海盗鸭

Rare warning signs are emerging in the U.S. housing market: by August, the degree of seller surplus has surged to 58%, the highest level on record. Even more worrying is that this record isn’t due to a spike in listings; it’s because buyers are accelerating their disappearance—people lining up to sell homes are plentiful, but would-be buyers are becoming increasingly scarce, widening the supply-demand gap to a historical extreme. The relationship with the crypto market is more direct than you might think: housing is the biggest pool of wealth for American residents. Once expectations for home prices weaken and the wealth effect reverses, a contraction in risk appetite won’t stop at the stock market. Look at Bitcoin over the past week: after a golden cross signal appeared, it pulled back to $77,000; then following the CPI release, it rebounded and reclaimed the $79,000 level. This kind of volatility in itself is proof that it is highly sensitive to macro conditions. If the housing market keeps deteriorating and ultimately forces a shift toward more accommodative policy, it could be a medium-term positive for $BTC ; but before that happens, we need to first withstand the test of tighter liquidity. Will you adjust your position in response to the housing market signals?

After the CPI data is released, $BTC doesn’t fall but rises again, reclaiming the $79,000 level. This price action is worth breaking down because it runs counter to most people’s intuition. The data itself is on the hot side; following textbook logic: hot inflation → expectations of tightening → risk assets fall. But the market’s tape is giving the opposite answer, which suggests that what the market is trading right now isn’t “how high inflation is,” but rather “whether the Fed dares to act.” The logic chain is: as long as the Fed holds steady in September, the rise in real rates gets pushed out, worries about tighter liquidity ease temporarily, and the previously accumulated short positions become targets for being squeezed—part of this rebound’s momentum comes from short-covering, not entirely from fresh long entries. You can tell from how fast the rebound happened. This is also an honest test of the “digital gold” narrative: if Bitcoin were truly an inflation-hedge asset, a hot CPI should push it straight up. The reality is that it’s far more sensitive to rate expectations than to inflation itself; the up-and-down rhythm looks more like a high-beta growth stock than like gold. This distinction matters a lot for position management: don’t use the logic you’d use to allocate to gold to set your Bitcoin position size. In terms of levels, $79,000 is only a reclaim of lost ground; what really matters is the psychological $80,000 integer mark. If it holds above $80,000, the low point of CPI Night is likely to be the bottom of this pullback. If it can’t hold, the $77,000 area will likely have to be tested again. Next, the biggest variable isn’t the chart—it’s the September Fed meeting itself. Volatility typically compresses before the meeting, but the real pricing happens around and after the decision is delivered. Which side are you on—longs or staying on the sidelines? #BitcoinCPIReboundReclaims79000

Core CPI Month-over-Month at 0.3%, above expectations—this set of data is worth every trader taking seriously, because it explains why the “inflation is cooling” story has been told for so long, yet the market’s easing trade has never dared to put on heavy bets.

Let’s start with the mechanism. Core CPI excludes food and energy and reflects the “background tone” of inflation. A 0.3% MoM reading annualized is clearly well above the Federal Reserve’s 2% target path, and the main driver is almost always the same: the housing component. Here’s a mechanism that many people don’t know: the rent tracked in the CPI reflects existing rental leases, while market rents (prices in newly signed leases) often lead the official data by about six months to a year. So you’ll see a split: new-lease prices are cooling, but the housing component in the CPI is still rising. This isn’t a problem with the data—it’s a lag from the statistical methodology.

The policy implication is this: as long as housing “stickiness” isn’t cleared, core inflation is hard to return to target smoothly, so the Fed can only use “gradualism” to hedge uncertainty. That’s also why after this kind of data is released, market reactions are often not panic but recalibration—risk assets face short-term pressure, Treasury yields move higher, and then things are priced in more gradually.

What to look at next? Three points: First, whether subsequent PPI and PCE move in the same direction as CPI—to confirm whether it’s a trend or just one month of noise. Second, the balance of emphasis in the September FOMC statement: is it focused on “inflation risks” or “employment risks”? That will determine the trading axis for the next quarter. Third, how liquidity-sensitive assets like $BTC digest the news—if negative data can’t knock prices down, it suggests the underlying risk appetite has already changed. This “small matter” of macro data is often the fuse for the next round of the market.

#U.S. August core CPI rose 0.3% MoM, above expectations

This Bitcoin golden cross is a bit “ill-timed”: the 50-day moving average has just crossed above the 200-day moving average, yet the price has fallen back to around $77,000. Many people’s first reaction is “the golden cross failed,” but when you lay historical data out, this is precisely the norm rather than the exception. After the 2015 golden cross, the price first went sideways for several months; after the mid-2019 golden cross, it was followed by nearly half a year of downside drift; and after the 2023 golden cross, it also initially pulled back before starting up again. The reason isn’t complicated: a golden cross is a lagging indicator. It confirms that “the average buying pressure over the past 50 days has been stronger than the past 200 days.” By the time the signal appears, the first wave of bargain hunters often has already taken profits. That’s why real technical traders don’t focus on the crossing moment—they look at two things after the cross: first, whether the pullback to the 200-day moving average can hold with reduced volume; second, whether during the second push higher, trading volume expands in sync. The practical impact on the broader market is this: the golden-cross narrative pulls a batch of trend-following capital and quantitative strategies back into long positions. These funds tend to have strict stop-loss discipline. Once $77,000 breaks, they will leave without hesitation, and then altcoins’ high beta will amplify the drawdown. Conversely, if <t-2/>$BTC manages to complete sufficient rotation around $77,000 and then strengthens again, this batch of trend capital will accelerate its entry, becoming the fuel for the second wave. In other words, the current pullback isn’t a sign that the signal has been disproven—it’s the golden-cross signal’s first pressure test. The pass marks are lower volume and holding the line, not immediately printing new highs. #BitcoinGoldenCrossPullsBackTo$77000

OpenAI releases Agent API beta testing, and many people’s first reaction is “yet another new term.” But this time it’s really not that different—its difference from the ChatGPT chat box is like the difference between a “customer service hotline” and “hiring an employee.”

In the past, when you asked an AI a question, it would return a piece of text and that would be the end of it. The logic of the Agent API is completely different: you give it a goal, and it breaks the work into steps, calls tools, checks results, discovers when it has gone off track, and then goes back to fix it—until the job is done. Searching, checking files, running code, operating interfaces—actions you previously had to string together manually—can now be executed continuously by it.

So what should ordinary users pay attention to? I’ll give you three actionable points.

First, don’t rush to write code—think first about “what repetitive work do I have.” Agents are not best at chatting with you; they excel at handling processes with clear rules: weekly information summaries, cross-verifying data across multiple sources, and organizing materials into a fixed format. Any repeated tasks you can list out are exactly where an agent shines.

Second, look at functionality, not demos. Demos during product launches often wow people, but what truly determines the experience is reliability—when it executes ten steps in a row, and the ninth step fails, does it cascade into getting everything wrong? In the beta phase, the most worth watching metric is “error-correction ability,” not how smart the response is in a single moment.

Third, be wary of over-authorization. For an agent to do work, it needs permissions: to read your files, send messages on your behalf, and even touch your accounts. The more permissions you grant, the higher the cost when it makes mistakes. In the beginning, it’s recommended to grant only read-only permissions, then gradually expand as you test and get comfortable.

In short, this beta testing is taking “AI that can get work done” out of large companies’ exclusive capabilities and turning it into a standardized interface. Ordinary people don’t need to jump in and use it right away—but it’s worth starting to observe now: in your industry, which parts will be taken over first by tools that “deliver once given a goal”? This kind of judgment is more valuable than chasing any hot trend.

What jobs do you have that you’re willing to hand over entirely to an agent? Let’s discuss in the comments.

#OpenAI推出智能体API公测

What Dell’s latest signal disclosure is worth noticing: AI server orders are surging, storage business is growing in parallel, and the supply chain is clearly tight. Translate it—on the demand side things are accelerating, but the supply side can’t keep up. This is usually the combination that shifts bargaining power toward sellers, where price hikes and queues often show up at this stage.

For $DELLB , this is a rare opportunity for an identity upgrade. In the past, the market valued Dell as a "PC and traditional server vendor." Now, as the share of AI infrastructure orders keeps rising, the story is shifting toward an "enterprise AI hardware integrator." The more order backlogs there are, the higher the revenue certainty for the coming quarters.

Of course, there are also concerns: component supply tightness could limit the delivery pace, and any improvement in profit margins will need time to be validated. But at least for now, what the market is seeing is real demand—not just storytelling. If AI capex continues to move higher, Dell could be one of the traditional hardware makers with the greatest flexibility. In your AI hardware allocation, is there Dell?

The thing about BNB breaking above $730—viewed over just one day, it doesn’t seem like much, but put it into the timeline and it gets interesting: since February this year, it has been trading within a box above $700; in May it tested around $720; and by mid-September it closed around the $730 line. Over more than half a year, this “stair-step up” style is rare among major coins—no storytelling, just climbing the stairs. Behind this slow bull run usually lie two engines. One is the supply side: BNB’s quarterly burn mechanism keeps reducing the circulating supply. The automated burn formula is tied to the coin price and on-chain block production; the long-term goal is to squeeze the total supply down to 100 million. The deflation logic is clear, on-chain verifiable, and it doesn’t rely on emotion. The second is fundamentals: in 2025, BNB Chain reduced block time from 1.5 seconds to 0.75 seconds (the Maxwell upgrade). On-chain interaction costs and confirmation experience have improved noticeably. The chain’s real activity volume is the underlying support for the price. What should regular users focus on? I suggest watching three things: first, the bottom of the $700 box—how long it holds determines whether this leg is merely a rebound or a trend. Second, the burn amount in the next burn announcement—that’s the most direct supply-side catalyst. Third, the volume expansion in the $730–$750 range: a breakout above the top of the box needs real volume to confirm. Breakouts that happen on low volume are more likely to be false breakouts, while pullbacks followed by confirmation are generally the healthier path. One more reminder: the biggest risk for a slow-bull asset isn’t the pullback—it’s getting shaken out the night before the breakout. Holding is conditional on your position size letting you sleep well; if your direction is right but your position is wrong, you still won’t be able to make money. $BNB #BNB涨破730美元

US August core CPI month-over-month rose 0.3%, exceeding market expectations; the final mile of inflation is more stubborn than imagined. After the data was released, US Treasury yields and oil prices rose in tandem, and the Dow fell by about 300 points. Tech stocks saw internal divergence: Nvidia weakened while Apple rose against the trend. This CPI directly affects the September FOMC decision: core inflation came in hotter than expected, meaning the threshold for Fed rate cuts has been raised a bit again. For the crypto market, this helps explain the recent back-and-forth: after the CPI release, Bitcoin briefly rebounded and climbed above $79,000, but as long as interest-rate expectations stir again, the market could just as easily pull back. Risk assets are all waiting for the same thing right now: whether the Fed will give September a foothold. The $NVDAB $AAPLB divergence is essentially the market resetting the pricing of rate-sensitive assets. Do you think a hotter-than-expected core CPI will change the September rate-cut path?#US August core CPI rose 0.3% month-over-month, beating expectations

I was genuinely a bit shocked by this round of joint signatures by Terence Tao and others. In the past, when it came to public letters like this, it was usually scholars who were just hopping on the hotspot—filling in numbers. This time, though, everyone who signed had actually done real work. The people behind these efforts have spent their whole lives being most annoyed by “movement-style” public stances. And now they’re willing to sign together, which really shows the matter has hurt them.

From what I understand, it means something like this: AI companies are going crazy grabbing papers, question banks, and proof processes to train their models. As the models improve, they can increasingly spit out answers directly. So in the future, how many young people will still be willing to sit in that ten-year cold-plateau grind, chipping away at one problem? Anyway, if you ask an AI, you basically get an answer with a pretty reliable direction.

In the short term it’s about efficiency. In the long term, it’s like pulling out the foundations of the discipline. The most ironic part is that among the job categories that were “freed” by AI first, there are mathematicians. A few years ago, people were still joking that doing mathematics is the safest, because machines can’t extract abstract thinking—yet the first part to crumble is this one.

I don’t think mathematics will die—math can’t die. We’ve been getting choked off for two thousand years already. What I’m worried about is the middle layer: young researchers who can’t reach the level of the masters, but also can’t outcompete the machines. If this group leaves, where will the next generation of Terence Tao come from? Nobody knows.

TSMC’s revenue hits a record high, and one thing is now essentially settled: AI chip demand really is booming—not a story from a PowerPoint deck. The very first place to see real orders is at the chain’s topmost upstream. Foundry capacity is running at full stretch—more convincing than any research report. When demand is at its hottest, the bargaining power at the foundry pricing negotiation table is quietly shifting, too.

But the next phase of the story has changed. The competition between NVIDIA $NVDAB and AMD $AMDB is shifting from “whose card is stronger” to “who can protect profit margins.” NVIDIA’s premium remains firm, while AMD is taking share with cost-effectiveness. And regardless of who wins, wafers still have to queue to place orders with TSMC $TSMB . The upstream earns the money of certainty, while the downstream earns the money of a fight.

This is also a typical sign that the AI cycle is entering its second half: β (overall industry volume) has already played out; what comes next is α (the competitive landscape). Will you allocate your position to the upstream’s certainty, or to the downstream’s upside potential?

In this revision of the CLARITY Act, what looks like a technical adjustment is actually one of the most critical watershed moments in U.S. crypto regulation: whether people who write code should be held responsible for wrongdoing committed by others using that code.

First, some background. The CLARITY Act (H.R. 3633) is a bipartisan digital-asset market framework whose core mission is to draw clear jurisdictional lines between the SEC and the CFTC. But the bill includes a “control-person liability” provision: if you “control” an entity that carries out fraud or manipulation, you are jointly and severally liable in civil cases. The question is—who counts as a “control person” in DeFi? Does a smart-contract developer count? Does someone who contributes code count? Does someone who runs nodes count? Under the original wording, in theory, they could all be swept in.

This revision is aimed directly at that loophole: it clarifies that this clause primarily applies to non-DeFi control persons—those who truly have actual control over the violating party, rather than merely people who develop or publish source code. Developers and node operators will no longer automatically be saddled with secondary liability just for writing code.

That’s where both sides diverge. Supporters (such as DeFi education foundations) see it as a return to common sense: without actual control and knowledge, you shouldn’t be held liable; otherwise open-source development in the U.S. would effectively be frozen to death. Opponents, however, worry about another extreme: some projects may deliberately market themselves as “decentralized” and use the appearance of no control to evade substantive trading or operational responsibility. Fraudsters hide behind a “decentralized theater,” and victims are left with an even harder time pursuing accountability.

Both concerns are valid—this is exactly what makes legislation difficult. If the liability threshold is too low, innovation is stifled; if it’s too high, wrongdoing is effectively excused. The revised core language is “actual control,” but what counts as “actual” ultimately will be defined by SEC and CFTC enforcement practices—that is, precisely, the most uncertain part.

Next, there are two things to watch: corresponding developments on the Senate side, and the first batch of cases in which regulators will take “control-person” provisions as the target. Those will be the real litmus test.

#CLARITY Act revision would规范非DeFi control persons

As soon as August core CPI printed a 0.3% month-on-month figure, the question people originally didn’t take seriously—“Will there be a rate hike in September?”—suddenly became a serious topic on trading desks. First, let’s do the math: a 0.3% MoM reading roughly annualizes to around 3.7%, nearly twice the 2% policy target. And this isn’t a one-month anomaly—core inflation persistence has been going on for several quarters; the services and housing components have consistently failed to cool down. But “should they raise rates” and “will they raise rates” are two different things. Historically, the Fed almost never pivots immediately based on a single CPI print—especially when the direction runs counter to the current policy cycle. One month of data is too noisy; the Fed pays more attention to the compounded trends over three and six months, and whether inflation expectations are losing their anchor. So what’s truly worth watching for the September meeting isn’t the outcome of that day’s decision itself, but two things: first, whether anyone in the rate dot plot has revised the path upward; and second, the wording from the new leadership—the chair Powell’s term ended in May, and the market’s biggest unknown is the new administration’s tolerance for inflation. For crypto markets, the transmission is fairly direct: rate-hike expectations rise → real yields move higher → the U.S. dollar strengthens → risk assets come under pressure. Long-duration assets like $BTC are especially sensitive to liquidity. On the day CPI was released, the market’s hesitation to drop below 77,000 and then the rebound back above 79,000 essentially reflects a bet on the window where “the data is hot, but the Fed isn’t willing to act.” In the week ahead, any speech from a Fed official could move the market more than the data itself. #CPIdataarrivesCanittriggeraSeptemberratehike

The same CPI, three sets of numbers telling three different stories—this is the most interesting part of the August report.

Story one: Headline CPI year over year rose from 2.7% to 2.9%, above expectations, suggesting inflation is rebounding. Story two: Core CPI year over year fell from 3.2% to 3.1%, below expectations, suggesting inflation is cooling. Story three: Core CPI month over month came in at 0.3%, above the expected 0.2%—the largest monthly increase since February—and it also looks like inflation is accelerating.

Why do the three figures seem to conflict? The mechanism is the “base effect.” The year-over-year reading equals today’s prices divided by last year’s prices from the same period. If the denominator fell sharply last year, the year-over-year figure will be automatically lifted this year; if the denominator rose sharply last year, it will suppress the year-over-year figure. So the rebound in headline year over year is, to a large extent, a statistical artifact driven by a low base—whereas the 0.3% month over month figure better reflects current momentum. Annualized, it’s roughly 3.6%, still clearly higher than the pace consistent with the 2% target.

This is exactly why the Federal Reserve watches month over month: year over year tells you what “happened in the past,” while month over month tells you what “is happening now.”

Next, watch three things: first, how the base effect evolves—this “automatic beautification” in the year-over-year numbers won’t last forever, and eventually the readings will become more reflective of reality; second, whether core services and the housing component can pull back—this determines whether the 0.3% is noise or a trend; third, don’t forget the Fed’s official target is PCE, not CPI—these two measures often send different signals, so don’t focus on just one.

One-sentence summary: This report isn’t “inflation is out of control,” but rather “the last-mile struggle of inflation is more grinding than people imagined.” Do you think core month over month can return to 0.2% before the end of the year?

#U.S. August Core CPI rose 0.3% month over month, exceeding expectations

BNB is up nearly 4% this week. The price briefly broke above $730, and its market cap also moved close to the $100 billion mark. However, this level still hasn’t been able to hold. Why is $100 billion so hard for $BNB ? Let’s look at the supply side first: BNB continues to buy back and burn, shrinking the circulating supply year by year. With a smaller supply, the same market cap requires a higher unit price to support it. Now the demand side: on-chain Gas, Launchpool, and ecosystem applications—each round of scenario expansion brings real consumption. Right now, market sentiment is fairly cautious. After the CPI is released, funds are waiting for signals from the Federal Reserve. Among major coins, there aren’t many that are strengthening against the trend, and BNB’s relative strength itself indicates that capital is continuously allocating to it. After it breaks above $730, the next test is turning the $100 billion market cap from a touch into a level it can defend. Do you hold BNB? At the $100 billion threshold, would you add more, or take profits? #BNBbreaksabove$730

The Dow fell more than 600 points in a single day; Wall Street then slid for three straight days. The market narrative that day was: oil prices broke above $100, U.S. Treasury yields hit multi-year highs, and fighting in the Middle East escalated. Then the next day, CPI came in line with expectations, oil prices pulled back by about 3%, and the Dow surged back up more than 600 points. With one move and then the reverse, the three things that most easily cause ordinary investors to lose money were basically staged for everyone to watch.

First, a big one-day bearish candle carries almost no information. The Dow is a price-weighted index of 30 component stocks. A 600-point move translates to roughly 1.1%, and in the volatile environment of 2026 that’s just routine swings. Changing positions based on a single day’s up or down is like paying a fee for noise.

Second, what’s truly worth tracking is the transmission chain: oil prices rise, which drives up energy inflation; the Fed is forced to hike rates (market odds for a rate hike at next Wednesday once reached as high as 90%); and finally, valuations get suppressed. Every node on this chain has data you can check: U.S. diesel prices have hit a record high of over $6 per gallon, and core CPI’s month-over-month reading of 0.3% came in above expectations. If the chain breaks, the sell-off is an opportunity; if the chain doesn’t break, the rebound is a window to “escape.”

Third, examine how sensitive your holdings are to oil prices. In this round, energy-themed ETFs set new 52-week highs, while high-valuation growth stocks were hit the hardest by rising rates. If your assets are positively correlated with oil, this move is a tailwind; if negatively correlated, you should have your strategy ready before the next 600-point drop—not make emotional trades that same day.

In one sentence: people who panic-sold that day had to buy back the next day at a higher price. The market never punishes waiting; it only punishes reactive behavior. #DowFallsOver600Points

Bitwise announced it will shut down its Dogecoin ETF (code: BWOW) on October 14, less than one year after its listing on November 26, 2025. At that time, remaining shareholders will receive cash based on the net asset value as of October 21, with the payment date being October 22. No action is required. The way this product is being wound up says a lot: as of September 8, the fund’s net assets were only about $722,000, holding roughly 8.2 million DOGE—an amount that even falls short of many retail investors’ own positions. When it was launched, Bitwise CEO also imagined that millions of DOGE holders would need ETF-style exposure, but in reality, demand never reached sufficient scale. This is more like a foregone conclusion of the ETF “crowding-out” era: once similar products flood the market, only the leading ones can retain liquidity, while long-tail products are destined to be merged or shut down. DOGE itself isn’t dead; what died is the business model of “raising big fish in a small pond.” Next, it’s likely that more mini crypto ETFs will quietly exit. $DOGE

Tonight’s CPI: rather than asking “how bad is the data,” it’s better to ask “whether the Fed dares to pull the trigger.” First, let’s set the facts straight: the U.S. August CPI year over year came in at 3.4%, unchanged from the prior reading and in line with expectations; month over month it rose 0.4%, the highest since June; what really “blows up” is core CPI, which rose 0.3% month over month—above the expected 0.2%—and marked the largest single-month increase since May. In terms of structure, gasoline was up 3.9% in the month, and energy overall rose 2.1%. The driver of this inflation wave is not on the services side, but in oil prices—Brent is already above $100.

After the data was released, CME’s “FedWatch” showed the probability of a 25-basis-point rate hike in September jumping from about 70% to nearly 90%. But one detail is worth noting: the 2-year Treasury yield has been moving higher, while the 10-year has eased back to around 4.95%. What the market is pricing is “another hike in the short term,” not “long-term rates running out of control”—the risk profiles of these two scenarios are completely different.

My take: this CPI is “bad in line with expectations,” not “worse than expected.” The year-over-year figure hasn’t deteriorated; the stickiness is concentrated in energy. And rate hikes can’t cure the supply-side issue of oil prices—this is more a political problem than a monetary one. What the Fed truly needs to weigh is this: with midterm elections approaching, not hiking could lead to questions about the Fed’s independence; but hiking also faces the dilemma that inflation driven by oil prices can’t be brought down at the root. Therefore, on September 16, the FOMC dot plot and the post-meeting language matter far more than this CPI release itself.

For Bitcoin $BTC , at the moment the data hit, it briefly poked into the $76,000 area, then snapped back in a V shape toward the $79,000 line—both longs and shorts got squeezed. The extent to which the negative news has already been priced in may be deeper than most people think. Are you betting on a hike, or holding steady? Let’s chat in the comments. #CPIDataIncomingCanItTriggerSeptemberRateHike

Same track, two different destinies. Over the past 52 weeks, $MRVLB has significantly outperformed $AVGOB , with the latter only slightly better than the S&P 500; but when you stretch the timeline to five years, Broadcom leaves both behind. Wall Street is still issuing strong buy ratings for both: Broadcom’s 42 analysts have an average score of 4.69, with a highest target price of $715; while M* w e ll has 36 analysts and a highest target price of $400. The disagreement lies in their business structure. The custom ASIC market is expected to grow at a 27% CAGR to reach $118 billion by 2033, and Broadcom holds 60–80% of the share. Customers include Google, Meta, and OpenAI. In the last quarter, AI semiconductor revenue was $16.7 billion, up 221% year over year. M* w e ll’s share is about 20–25%; data centers account for 79% of revenue, but 10 large customers contribute 82% of revenue. The author’s rationale is this: Broadcom has $8.7 billion in recurring software revenue from VMware as a backstop, while M* w e ll is a more pure play—and a more volatile one—on AI exposure. Would you choose the safer one or the faster one?

The CPI night’s script should have been bearish: the core month-over-month reading came in at 0.3% above expectations, the U.S. dollar index broke above 98, Treasury yields climbed, and spot gold dipped in the short term. According to the old logic of “Bitcoin follows liquidity,” under this combination, $BTC should have been playing catch-up and lagging. But instead, it rebounded from the lows and regained $79,000.

How should we understand this “anti-macro” move?

First explanation: bad news was priced in early. Before the data was released, $BTC had already pulled back from its high after a golden cross to around 77,000, and the short positions that needed to be built were already in place. Once the data was actually logged, what happened was “uncertainty liquidation.” This also explains why many seemingly terrible data prints were followed by gains—the market isn’t trading the data itself, but the gap between the data and expectations, as well as the position unwinding that comes when the uncertainty disappears.

Second explanation: capital is re-evaluating the “reaction function.” This CPI isn’t enough to shift policy back toward further rate hikes, so the “worst-case scenario” has been taken off the table, giving risk assets a window to breathe.

But a rebound is not the same as a reversal. $79,000 only recovers part of the lost ground. What really needs confirmation are three things: one, whether the 77,000 area can turn from a pressure zone into support—only a retest that doesn’t break would count as holding; two, whether the rebound has volume behind it—low-volume rebounds typically don’t go far in a range-bound market; and three, if the dollar and yields keep strengthening, the headwind will eventually press it back down.

My take: Right now, $BTC is in a state of “macro is unclear, but technicals are offering support.” The 77,000—79,000 range is the main battlefield for bulls and bears; whoever breaks the range first will gain control of the pricing power for the next phase.

Which side are you on?

#Bitcoin Rebound After CPI, Regains $79,000

Qualcomm and AWS have signed a multi-generation custom chip cooperation agreement, targeting AI inference and the CPU market in data centers. But the deal—interpreted by some as a “challenge to Nvidia”—contains quite a number of details: Qualcomm issued Amazon 25 million warrants. The exercise price is $161.26, or about $4 billion. Of these, only 3.75 million shares vest immediately; the rest are tied to actual performance—including a real chip procurement volume with an upper limit of $60 billion. In other words, $60 billion is not a promised revenue figure, but an execution cap. More importantly is the positioning: this collaboration deliberately avoids the GPU training market where Nvidia dominates, and instead focuses on the inference side. Qualcomm raised its target for non-mobile revenue in fiscal year 2029 from $22 billion to $40 billion, and set a data center sales target of $15 billion, though there is still no shipment validation at present. The trust vote from hyperscale cloud providers is real—but whether it can be converted into steady revenue depends on execution. The story of the second tier in AI chips is only just beginning. $QCOMB $AMZNB $NVDAB