每周编辑精选 Weekly Editor's Picks（0328-0403）

The information flow is too fast—deep analysis articles can easily get drowned out by trending topics. The “Weekly Editor’s Picks” section pulls this decision-worthy content out of a massive sea of information, helps you filter out noise, keeps the insights, and sparks inspiration.

Macroeconomic Conditions

A global economic recession—has it quietly already begun?

The author redefines recession, not as an “economic outcome,” but as a “strategic state.” It doesn’t just compress growth and employment—it also weakens a country’s bargaining power, ability to attract capital, and external credibility, causing it to lose initiative in global games. That’s also why governments in various countries are replacing monetary tools with fiscal, diplomatic, and even geopolitical measures—essentially buying time for growth to slow and avoiding being forced into negotiations during a recession.

Within this framework, what matters to the market is no longer the interest-rate path itself, but rather: who can break free of constraints, and who is still trapped within them.

The difference first shows up in FX and interest-rate markets, and then transmits to asset prices and capital flows. When valuations keep rising even amid slower growth, the cause may not be an improvement in fundamentals, but rather policy expectations that a recession won’t be allowed to happen.

After Buffett steps down, his first visit: now is not the time to bottom-fish; nuclear weapons will be used sooner or later

Apple sold early—but now it also doesn’t buy.

Don’t bottom-fish in the U.S. stock market at this time.

The most dangerous scenario is when the person holding the nuclear button is about to die, or is facing massive humiliation. In that kind of situation, no one can predict what someone will decide. In the next 100 years, maybe 200 years, nuclear weapons will be used.

$700B poured into AI—Americans already tasted the bitter lesson of inflation first

The more you invest in AI, the higher inflation goes, the farther rate cuts recede, and the higher the cost of financing—but investment is still accelerating.

The defense-contractor arms race among large companies can’t be stopped. One data center can top the electricity usage of an entire state.

$700B is flowing into AI infrastructure. Is this money the cause of inflation, or the prelude to a productivity revolution? It depends on a question nobody can answer yet: will the models running in these data centers make the economy more efficient?

Interview with Pantera’s founder: Bitcoin has reached escape velocity—traditional assets are being left behind

It’s not gold setting new highs—it’s fiat money hitting historic lows.

The average age of first-time homebuyers in the U.S. has been pushed back from 28 to 40.

We’re facing a generational turning point where money separates from the nation. Stablecoins are very likely to take half of bank deposits within a decade. Bitcoin has reached escape velocity.

Investing & Entrepreneurship

Interviewing 10 Dubai office workers: some are required to sign a “life-or-death statement”

On the ground, Dubai is already on the battlefield.

Some Web3 workers are fleeing, others are holding their ground, and some sign a life-or-death statement (company liability waiver) and return to Dubai—hoping to win high career returns under high risk.

ve token model “reflux”: why the three major protocols voluntarily abandoned their former ace?

The reason isn’t theoretical error, but execution-level failure: low participation rate, governance captured, emission flows going to unprofitable pools, and token prices falling hard even as usage grows.

And yet Curve’s veCRV and Aerodrome’s ve(3,3) are still developing healthily. But this model only works where the emissions it incentivizes can create real economic demand for liquidity.

Meanwhile, other protocols are choosing alternatives for the ve token model: revenue-supported buybacks, deflationary supply mechanisms, or liquidity-governance tokens.

Tiger Research: analysis of the current state of retail investors in Asia’s nine largest markets

Tiger Research covers nine Asian markets with the largest user volumes, analyzing barriers to entry for retail participants and exchange response strategies. The core findings: there are structural differences in access barriers across markets such as South Korea, Japan, and Vietnam, and the exchanges’ localization strategy is the key variable to whether it can be executed successfully. This is valuable as a reference for teams looking to go to sea from Southeast Asia.

Also recommended: 《Odaily exclusive interview with SharpLink: “productive capitalists” of Ethereum》.

Web3 & AI

What exactly are AI Agents doing? Full解析 of the 500k-line Claude Code leak

The competitive moat for AI products may not be in the model layer—it’s in the engineering layer.

For AI products, the model inference cost may not be the most expensive layer; failure in cache management is.

Followed AI tools for a year—zero output: a reflection from a serial entrepreneur

Don’t treat “trying new tools” as “you building things.” When everyone can use the same model, the only moat is taste and depth—and taste can only be won through real outcomes and sustained focus.

Tiger Research: what AI services do crypto companies provide?

Different sub-industries have different adoption motivations: exchanges aim to prevent user outflows; security companies aim to fill audit blind spots; payment infrastructure targets the emerging agent economy.

In the AI field, “FOMO” and competitive pressure are accelerating adoption—far beyond actual needs. Both real demand and competitive anxiety are playing a role.

Distinguishing adoption that creates value from adoption that merely slaps on a label is the key question.

Prediction Markets

A 1-dollar return rate of only 43%—why are 87% of Polymarket players losing money?

“Contrarian preferences” are one of the most expensive mistakes in prediction markets. Traders often systematically overestimate low-probability events and pay too high a price for contracts that look cheap.

Traders who truly outperform prediction markets over the long run aren’t necessarily the ones who are “most accurate” in judgment; they’re the ones who can adjust their views the fastest and most reasonably when new evidence appears.

Bayesian methods essentially provide a measure of that “adjustment speed.”

Don’t bet on a hunch: AI is “picking up money” on Polymarket

The article introduces a method to identify arbitrage opportunities on Polymarket and systematize executing them: use Perplexity to conduct research and identify deviations between data and market pricing; use Claude to build trading logic, control risk, and execute automatically; and ultimately complete trading and monetization on Polymarket.

For ordinary participants, a more realistic path is to first find determinism through research, and then amplify returns with a system.

Also recommended: 《Polymarket smart money panorama: 26 long-term trackable addresses (broken down by track)》.

CeFi & DeFi

Stripe up, PayPal down: a new king of payments is crowned

Stripe’s view is straightforward: once AI Agents start making purchase decisions for humans, whoever controls the payment rails will be the first to seize the core lifeline of the AI economy. Stripe’s forward-looking perspective has put it ahead of the entire payments industry.

PayPal, meanwhile, is stumbling. PayPal’s business model lives on “fees from cash flow,” while the stablecoin business model depends on “parked assets earning Treasury interest.” There’s a natural conflict between these two logics. Every time PayPal promotes a PYUSD stablecoin payment, to some extent it’s actually cannibalizing its own traditional fee revenue. In PayPal’s current business framework, this problem is hard to solve.

BTC is on its last breath—why has HYPE surged 20% against the trend?

A new source of value support is provided by Hyperliquild’s RWA trading market, empowered by HIP-3.

A deep, ten-thousand-word breakdown of Hyperliquid HIP-4: borrowing prediction markets and options trading to swallow traditional finance

The market overall is in a downward trend, yet HYPE has shown very strong stability. The important reason is undoubtedly Hyperliquid’s solid fundamentals, its focus on creating revenue, and the continued use of profits to buy back HYPE.

HIP-3 (a perpetual contract market deployed by builders) has already shown a clear rule: when infrastructure is permissionless and has been validated by the market, liquidity tends to cluster around stronger teams—regardless of whether they receive extra ecosystem support.

The same logic will apply to HIP-4. HIP-4 focuses on “outcome trading.” It brings prediction markets and certain specific types of options into Hyperliquid—these products can provide nonlinear outcome payoffs, while also having no liquidation risk.

In the past, no protocol was good enough to truly bring sustainable options trading into the crypto world. Hyperliquid has done it, and the way it operates is different from every other platform. This platform has no investors, isn’t affected by any external pressure, and Jeff can freely decide what the company wants to do. In this sense, Hyperliquid is a lot like Telegram: you don’t need to spend too much money on marketing—the key is belief itself. If the product is good enough, people will come to use it eventually.

Hyperliquid has the opportunity to take over the entire crypto options market.

The winners will build real-world product experiences—embedding crypto as the underlying implementation detail. The losers will keep holding onto the old narrative of “crypto for crypto’s sake,” and expect the whole world to adapt to them instead.

Airdrop Opportunities and Interaction Guides

Hot interaction collection | Abstract new badge quest; Noise Beta version is live (April 2)

Meme

Making A7 per month as a post-00s kid—trapped in three screens

Odaily interviewed a Meme P newcomer who can still get big results in 2026, and a former Meme diamond-hand, to understand their path to making money and their work-and-life approach.

Bitcoin

Lose $19k for every coin mined—Bitcoin miners collectively defect to AI

CoinShares’ latest mining report shows that the weighted average cost for listed mining companies to mine one Bitcoin has risen to roughly $80k, while BTC’s current price is $68,000–$70k—losing $19k for every coin mined.

The Bitcoin mining industry is undergoing the most fundamental transformation since its founding. The clearest signal isn’t compute power or difficulty adjustments—it’s the balance sheet, and the way out is a full pivot to AI infrastructure.

When Bitcoin mining rigs fly into space

For the Bitcoin mining companies currently operating, space mining doesn’t yet pose a practical competitive threat in the short term—but many startups are still continuously trying, which also reflects the large cost-reduction space it represents behind the scenes, and the industry’s continued appeal and imagination regarding it. This also indirectly shows that the whole industry is facing structural cost pressure.

The logic of space mining is the ultimate extrapolation of the trend above: if cheap electricity on the ground is ultimately squeezed by competition for demand, then go where energy is most abundant—i.e., the universe.

Security

Google’s major breakthrough in quantum computing—crypto world may need to “change locks” early

Led by top quantum circuit experts including Craig Gidney, Google’s quantum research team (Google Quantum AI) has rolled out two actions in succession.

First, on March 25, Google officially proposed its post-quantum cryptography (PQC) migration timeline with 2029 as the target year. Second, on March 31, Google Quantum AI specifically released a research report for the crypto industry, stating that based on the latest research results, the resources required for future quantum computers to break the elliptic curve cryptography that protects cryptocurrencies will be far less than previously thought.

The industry’s past assumptions about timelines need to be revised. Google has set its migration target for 2029, and in its article, Google Quantum AI also mentioned that it is working with institutions such as Coinbase, the Stanford Blockchain Research Institute, and the Ethereum Foundation to responsibly advance according to the 2029 schedule.

For every crypto project, this means a brand-new security checkpoint. Whichever can recognize the problem earlier, push through upgrades, and complete the “change locks,” will have a better chance to keep its security boundary intact in the next era.

Also recommended: 《Odaily exclusive interview with Yu Xian: Anthropic’s nuclear-bomb-level new model leak—how will it impact crypto security offense and defense?》.

A week of hot topics to catch up on

Policy & Macro Markets

U.S. lawmakers’ offices plan to release next week a draft of stablecoin yield terms, and the industry is preparing counter-proposals;

CFTC clarifies the top 5 enforcement priorities: “cracking down” on prediction market insider trading and market manipulation;

Opinions & Statements

Trump: negotiations have made major progress; Trump threatens to strike Iran’s energy facilities, oil prices keep rising, and U.S. stock index futures fall; Iran’s president: we are prepared to end the war, but hope for assurances; Iran will charge tolls to ships transiting the Strait of Hormuz;

Trump: Bitcoin has an important position, and the U.S. needs to stay ahead;

10x: under the conflict between the U.S. and Iran, Bitcoin’s “safe-haven myth” fails, and ETF funds reshape the pricing logic;

Trader Eugene: cut losses and exited, plans to reduce trading frequency;

Institutions, Big Companies & Top Projects

SpaceX is expected to list in June; IPO target valuation above $2 trillion; plans to raise up to $75 billion; OpenAI completes $122 billion in financing, reaching a valuation of $852 billion;

Bitfarms plans to offload all Bitcoins on its balance sheet and fully pivot to AI infrastructure;

Aster adjusts its tokenomics structure and sharply reduces the monthly unlock amount;

edgeX airdrop Waterloo;

MagicEden: the wallet will be taken down tomorrow and enter only-export mode;

Dmail Network will gradually shut down starting May 15; users need to export their emails before then;

Security

Drift Protocol was attacked, with losses of at least $200 million (details).

Attached: the “Weekly Editor’s Picks” series transmission links. See you next time~