链研社lianyanshe

Don't go against the cycles; this is the most important lesson I learned from the last cycle. There is no super cycle, and don't blindly trust the big players; several of them get sacrificed each round.

In the past few days, I reviewed the current bear market, which is basically following the same pattern as the last bear market.

Last cycle (May 2022): $30,000 was considered the iron bottom of the super cycle at that time because it was the starting point after the disaster on May 19, 2021, and also the support of the weekly 120. I particularly remember that when it broke below $30,000, the fear index was already below 10, and a lot of bottom-fishing funds believed it couldn't drop further or that it was time to rebound.

Current cycle (February 2026): $80,000, which used to be a strong support level, has now become a strong resistance level at the weekly MA120. $80,000 is also the cost price for Bitcoin mining companies. The current $76,000 feels like the brief struggle after breaking below $30,000 in the last cycle; the market may very well have one more action that thoroughly shatters confidence, such as touching the peak of the last cycle at $69,000.

If it falls below $70,000, it will trigger a larger-scale stop-loss and liquidation, which could potentially exceed the massive trading volume from November 2025. Without extreme panic, rebounds often just serve to entice buyers.

During the cycle, never let emotions lead you astray; in extreme trend markets, emotional indicators can be misleading. A fear index of 10 indicates that retail investors are already hopeless, but the main players may still be using this despair for one last deep squat. Trend lines are more direct than any indicator; as long as the price remains below MA120, all upward movements are merely rebounds rather than reversals.

In the past few days, Bitcoin has also been constantly hitting new lows, recalling the trend of the last cycle. The expectation for a rebound after breaking down with volume in place is 20%, back then from $26,700 to $32,399. Based on the current situation, it can only be inferred that only after breaking $70,000 will there be significant volume released, with rebound expectations reaching slightly above $80,000 at the weekly MA120, anticipating a 15% increase.

Additionally, if viewed from the cycle perspective, Bitcoin should bottom out in November to December. What could the price be then? $70,000? $60,000? Or even lower?

In the storage sector, those who fear heights are the unfortunate ones, and the ones who FOMO out are just clowns; only the ones chasing the peaks are truly happy.😭

I'm the clown

The storage sector on Binance has become the best launchpad $MU $SNDK $EWY

Be a time buddy in the crypto game to achieve financial freedom. From 2020 to 2026, while locked up, stocks skyrocketed 400 times. Can you guess which Korean stock he bought? If only we could trade Korean stocks on Binance, I'm seriously eyeing Hynix.

Four. The AI Sprint for Meme has wrapped up. A 30-day hackathon resulted in 196 BUIDLs, with only 5 projects making it to the final Demo Day, a selection rate of less than 3%.
From these 5 projects, you can catch some clues on how AI Agents are set to land on-chain. What’s the future trend for AI Agents?

These 5 projects conveniently cover five distinct directions in the AI Agent track:
4lpha AI is focused on intelligence + trading Agents.

Build4 has gone further, creating the infrastructure for an autonomous agency economy.

Clawdyland is all about the Agent arena, letting AI Agents compete for BNB rewards in poker and debate contests.

ClipX is merging the social layer with the trading layer.

elizaOK is following the value layer route based on elizaOS. It utilizes an AI treasury and a national treasury flywheel mechanism to allow Agents to complete an economic cycle from birth → profit → pay fees.

Looking at these five projects, a common thread emerges: they are all tackling the real-world challenges AI Agents face when executing on-chain.

What’s next? I think there are a few points worth keeping an eye on:
First, how far the developer ecosystem for the Agent Skill Framework can go. If 2-3 projects from this hackathon can evolve into continuously iterating products, the confidence in the entire ecosystem will shift dramatically.
Second, the user experience on the AI Assistant side. While the API openness caters to developers, whether end-users can really use natural language to create tokens and interact in the market will determine the minimum threshold.
Third, the fast track for the MVB accelerator. Quality projects can leap directly into the top acceleration system of the BNB Chain.

On May 13, Demo Day will reveal the final rankings. Regardless of who takes first place, what matters is that this path has been paved. AI Agents are not just for storytelling; they should execute, trade, provide decision support, and facilitate economic cycles. At the very least, they need to be serious about these tasks, and the rest will be validated by time.

$FORM

On May 11th at 8 PM (Monday), we’ll be live streaming on X Space and Binance Square. Let’s dive into the changes Sui has gone through over the past three years, those undervalued key developments, and what’s really worth anticipating in 2026.

🎙️Guests and Host
- Rachel, Head of APAC Markets for Sui
- Top content creator lianyanshe
- Sui supporter MindfrogCrypto
- Yammy_1222 (Host)

Highly recommended, the Binance Online Investment Summit on May 13 is a goldmine for gaining that information edge. You can catch insights from traditional finance giants managing trillions and crypto pioneers on how they view the current market. The best part? It's totally free and streamed online live in the plaza.

I had the chance to attend a closed-door session with friends once, and the information edge we gained was six months ahead of the market. Back then, Grayscale's GBTC was trading at a negative premium, and they were talking about ETFs. It was hard to believe during the bear market, and we missed the biggest opportunity.

No worries if you don’t understand English at this summit; the plaza has built-in subtitle translation, which is super advanced. I also recommend using Doubao to jot down key points from the meeting; it automatically summarizes and generates highlights, which is really handy.

For retail traders, this is a real chance to shorten the knowledge gap. How does BlackRock's COO view tokenization? What’s the Solana Foundation's roadmap? How does Adam Back see BTC evolving? Understanding the cutting-edge crypto narrative process is crucial, as the main themes for the next cycle are RWA, stablecoins, and how AI×Crypto will materialize.

Here are a few of the key guest speakers:
- Rob Goldstein, COO of BlackRock. By the end of Q1 2026, BlackRock’s AUM is projected to hit $13.89 trillion (based on company reports), making it the largest asset management firm globally. His talk will focus on tokenized financial markets.
- Brad Garlinghouse (Ripple CEO), Lily Liu (Solana Foundation Chair), CZ, Anthony Pompliano, Chamath Palihapitiya, He Yi.

TradFi, crypto natives, and public chain infrastructure all in one place.

详细信息

#木头姐与cz谈ai和稳定币

Saylor finally let slip selling BTC to pay dividends: but the narrative of never selling has started to lose its premium.

During the May 5th call for the Q1 2026 earnings report of Strategy (formerly MicroStrategy), Saylor dropped that line the market has been waiting five years to hear—"we might sell a small amount of BTC to pay dividends, just to give the market a shot and silence the shorts." MSTR dropped 4% that day, BTC briefly fell below $81,000, and the probability of selling BTC by the end of the year on Polymarket shot up from a low to 40%.

Strategy currently holds 818,334 BTC, roughly $66 billion. The preferred shares (STRC 11.5% floating, STRK 8%, STRF, STRD) have an annual dividend obligation of about $1.5 billion. Q1 software business revenue was $124.3 million, up 11.9% year-on-year—almost negligible in the face of the $1.5 billion dividend. Plus, Q1 recorded a net loss of $12.54 billion, making the structural cash gap crystal clear.

The market treated this as bearish news and sold off, but I think they missed the essence.

If they really do sell, 1% of their holdings would amount to 8,183 BTC, about $660 million, which would cover a year's worth of dividends and have almost no impact on spot liquidity. CME and spot trading volumes far exceed this level in a single day. Saylor himself framed the move as a liquidity proof + a counterattack against the shorts, hinting that while selling, a new round of fundraising would continue to buy more, keeping the company’s net position increasing.

The real turning point is in the narrative. The promise of never selling has been the core source of premium for Strategy over the past five years. The high premium of MSTR relative to NAV is essentially what the market paid for this commitment. Once Saylor starts to ease up, even if it's just a small, strategic easing, the foundation of that premium logic begins to shake.

Deeper still is the business model itself. Preferred shares + continuous issuance + the appreciation of BTC formed a self-reinforcing cycle in the past; but when software revenue only covers 8% of the dividend obligations and Q1 recorded a massive loss, the cycle has shifted from "I win twice when BTC rises" to "if BTC doesn’t rise, I must sell." This isn’t about being a Bitcoin Maxi; this is a realism forced by the obligations of preferred share dividends.

In the medium to long term, MSTR’s NAV premium will systematically converge. There’s no major issue for BTC itself, but MSTR holders need to recalibrate their expectations. Saylor has never been a religious leader; he’s a masterful capital operator. The market only just started to realize this.

NVIDIA is facing a dilemma: demand is through the roof, but the market has started to price in de-NVIDIA-ization.

On April 30, NVIDIA dropped 4.63% in a day, losing around $230 billion in market cap, breaking the $5 trillion mark. On the same day, the earnings report was still explosive: FY2026 Q4 revenue around $68.1 billion (up 73% year-over-year), FY2027 Q1 guidance around $78 billion, not even counting revenue from Chinese data centers. Blackwell's monthly output is nearing 1 million units, targeting ~20 million units shipped by 2026, with a GAAP gross margin still at 75.2%. It's like a legal money printer that’s been churning out cash for years and still has such high profits.

Demand is not the issue; the problem is what the market is seeing.

The Big Four US Clients are Voting with Their Feet

Four massive cloud players are collectively doing one thing: moving inference workloads off of NVIDIA GPUs. NVIDIA's moat in the training domain remains unshaken.

1. Google TPU v7: self-developed seventh generation, plus a four-way supply chain combo, almost fully self-sufficient in internal inference.
2. Meta MTIA: announced they will launch 4 generations of self-developed chips within two years, and in April signed a joint development agreement with Broadcom.
3. Amazon Trainium / Inferentia: Meta just inked a multi-million unit deal for Graviton with AWS this month.
4. Microsoft / OpenAI: self-developed ASIC roadmap is advancing in sync, with OpenAI also placing multiple bets with Broadcom and AMD.

According to data from TrendForce, by 2026, hyperscaler self-developed ASIC shipments are expected to grow by ~44.6%, while GPUs will only grow by ~16.1%; the custom chip market size is projected to reach ~$118 billion this year. Morgan Stanley's share structure is also changing: NVIDIA ~85%, ASIC slightly over 10%, AMD slightly below 5%—10% doesn’t seem like much, but this number was single digits last year.

My three judgments are:

1. NVIDIA won’t be able to eat off the training market and the Rubin roadmap for another two or three years; the trillion-dollar orders are real.
2. However, the end of the Davis double-click is likely here; future profits will come from EPS growth rather than valuation expansion.
3. The real alpha is in the players positioned to benefit: Broadcom (doing ASIC for Meta, Google, OpenAI), Marvell, TSMC advanced packaging capacity, and HBM—these are the stocks that will genuinely benefit from the de-NVIDIA-ization narrative $NVDA

BTC breaks above 76,300: Consolidation is a good thing, the real variables are outside

As we enter May, BTC is priced around ~76,300 USD, recovering about ~25% from the early April low of just over 60,000. More importantly, this recovery isn’t a massive bullish spike driven by emotional trading; it’s a steady climb that’s consistently holding above the 50-day moving average. I prefer this kind of price action for a simple reason: support built at the bottom is much more reliable than highs that are spiked up.

Many are eyeing the 80,000 USD mark, hoping for a direct breakout. However, from a rhythm perspective, it’s actually bullish for BTC to hang around the 70,000—75,000 range for a while. There are two reasons for this:

1. Chip turnover: The panic selling from April needs time to be absorbed; a rapid surge would only pull weak hands back in.
2. Pricing based on expectation gaps: The market isn’t uniformly bullish right now; discrepancies are the fuel for future price action; safe assets usually rise slowly from these gaps.

The real variables are not on-chain but out in the broader market. Current macro signals are mixed: tech stocks are hitting new highs while other sectors are lagging; oil prices are rapidly climbing, and the dollar is strengthening in tandem. This is a classic case of localized risk appetite; money is flowing, but only into a few select tracks. Ahead of important U.S. economic data releases, crypto prices appear calm on the surface, but the options structure on Deribit is already telling you the story: traders aren’t aggressively chasing higher prices but are rather increasing their downside protection positions. This is a cautious signal, not necessarily bearish.

So my judgment is:
- In the short term, without strong catalysts, BTC will likely continue to oscillate within the 70,000—78,000 range.
- Moderate data → Market maintains current structure, slowly grinding upwards.
- If inflation or oil prices continue to heat up → There’s a high probability of first dipping to around 70,000 to find support, and then we’ll see if funds step in.

In many chain games, user behavior primarily stays at the usage level. You complete tasks, participate in events, and earn rewards; these actions affect the experience but rarely directly alter the allocation of funds.

@Pixels takes this relationship a step further. The system records user actions across different games and adjusts reward distribution and budget flow based on these behaviors. The white paper mentions that data such as retention, payment, and participation paths will be used to continuously optimize the incentive model.

This brings a change: behaviors begin to be directly linked to the funding structure. Actions deemed more valuable by the system are more likely to receive incentives; those with lower conversion rates will see their weight reduced.

From the results, users are not just participants; they indirectly influence resource allocation. A game that gathers more high-quality behaviors may see an increase in the budget it can access; conversely, resources will gradually decrease.

This structure is akin to converting behaviors into a measurable signal and using that signal to allocate funds. Behaviors are no longer just results but begin to participate in the pricing process.

However, this mechanism also relies on prerequisites. The data needs to be sufficiently accurate, and the model must differentiate between short-term actions and long-term value. If there is a misjudgment, funds could be directed towards inefficient paths.

So, from this perspective, it is indeed attempting to incorporate user behavior into a more direct funding allocation system, but how effective it will be depends on the data and the model's performance in real-world operations. #pixel $PIXEL #广场征文