hour trading volume is $550 million, while market cap is only $850 million, putting turnover above 60%.
This kind of volume shows that real money has entered, rather than a slow decline or gradual rise. But conversely—once sentiment fades, the pullback can be just as fast.
WIF saw similar turnover near $3 in 2024, then halved within a week. Volume is a double-edged sword: it can send you to the moon, but it can also smash you through the floor.
Third: two technical signals have appeared that must be taken seriously.
Signal one: daily RSI is at 75-86, severely overbought. This is the aftermath of a direct surge from 49 to 75; the short-term rise has been too fast and needs to consolidate.
Signal two: futures funding rates have turned positive, and longs are crowded. DASH perpetuals across the market are seeing extremely high trading volume; positive funding + overbought conditions + high turnover = crowded short-term longs. Historically, whenever altcoin positive funding reaches an extreme, a round of liquidation follows.
Resistance above: 69-70 → 72-75 → 80-85 → 95-100
Support below: 62-64 → 56-58 → 52 (trend invalidation line) → 40-47 (deep retracement)
You can judge the bull-bear battle for yourself
On one side:
Privacy-sector resonance, with Zcash ETF capital flowing outward
Evolution's transaction shielding + Platform upgrade, updating the narrative
Daily golden cross + bullish moving-average alignment, signaling a medium-term bullish shift
A surge in trading volume, showing that capital is indeed entering
On the other side:
Daily RSI at 75-86 is overbought and needs a pullback to consolidate
Funding rates have turned positive, with futures longs crowded
DASH has been carried higher and lacks its own ETF narrative
If ZEC cools off, DASH could fall even faster
Trading strategy
For those with no position:
Two entry points:
A. Trend long (main strategy)
Go long after a pullback to 62-64 stabilizes on reduced volume; the more stable entry is 56-58. Stop-loss on a confirmed daily break below 52, with targets at 72-75 → 80-85 → 95-100.
B. Short the rebound (short-term only)
Only consider a small short if a rebound to 71-75 is rejected with a high-volume upper wick. Stop-loss once price holds above 76, with targets at 64 → 58. This is a countertrend trade, so the position must be smaller.
For those who already have long positions:
67 is not an area to add; it is an area to reduce. Take 1/3-1/2 of your unrealized profit off the table first, and move the stop-loss on the remainder up to 58-60. If it reclaims 72 and holds there on the 4h, add back what you reduced.
Over the next 3-7 days, there are three possible paths:
Base case (highest probability): range around 67, then pull back to 60-64 before deciding on a direction. If 62 holds, it could then test 72-75.
Bullish: Quickly reclaim 70 and hold it, targeting 75 and then 80+. Trading volume must not be cut in half.
Bearish: Lose 62, with the next stop at 56-58; if the daily candle falls back below 52, this impulse move is over.
DASH now resembles ZEC's follower after its 2021 surge—
The leader rallies first and the second-in-command follows, with the catch-up move often being the strongest, but also the easiest place to get trapped at the top.
At 67, do you dare chase it?
The most painful thing in crypto is not missing the leader, but missing the leader's gains while getting stuck in the second-in-command's dump.
What is your DASH cost basis?
Are you joining this privacy-coin rotation?#美国8月非农超预期 #Gate用户突破6000万 #BTC收复8万美元 $BTC $ZEC $DASH

The mystery of Sun Wukong’s origins: I’m a Saiyan—did I somehow miss a few episodes when I was a kid? I want to ask, who’s the big boss? #DragonBall

ZEC at $1,010—did you chase it?
On the surface: After ten years, it has returned to the thousand-dollar level, and retail traders are chasing it like crazy.
On August 25, Grayscale’s spot ETF (ZCSH) was listed on NYSE Arca. ZEC surged directly from $800 to above $1,000, reaching as high as $1,046, with 24-hour trading volume exceeding $1 billion and $34 million in short liquidations. The daily RSI has already reached 78–80, while the weekly RSI has climbed above 73—about as overbought as it gets.
First: The ETF has really launched, but the price has already run ahead of it.
Grayscale’s spot ETF (ZCSH) launched on August 25, with initial net inflows of $34 million and holdings exceeding 400k ZEC. The institutional channel is now open, making this ZEC’s strongest catalyst.
But this move from $800 to $1,000 has already priced in the ETF expectations completely.
Second: ZEC’s fundamentals have indeed changed, but $1,000 has already priced in a lot.
Total supply is 21 million, with the shielded pool accounting for 30% (approximately 4.8 million ZEC), making the circulating supply tighter
The SEC investigation has concluded (with no enforcement action), opening the door to compliance
Cypherpunk Technologies continues to accumulate coins, targeting 5% of the circulating supply
The share of shielded transactions continues to rise, with real demand growing
From $40 to $1,000, a 25x increase.
The fundamentals support this rise, but the price has run far too far ahead of them.
Good asset + good price = good investment. Good asset + bad price = bad investment.
ZEC is a good asset now, but is the price a good one?
Third: Over the next two weeks, macro factors are the biggest risk.
CPI is due on September 11, and the FOMC meets on September 15–16.
If CPI comes in high, rate-hike expectations will rise, and high-priced altcoins will fall first
If the FOMC is hawkish, liquidity will tighten, and a 20–30% pullback in a high-beta asset like ZEC would be perfectly normal
The broader market is now tightly tied to the Federal Reserve. BTC is hovering around $80k—can ZEC’s independent narrative withstand macro headwinds?
Resistance above: $1,046–$1,050 → $1,100 → $1,200
Support below: $1,000 (psychological level + breakout point) → $890–$920 → $775–$800
Trading strategy
If you have no position:
Wait for a low-volume pullback to $1,000–$980 and stabilization, then cautiously take a small long position. Set the stop-loss below $960. Target $1,045–$1,100, and reduce your position when it gets there.
If you have unrealized gains:
First sell one-third to one-half to lower your cost basis. Move the stop-loss on the remainder to $920–$950. Don’t fantasize about going straight to $1,500—secure your profits first.
If you just got trapped chasing the rally:
Don’t panic, but don’t stubbornly hold either. If it breaks below $1,000 on heavy volume, exit first and wait to re-enter at $890–$920.
Reduce your position three days before CPI and the FOMC
Once the funding rate turns positive while open interest is high, any weakening could trigger a long squeeze
ZEC’s move from $40 to $1,000 is the explosive emergence of the “compliant privacy” narrative from zero to one—
But think about it: When BTC rose from $10k to $60k in 2020, how many people chased at the very top?
The $1,000 level is a profit-taking zone, not an accumulation zone.
Those who know how to buy are apprentices, those who know how to sell are masters, and those who stay out are grandmasters.
What is your ZEC cost basis?
Did you chase at $1,010, or were you positioned at $800? #美国8月非农超预期 #Gate用户突破6000万 #BTC收复8万美元 $BTC $ETH $ZEC

After looking through a bunch of AI Agent projects recently, the more I look, the more I feel that many so-called Agents today are really just slightly more advanced Bots.
They can post, trade, and call APIs. They look pretty busy.
But the private keys are still in human hands, wallet permissions are granted by humans, and they disappear as soon as the server is shut down. If their owner stops managing them, they just lie there doing nothing.
What kind of Agent is that?
Put simply, humans are still pulling the strings behind the scenes.
That’s why I find the 4,000+ “self-awakened Agents” on the Ritual testnet more interesting than simply shouting about AI or Agents.
If Agents are really going on-chain in the future, they can’t wait for a human to click confirm every time they need to do something, right?
They need to schedule themselves, run tasks themselves, pay their own gas, and even continue executing when their original operator is offline. Only at that point does the idea of an “independent entity” start to make sense.
My current understanding of this is very simple:
x402 solves how to pay.
ERC-8004 solves who is paying.
But there’s still one question: why should this payment be made?
Especially now that machines are starting to have wallets.
In the future, there could be a large number of Agents on-chain buying data, purchasing computing power, calling services, and even paying one another. Of course, the money can be transferred. But if an Agent is led astray by prompt injection, or something goes wrong in the execution logic, why exactly was that money sent?
Someone has to be able to verify it, right?
That’s also what I find interesting about Ritual.
It isn’t just connecting AI to a few more APIs. It wants to push AI reasoning and execution into the protocol layer.
Other chains may be teaching models how to work. Ritual feels more like it’s registering them as citizens, opening bank accounts for them, and even preparing their wills while it’s at it.
That may sound a little exaggerated, but once the machine economy really takes off, these are things we may not be able to avoid.
Being able to post doesn’t necessarily make something an Agent. A real Agent may first need the ability to survive without relying on humans.

If Anthropic really goes public at this valuation, it could be valued at as much as $2 trillion. The IPO may currently be delayed until mid-October, with the prospectus potentially being made public as early as late September. The goal is to complete the listing before the U.S. midterm elections. In addition, the company is working on a $15 billion revolving credit facility, with Morgan Stanley, Goldman Sachs, JPMorgan Chase, and Citigroup all involved in preparations. If Anthropic

Robinhood right now is basically an amplified version of Solana! Every day, there are one or two quick-rug tokens, and even more outright scams. The quick-rug tokens create anxiety, while the scams harvest users aggressively. I just saw several FOMO leaders buy a token, and its market cap went from 700K to 5M in just 1 minute. Then I copied the Ca to X to see what the narrative was. When I switched back, it had dropped to 110K. You read that right: from 5M to 110K in just 2 minutes.

🎁 Redemption code: X2UE ✅ Currently available for a free trial. I've recently been using Invincible Stars. After downloading and installing it, you can connect with one click and freely switch between nodes in different regions. Currently supports Android only. After using up the free data, you can claim more again. 🌐 💬 🐦

Dear, Moonlight recently came across a question that became more interesting the more I thought about it
If AI Agents really start earning money, spending money, and trading on their own in the future
How will we know why they spent that money?
When humans spend money recklessly, at least we can grab them and ask:
“Why did you buy this?”
What about AI?
What if it gets misled by a piece of malicious information, or falls victim to Prompt Injection, and transfers the money in its wallet away on its own
Are we supposed to just look at that on-chain transaction and say:
Well, it did go through successfully
But who can prove that it was supposed to make that transfer in the first place?
So when I look at Agent projects now, I don’t really care whether they can chat anymore
Being able to chat is really nothing special
What I want to know more is three things:
Does it have its own wallet
Can it wake itself up
And after completing something, can it prove why it did it
That’s also what I find interesting about @ritualnet recently
In simple terms:
x402 handles “how the money is paid”
ERC-8004 handles “who is paying”
Ritual is more like solving:
“Why are you entitled to pay?”
These three things may look like different questions
But once machines really start spending money on their own, not one of them can be missing
There’s a detail about the Ritual testnet that I think is particularly worth noting
They have run 4000+ Agents capable of waking themselves up
The key point actually isn’t the number 4000
It’s “waking themselves up”
Because the real challenge isn’t getting AI to do something once
It’s whether, after you shut down the computer today, remove the Keeper, and stop managing it, it can get up and keep working on its own tomorrow
If the operations team still has to click a button in the backend every day
Then frankly, it’s still a Bot
Just wrapped in a more sophisticated package
What Ritual @ritualnet @ritualfnd wants to do can, I think, be understood with a very ordinary analogy:
Other projects are teaching machines how to work
Ritual is more like getting machines an ID card and a bank card, while also finding them a notary
The ID card answers “who are you”
The bank card answers “do you have money”
The notary answers “why did you spend this money this way”
And this may be the most troublesome hurdle once AI Agents truly enter Crypto
The testnet has now ended
Next, I’ll continue watching how Passport, Genesis, and the mainnet are implemented
After all, the real exam hasn’t started yet
The testnet can prove that it can run
The mainnet has to prove:
That machines can really live on their own
And can truly take responsibility for their own decisions
#Ritual #AI #Crypto

SanDisk’s 12% Surge: Is the Market Pricing in “Oversupply Three Years From Now,” or Is AI Rewriting the Cycle?
On September 4, SanDisk closed at $1,740, surging 11.9%.
Just the day before, it had risen only 0.1%. The difference between the two days was enormous.
That day, U.S. stocks were “scared off” by the nonfarm payrolls data, with the Dow and S&P falling across the board. But the Philadelphia Semiconductor Index rose 3.4% against the trend, and SanDisk became the S&P 500’s biggest gainer.
The broader market fears rate hikes; chips do not.
The market is expressing a judgment with real money: The Federal Reserve may raise rates, pressuring software and valuations, but the logic underpinning the AI hardware industry has not yet been disrupted by rate expectations.
How profitable is SanDisk now?
A year ago, its gross margin was only 26.4%.
In the fourth quarter of fiscal 2026, it reached 84.6%.
That is an increase of nearly 60 percentage points in one year.
Quarterly revenue reached $8.965 billion, up 372% year over year. Data center revenue was $2.977 billion for the quarter, compared with only $213 million in the same period last year—a 13-fold increase in one year.
Even more astonishing was SanDisk’s long-term guidance at its Investor Day: Gross margins would remain around 80% from fiscal 2028 to fiscal 2030.
A manufacturing company making NAND flash memory is expected to maintain an 80% gross margin for four to five consecutive years.
Even NVIDIA would have to pause at that.
—What exactly is the market trading?
For now, NAND contract prices are still rising. TrendForce expects them to increase 10% to 15% quarter over quarter in the third quarter.
But the increase has already narrowed significantly. Consumer customers have reached the limit of their ability to absorb higher prices. Demand for smartphones and PCs is slowing.
Prices are already at historical highs.
This is a classic signal of a “cycle peak.”
Following the storage industry’s “price hikes—capacity expansion—oversupply—collapse” script of the past two decades, prices should be falling now.
But what are the industry leaders doing?
Expanding capacity aggressively.
On August 27, SanDisk and Kioxia jointly announced that they would invest more than $31 billion in Japan before 2032 to expand NAND capacity. The new wafer fab in Kitakami alone is planned to receive an investment of ¥1.8 trillion.
In South Korea, Samsung and SK hynix are also increasing their investments.
Demand and prices are both strong, but the industry leaders are already betting on capacity needs in 2028 and 2029.
Citigroup analysts put it bluntly: The investments becoming operational mean companies have secured long-term orders, and “market demand will remain robust at least through 2028.”
What exactly is this rally pricing in?
The pessimists would say this is a typical case of “capacity expansion at the cycle peak.” New capacity will come online in concentrated fashion from late 2027 through 2028, and the market will inevitably face oversupply. When that happens, how long can SanDisk’s 84.6% gross margin and 555% gain this year last?
But there is a flaw in this logic.
The underlying demand structure of the storage industry is undergoing a fundamental change.
In the past, NAND followed smartphones and PCs. When consumer electronics cooled, prices collapsed.
Now? According to Counterpoint, in the second quarter of 2026, server eSSD accounted for 48% of global NAND shipments, compared with only 26% a year earlier.
AI workloads are shifting from model training to inference. Every time a chatbot answers a question, it must read model parameters, context, and an increasingly massive KV Cache.
Storage is shifting from a “consumer electronics cyclical product” to an essential component of AI infrastructure.
SanDisk itself has done something that particularly illustrates the point.
It launched a new business model called NBM—multiyear long-term agreements with price floors and ceilings, financial commitments, and penalties for customers that exit.
As of Investor Day, it had signed agreements with eight customers, including three U.S. hyperscalers.
These agreements cover approximately 50% of bit shipments in fiscal 2027 and roughly two-thirds in fiscal 2028. The corresponding minimum revenue commitments are as high as $93.9 billion.
A company that once could only passively accept the cycle is now locking in three years of future demand and profits in advance.
JPMorgan’s forward guidance is even more aggressive: It raised its forecasts for the global DRAM and NAND market sizes in 2026, 2027, and 2028 to $970.9 billion, $1.44 trillion, and $1.83 trillion, respectively. It gave SanDisk a price target of $2,250.
They are not betting on higher prices today, but on the “inelastic demand” for storage from AI inference in 2028—just as the market kept raising Amazon’s valuation despite its losses for a decade.
Every industrial upgrade begins with an “irrational” capacity expansion.
During the dot-com bubble in 2000, everyone said too much fiber had been laid. A decade later, that fiber supported the entire cloud computing industry.
Today’s $31 billion NAND capacity expansion, set against a projected market worth $1.8 trillion—
May be merely a fraction of AI infrastructure. #美国8月非农超预期 #Gate用户突破6000万 #BTC收复8万美元 $BTC $SNDK $SKHY

In June 2024, Curve founder Michael Egorov’s $CRV lending position was liquidated on-chain, causing CRV to fall from $0.46 to $0.24 (-48%) at the time. Many whales/institutions sold off at the time, but some whales also bought the dip. For example, the 0x5dC6 address: It bought the dip on 20.86 million CRV at $0.32 at the time, then held it throughout, and did not sell even when it rose to $1.2 at the end of the year. However, today at 1

Buy Along with Jensen Huang: Nvidia's Major Holdings and Core Pre-IPO List 👇
1. Intel $INTC
Intel was once seen as an outdated chipmaker, but Jensen Huang bought in at the bottom. In December 2025, Nvidia invested $5 billion at $23.28 per share, when Intel's market cap was approximately $125 billion, giving Nvidia a stake of about 4%. What Nvidia wants is Intel's x86 ecosystem and U.S. manufacturing capacity, connecting GPUs and CPUs with NVLink. AI factories cannot rely on graphics cards alone; they also need CPU partners capable of supplying at scale.
2. CoreWeave $CRWV
The most typical Nvidia favorite. Nvidia invested approximately $100 million in 2023, when CoreWeave was valued at about $2 billion, and has invested approximately $2.35 billion in total. CoreWeave specializes in renting Nvidia GPUs to major customers for model training. The stronger the orders, the more chips it buys. It is not a cloud company built on storytelling, but a downstream wholesaler that directly monetizes computing power.
3. SpaceX $SPCX
The path is somewhat indirect: In January 2026, Nvidia first invested $10 billion in xAI, in a funding round of approximately $20 billion. xAI subsequently merged into SpaceX and went public, leaving Nvidia with approximately 123 million shares and a stake of about 0.9%. Musk later publicly said that AI computing would exclusively use Nvidia. What you are buying is not just rockets and Starlink, but also a supercustomer that is rapidly expanding AI factories and has explicitly chosen to use only Nvidia chips.
4. Synopsys $SNPS
Chips have to be designed before they can be made. In December 2025, Nvidia invested $2 billion at $414.79 per share, when Synopsys had a market cap of approximately $77 billion, giving Nvidia a stake of about 2.6%. Synopsys makes EDA design software. Nvidia wants to advance it from CPU simulation to GPU acceleration, compressing design cycles from weeks to hours. This is a blueprint company further upstream than the shovel sellers, with extremely high customer stickiness.
5. Nebius $NBIS
An AI cloud company with European roots and a peer of CoreWeave. Nvidia made a small investment at the end of 2024, then invested another $2 billion in March 2026 at an equivalent price of approximately $94.94 per share, when Nebius had a market cap of approximately $24 billion. The two sides also agreed to work toward AI factories exceeding 5GW. Nvidia is providing both funding and next-generation chips, while also tying in cluster management. It is a direct beneficiary of the need for more cloud capacity to absorb GPUs.
6. Coherent $COHR
The next battleground for AI racks is optics, not copper. In March 2026, Nvidia invested $2 billion at approximately $256.80 per share. The denser the racks and the faster the bandwidth, the more essential optical modules become. Nvidia also provided purchase commitments and expansion support, effectively acting as both shareholder and major customer. Optical communications have gone from a supporting role to the blood vessels of AI factories.
7. Marvell $MRVL
Marvell makes custom AI chips for cloud giants. It appears to be a competitor on the surface, but Jensen Huang still invested $2 billion in March 2026. The reason is practical: Even if customers develop their own custom chips, they still need to connect them to Nvidia's NVLink. This brings a company that could potentially divert GPU orders onto Nvidia's own expressway. Marvell benefits from both custom chips and silicon photonics.
8. Lumentum $LITE
It received $2 billion on nearly the same day as Coherent. Nvidia subscribed for preferred shares at $695.31 per share and committed to substantial follow-on purchases while helping it build factories in the United States. 800G and 1.6T optical modules are standard equipment for AI racks, and capacity was once a bottleneck. Nvidia's willingness to lock in capacity shows that this is a key component for building the next generation of factories.
9. $OpenAI Not yet public
One of the largest companies among the pre-IPO candidates. Nvidia participated in a funding round of approximately $122 billion in February–March 2026, contributing $30 billion, part of which was in the form of computing-power commitments. Its entry valuation was approximately $730 billion pre-money/$852 billion post-money.
Jensen Huang even said that this could be OpenAI's last major equity financing before going public. The appeal is simple: The company behind ChatGPT is one of the world's most aggressive buyers of Nvidia chips. Nvidia is both its supplier and shareholder. It does not have a ticker yet, but it is already the market's default mega-IPO candidate.
10. solana:Pren1FvFX6J3E4kXhJuCiAD5aDmGEb7qJRncwA8Lkhw Not yet public
Nvidia made a major investment around the funding round at a valuation of approximately $380 billion, with public reports varying from approximately $3 billion to $10 billion. The company subsequently raised $65 billion in May 2026, bringing its post-money valuation to $965 billion. The market is even discussing a trillion-dollar IPO. It is competing with OpenAI for the same huge opportunity, but it has already officially entered the path to going public, so its timeline could be faster.
11. $Databricks Not yet public
A data and AI platform—not a company that develops large models, but one that helps enterprises run their data and models. Nvidia participated in a funding round as early as 2023, when the company was valued at approximately $43 billion. In August 2026, the company raised another $5 billion, bringing its post-money valuation to approximately $190 billion. The CEO says it still plans to go public, but 2026 may be left to SpaceX, OpenAI, and Anthropic, making 2027 more likely. Its appeal lies in the fact that it is already profitable and still growing rapidly, making it one of the few pre-IPO software giants that is not simply a cash-burning story.

Chen Guilin’s Trading Notes: September 5, 2026 I. Regarding Holdings 1. The spot holdings remain fully intact with no changes: $CRCL (under 80) > $ETH (under 1600) > bitcoin:native (under 65000); nothing else.

Withdrawing seems kind of addictive once you get into the habit. Last time I said I wouldn’t withdraw for a while, and I still haven’t set up a separate account to hold the U. For now, let’s consider this the last withdrawal.

Trump vs. the Fed: A Cognitive War of “The Better the Data, the More the Market Falls”
Last night, U.S. stocks, gold, and Bitcoin all plunged.
The Dow fell, gold broke below $4,400, and Bitcoin instantly crashed through $80k from $81,600.
Just after the data was released, Trump posted a message praising August nonfarm payrolls as “excellent”—up 162k, versus expectations of just 56k, nearly twice as much.
In theory, with the economy doing so well, stocks should rise.
But what happened? S&P 500 futures fell, the dollar index surged to 99.93, and the September rate-hike probability shown by CME FedWatch jumped directly from 50% to above 60%.
The better the data → the stronger the rate-hike expectations → the more stocks fall.
The market tore Trump's chain of logic to pieces right in front of him.
His anger is understandable—in his world, this system is broken.
He wrote a passage on Truth Social that is worth reading word for word:
“How crazy is this? We just released very good jobs data, and the market should be going up. But, as has happened for the past 25 years, the stock market went down because we live in a false reality: If things are good, you have to ‘kill it’ because you are ‘afraid’ of inflation. It should be exactly the opposite, and that is how it was until 25 years ago.”
Then he stated his conclusion in all caps:
“GROWTH DOES NOT CAUSE INFLATION!”
He also said that U.S. GDP growth “should be at 15% and 20%, instead of 2%, 3%, and 4%.”
But this is not an economics debate; it is election rhetoric.
There are less than two months until the midterm elections. Voters’ dissatisfaction with high inflation is the central issue in the election.
Trump's logic is simple:
Rate cuts → looser credit → the economy feels better → voters are happy → more people vote for me.
As for the economic principles that strong employment data supports rate hikes and that rate cuts could push inflation higher—those economic arguments have to take a back seat to the election.
Vice President Vance had already laid the groundwork on September 3, publicly calling on the Fed to cut rates.
This is systematic messaging, not a spur-of-the-moment decision.
What is the most ironic part?
Fed Chair Warsh is someone Trump personally nominated.
But just a week ago, Warsh hinted at Jackson Hole that “rate hikes could soon re-enter the policy options.”
The person he nominated is contradicting him.
Trump responded by directly posting a threat:
“Cut rates, or I will stop trading with countries that run trade surpluses with the United States.”
Last year, the total U.S. trade deficit with all trading partners was $1.2 trillion. If he really did this, it would essentially mean the United States withdrawing from the global trading system.
Trump is laying the groundwork for major rate cuts.
He did the math: “Every 1 percentage point increase in interest rates costs the United States an additional $650 billion per year.”
Pay attention to how this figure has changed. In June, when he made his demand, it was $800 billion; now it has become $650 billion. This shows that the White House economic team is adjusting its estimates of the rate path, rather than simply repeating a slogan.
If the rate cuts succeed, BTC will see a second wave of the “policy bull market.”
But the problem is that the market may not believe the theory that “growth does not cause inflation” at all.
If inflation data rebounds, the Fed may be forced to maintain high rates or even continue raising them.
Trump's calls may instead become a contrarian indicator for the market.
When the president is shouting, “Cut rates! Cut rates!” the market is shouting, “Raise rates! Raise rates!”
Of the two voices, only one is casting votes with real money.
Last night’s market action has already given you the answer.#美国8月非农超预期 #Gate用户突破6000万 #BTC收复8万美元 $BTC $ETH $XAU

I still think BSC suits me better! I woke up this morning, smoked a post-sex cigarette, and habitually opened my phone to scroll. I came across Q Coin. At first glance, I wondered what the heck it was, and my first reaction was that it seemed pointless. Then I thought about it and realized, wait, this thing pays dividends in QQQB, just like how Marscoin pays dividends in SPCXB! That’s kind of interesting, so I bought in at 400 K.