🤔 On Anthropic’s first day as a public company, how many shares did its founders and employees sell?
Dario sold about 3.38 million shares in the $ANT IPO, cashing out for roughly $280 million at the offering price. The seven co-founders cashed out for a combined total of about $2 billion.
Employees were also able to unlock 10%, giving them a total of 315 million shares they could sell—one-third of the total IPO offering.
On the one hand, they tell the market about AI’s long-term future and say they’ll cure all cancers; on the other, the founders cash out enormous wealth when the company goes public 😂
👉 Some of the betrayals before the IPO:
1️⃣ Microsoft invested $5 billion, only to help create its own competitor to Office
Anthropic launched Claude Sheets and Whiteboards, upgraded Claude Data, and began competing for Microsoft’s foothold in office software.
Now spreadsheets, whiteboards, and data analysis can all be done within Claude, and competition between the two at the application layer has clearly intensified.
2️⃣ Figma partnered with Claude to connect workflows, then Claude went and built its own design tool
On February 17, Figma announced an integration with Claude Code, allowing generated interfaces to be brought into Figma for editing.
On April 17, Anthropic launched Claude Design, covering use cases such as prototypes, design mockups, and presentations.
Jensen Huang’s departure from Nvidia caused its share price to fall $NVDA ; the most direct beneficiary was also “the largest IPO in history”—Anthropic.
Dario’s political maneuvering may also have played a part.
Talk about vision, and it’s all about the future of humanity. Talk about safety, and they act as though the fate of civilization rests on their shoulders.
But when it comes to business decisions, they poach business from partners, change users’ payment rules, and sell founders’ shares at the IPO—they haven’t missed a trick.
Although Anthropic’s large language models are genuinely useful, it’s hard to believe it’s a “good” company 😂
🤔 I was wondering why content mining rewards on Binance Square have been increasing lately.
I asked the Square security guard and found out W2E had just revamped its reward system. Here are the details:
👉 The main addition in this revamp is an “Engagement Reward”:
➤ Readers don’t need to trade: genuine comments, likes, and follows can also earn engagement rewards for content
➤ Rewards are allocated based on quality and quantity: different types of engagement carry different weights, and active readers’ interactions count for more
➤ Engagement farming doesn’t count: like-for-like groups, alt accounts, linked accounts, interactions without genuine reading, and similar activity are excluded. Repeated interactions are deduplicated
👉 Livestreams retain ranking rewards and add a tiered commission rebate mechanism
Image-and-text posts, short posts, long-form articles, and videos: ranking rewards are removed and engagement rewards are added
Livestreams: ranked weekly by base livestream commission, with tiered rebates— the top 30 receive a total rebate of 50%, ranks 31–100 receive 30%, and all others receive 20%
Extend more incentives to before a trade takes place: let great content attract people, genuine conversations keep them around, and trust naturally drive what comes next.
I think this might be the content flywheel Square is looking to optimize, so that more diverse, high-quality content can flourish on Binance Square!
🤔 When AI agents start managing money for people, who checks the data they use to make decisions?
Recently, while researching @SpaceandTime , Bump has been paying close attention to the common role it plays in these scenarios.
Space and Time is a data layer backed by Microsoft’s venture capital arm, M12. Its core focus is clear: providing verifiable data for AI, tokenization, and DeFi.
Its Proof of SQL can attach a cryptographic proof to supported query computations. When an application receives a number, it can also verify that the number was correctly calculated from committed data using the specified query.
It’s easy to understand how this works in real-world applications.
1️⃣ AI research needs verifiable evidence
Suppose an AI agent needs to evaluate a protocol’s usage. It can query fees over a specified period and cite an aggregated result with an attached proof. Applications using this analysis can also verify the calculations.
When AI analysis goes on to inform financial management, this capability has a practical purpose.
2️⃣ Assets are tokenized, and the books still need to be balanced
When a fund tokenizes its shares, it still needs to account for holdings, cash flows, and expenses each period to produce a net asset value / NAV report.
Space and Time supports connecting custom off-chain data and provides proofs for supported computations in reports. This gives holders and partners another way to check the calculations.
When off-chain assets are involved, it’s also important to examine how source records are collected and verified. The scope of computational proofs covers supported computations performed on committed data.
3️⃣ DeFi can enforce rules based on verifiable historical records
For example, if a protocol wants to offer fee discounts to long-term users or determine whether a wallet qualifies for a reward, it needs to look up past activity.
With proven query results, contracts can verify eligibility and then act according to predefined rules. Developers can use these historical records to design more tailored user incentives.
$SXT is Space and Time’s governance token. It is used to pay for workloads such as data writes and queries, and staking helps secure the network’s validators.
I’m looking forward to seeing Space and Time bring this kind of demand into real financial workflows, so that the data calculations cited by AI, fund NAV reports, and protocol eligibility assessments all have verifiable evidence behind them. #sxt
😆 The Bangkok BBW guerilla promo video is here! This time, Pumpump is transforming into a Pokémon Trainer and exploring Bangkok by sea, land, and air~
Word is that @Four.meme-華語搬运号 and BNB Chain are taking over Bangkok by sea, land, and air, with Fora as our local guide.
➤ A yacht cruise on the Chao Phraya River in collaboration with Four 🛥️ ➤ Free tuk-tuk rides within a 5-kilometer radius of the BBW venue 🛺 ➤ And a drone light show at night to light up Bangkok 🛸
My favorite TCG project, Renaiss, will also be an ecosystem partner and sponsor of the event.
The mysterious Pokémon Professor Winchman slipped me a few Pokémon cards
and said they could help Pumpump unlock the ability to get around by “sea, land, and air.” I’m curious to see how that works 🤔
Binance Blockchain Week kicks off in Bangkok at the end of the month! It’s the crypto event I look forward to most every year. Hope I get to see you all there~ #FourTakesBangkok #BNBChain #BinanceBlockchainWeek $FORM
👀 List of Top Global Companies' Founders Who Abruptly Resigned
Jensen Huang spent 33 years leading a company to the center of an era; the board's decision to force him to resign was a huge mistake.
Such power shifts have occurred many times in large companies, from Jobs and Musk to Altman.
Elon was removed from his position by the board of PayPal, the company he founded, in 2000.
Steve Jobs resigned after being stripped of his leadership by Apple in 1985, returning in 1997.
CZ (nickname @CZ ) voluntarily resigned as CEO in 2023 before going to the US to deal with an anti-money laundering case.
Sam Altman was briefly removed from his CEO position by the board.
Jensen is only in his 60s, the prime age for a leader; he has extensive experience, is a top figure in the semiconductor industry, and still has energy and passion.
Remember last year's BBW (Baidu, Apple, and Microsoft) top female CEO? @Yi He said: The company has grown into a towering tree, and we hope its ecosystem will become an "Amazon rainforest."
Today, the semiconductor and AI computing infrastructure, along with countless developers, research teams, and startups, have thrived in this environment, all thanks to Jensen's contributions.
Great companies can have lifecycles spanning centuries or even longer; the company itself is a continuation of the founders' will.
Hopefully, one day we'll see Jensen again, wearing that familiar leather jacket, returning to Nvidia.
🤩 Wow, I just saw Neosoul @NeoSoulAI get listed on the Korean exchange Bithumb! They’ll be expanding aggressively into the Korean market next.
It’s clear that both the Neosoul team and the market maker are highly capable.
It’s been about a month since TGE, and the price of $DNA ’s token is still holding steady. It rose a bit after the listing news, and the market hasn’t pulled back much.
Development of NeoTrade and NeoConnect is also progressing steadily. I’ll continue to follow the growth of the NeoSoul AI Agent ecosystem.
At Bangkok BBW, you can ask PumpPump for the new Neosoul collaboration merch. See you in Thailand at the end of the month~~
😂 The 63-year-old AI godfather can’t escape a 49-year-old blonde woman either
Huang Le has one hell of a résumé: from CCTV reporter to banking, securities and private equity, then to a Singapore-based AI computing company—spanning multiple worlds:
Hunan TV, CCTV reporter → HSBC Private Bank → CITIC Securities → Shanghai private equity → Singapore AI computing company
Hexi Investment invested in 7Road, while Megaspeed was connected to 7Road’s overseas business
In March 2023, Huang Le became Megaspeed’s managing director; that November, she resigned as a director
After years of moving from finance into the GPU business, the two had finally built the connections to enter the same business circles
1️⃣ September 20, 2016: The “Hequ Shengcai Steady Growth Fund” was established. Fund databases listed Huang Le and Kang Wei as investment managers
2️⃣ March 2023: 7Road International, originally registered in Singapore in 2015, changed its name to Megaspeed
3️⃣ During Computex in June 2024: Huang Le and Jensen Huang met with other tech-industry figures at a Taipei bar, leaving behind publicly available photos
4️⃣ May 22, 2025: The two were photographed together again at a business dinner at Sanliu Restaurant
5️⃣ October 9, 2025: The New York Times published an investigation into Megaspeed’s chip procurement and where the chips went
6️⃣ October 29, 2026: US media publicly reported allegations that Huang Le was involved in intelligence activities; Nvidia said the interactions were normal business dealings
7️⃣ November 4–7, 2026: Congress asked Jensen Huang to attend a closed-door briefing
8️⃣ November 8, 2026: Jensen Huang admitted to being unfaithful to his wife, and the board launched an independent investigation
As AI gets swept into a struggle over national destiny, it’s not just computing power that may become a battleground—the human heart could be one too
A tailor-made “master plan”: even the AI godfather who understands computing power can’t calculate the human heart
The three great delusions of business karaoke: She really loves me; I’m irresistible; she’s different
Jensen Huang: She really is different—she knows graphics cards 😂
🔶 Big Sis @Yi He joined the 「Binance Learning Camp」 dinner remotely, speaking without notes the whole time, but sharing plenty of insights
Beng Beng pulled together a few key takeaways to share with everyone:
1️⃣ Hacking incident: Security is an ever-escalating battle between attackers and defenders
It’s said that on the day of the theft, both Cousin @CZ and Big Sis stayed unusually calm and just wanted to resolve the incident as quickly as possible.
She admitted that Binance had indeed fallen short on security this time, but the losses were still within what the platform could absorb.
2️⃣ Bitcoin and quantum attacks: Technical challenges are always there
He Yi believes quantum computing could indeed pose a threat to Bitcoin, and Binance will also invest resources in related research.
3️⃣ Issuing tokens is like issuing debt: Financial freedom may actually weaken the drive to build
He Yi mentioned that she had believed as early as 2017 that:
Issuing tokens is essentially similar to issuing debt: it means raising funds on the basis of an individual’s or project’s reputation, and also entails long-term responsibility.
4️⃣ Why did token listings used to send prices up, but now they can easily mark the top?
5️⃣ Crypto’s shortage of assets: Profitable projects may not issue tokens, and projects that issue tokens may not be profitable
He Yi mentioned that the industry is not short of businesses with genuine revenue.
For example, some wallets, trading tools, DEXs, and others already have solid profitability, yet may not issue tokens.
6️⃣ Binance’s future: More than just a Crypto exchange
He Yi believes user needs are becoming more segmented.
Some focus on Meme, some like swing trading, some invest for the long term, and others care more about returns from wealth management.
7️⃣ AI products: Start with a demo and keep iterating
Her ideal AI financial assistant would directly understand users’ intentions, develop strategies, and execute trades.
8️⃣ User trust and long-term operations matter more than short-term buzz
🔶 At the learning camp, we heard Anco share how Binance Wallet is becoming a super hub for the on-chain ecosystem
1️⃣ A brand-new Meme social experience:
We’re continually improving the experience for Meme traders on the App and Web, making trading smoother and more intuitive.
Here’s a preview: a brand-new Meme trading mode is coming to mobile soon. Stay tuned!
2️⃣ Embracing AI, with rapid growth in Agent Wallets:
We hear the number of Binance Agent ID redemptions has now surpassed 170,000+.
This massive traffic has driven developers to create many spin-off experiences and generated wealth opportunities. More ideas are taking shape in the community.
3️⃣ Diverse prediction market experiences:
Through deep integration with predict fun @Predictdotfun , trading volume in the 5-minute price movement market has already surpassed Polymarket’s.
Since integrating with the wallet, predict has reached a cumulative trading volume of $946 million, with 446,700+ cumulative trading users.
User needs in prediction markets are becoming more diverse. Beyond politics, macroeconomics, and sports, short-term crypto price predictions are also developing into their own trading category.
4️⃣ Opening up capabilities to empower developers and unlock more productivity:
We’re bringing the vision of Agent ID and Agentic Wallet to life, giving AI Agents their own independent MPC wallets.
They can perform on-chain operations while adhering to safeguards such as segregated funds, independent daily limits, and authorization controls.
The wallet has also opened up many SKILL and MCP capabilities to empower the developer community, including on-chain smart money signals, token audits, and Alpha discovery.
In the age of AI, software is no longer the barrier. What’s scarce is the spillover of “high-quality data” and “open capabilities.”
----
In the past, wallets mainly answered the question: “Where should I store my assets?”
Now, wallets are looking to answer a bigger question: “What can I do with my assets here?”
And in the age of AI, you can choose which on-chain operations to let an Agent handle for you in Binance Wallet @Binance Wallet .
Looking forward to more new features!
👉 Available exclusively in Binance Wallet. If you don’t have a Binance Wallet yet, please:
How North Korean hacker groups divide their work, and a look at the methods behind their past “major cases”
North Korea may be the only country to have turned the hacking industry into a pillar of its national economy
Groups work together, targeting everything from banks and other financial institutions to software supply chains
The scale of their workforce, the breadth of their infiltration, and the extent of the damage they cause are likely all unmatched
👉 Yesterday, Binance published its investigation report on the hack
The methods show similarities to past attacks by North Korean hackers
1️⃣ Lazarus Group: broad-spectrum cyberattacks
Covers intelligence theft, financial theft, and destructive attacks, and is the main actor behind multiple related operations
Notable cases: Sony Pictures (2014), WannaCry (2017), the 3CX supply-chain attack (2023), and the “Dream Job” recruitment-bait campaign
2️⃣ TraderTraitor: focused on cryptocurrency theft
Primarily targets exchanges, wallet service providers, and employees with access to critical privileges
Often approaches targets through fake job offers and technical tests, then uses stolen identities and credentials to gain access to wallets, transactions, or signing processes
3️⃣ Bluenoroff / APT38: initially targeted banks via SWIFT and ATMs, then shifted to cryptocurrency
Primarily targets banks, financial institutions, payment and transfer systems, and also cryptocurrency assets. A typical method is to infiltrate an institution, abuse transfer privileges, manipulate fund transfers, and potentially disrupt systems to cover its tracks
Notable operations: Bangladesh Bank (2016), FASTCash ATM cash-outs, and macOS malware targeting venture capital firms and project teams
4️⃣ AppleJeus / UNC4736: posing as trading software and fake investment firms to infiltrate DeFi teams over the long term
Poses as trading software and fake investment firms to infiltrate DeFi teams over the long term
Notable operations: Radiant Capital (2024), Drift (2026)
4️⃣ Andariel / APT45: stealing military and industrial technology, and raising funds through ransomware
Primarily targets defense, aerospace, nuclear energy, and engineering organizations to steal sensitive technical information; also attacks medical institutions for ransom
Notable operations: Maui ransomware, Upbit cryptocurrency theft (in collaboration with Lazarus)
5️⃣ Kimsuky / APT43: intelligence gathering focused on think tanks, diplomats, and nuclear policy experts
Primarily targets government officials, think tanks, academics, research institutions, and media organizations, stealing emails and research materials through spear-phishing, impersonation, and other methods
🔶 Back on Phu Quoc after just half a month, and I’ve checked into what may be the world’s most beautiful Marriott
As soon as I walked into the room, I spotted practical goodies from Binance that everyone can use: a handheld fan, mosquito repellent, sunscreen, a waterproof phone pouch ... So thoughtful!
As a sponsor, TrustWallet prepared nasal inhalers, cooling balm, and even hangover medicine for everyone.
It started pouring as soon as I landed on Phu Quoc last night, providing everyone with a huge supply of natural liquidity 😆
I hope this past October was the market bottom, and that everyone will see liquidity go through the roof from here on out 🙏
Today’s challenge: spend all day using Binance Pay!
🔶 U.S. earnings season is coming! Get earnings previews and results analysis on Binance Square!
There’s also the #币安神算子 event, featuring prediction markets where you can bet on earnings expectations, powered by @Predictdotfun .
👉 PumpPump has put together a simple analysis of these earnings prediction markets:
1️⃣ Microsoft $MSFTB / Earnings date: 10.28
Last quarter, EPS of $4.74 significantly beat expectations of $4.24, while revenue of $90.01 billion also exceeded expectations of $87.62 billion.
Demand in the Azure market is very strong, and revenue growth is expected to accelerate again in 2027, making a beat highly likely.
2️⃣ Google $GOOGLB / Earnings date: 10.28
Most of Q2’s gains came from SpaceX and Anthropic.
Excluding investment gains, IBD estimates core Q2 EPS at around $2.85.
The Q3 consensus is around $3.04. Strong Cloud performance supports earnings, but gains from equity investments could be a drag.
3️⃣ Visa / Earnings date: 10.27
The estimate is around $3.44, with a slight beat expected.
Cross-border payments remain resilient, but customer incentives, expenses, and tax rates could limit the scope for a beat.
4️⃣ Amazon $AMZNB / Earnings date: 10.29
AWS grew by around 37% last quarter, supporting continued profit growth. I’m confident it will beat expectations.
5️⃣ Eli Lilly LLY / Earnings date: 10.29
The company has forecast around $4.2 billion in third-quarter expenses related to acquisitions of drug candidates in development. Tirzepatide has also been reported to carry a very slight risk of causing blindness.
Weight-loss drugs are selling well, but that may not be enough to offset the risk of results falling short of expectations.
This is just a brief analysis based on my own understanding. Feel free to copy my homework if you like.
But on Binance Square, many experts from institutions and highly skilled traders will share more professional analysis with everyone.
😆 Traditional brokerages are embracing crypto—even Futu has added an on-chain gateway for BNB?
Futu has officially enabled deposits and withdrawals for $BNB . Professional investors in Hong Kong can now also deposit funds into their Futu crypto accounts via BEP20.
👉 The business case behind traditional brokerages embracing crypto is very real:
The more of users’ asset allocation needs a platform can meet, the better its chances of retaining those users and their trading activity.
➤ For Futu, offering BNB trading and deposits and withdrawals adds another way to serve BNB holders and reach on-chain users.
➤ For BNB, this is also an expansion of its reach.
Brokerages already have customer relationships, trading interfaces, and service systems in place. By integrating with one, BNB gains another gateway to traditional finance users.
🔶 Binance Wallet x USDai Finance Flash Campaign: 17% combined APR, just a few days left to share the $CHIP prize pool
This round of the campaign ends in less than a week (September 29, 2026).
The source of USDai's annualized yield is somewhat unique: sUSDai's yield primarily comes from GPU financing spreads, not funding rates.
USDai is a stablecoin protocol built around financing AI infrastructure, and is pegged 1:1 to the US dollar.
sUSDai is the yield-bearing receipt you get when you stake USDai. Its yield is reflected in an increase in sUSDai's NAV, without relying on token issuance.
👉 What makes up the 17% combined APR:
1️⃣ Base yield: approximately 7%, from GPU infrastructure loans and stablecoin reserves.
If the base yield falls below 7%, the Yield Reserve makes up the difference, providing a floor.
2️⃣ Binance Wallet Boost: up to approximately 10%, with a $300K CHIP prize pool distributed in CHIP and a daily prize pool of $10,000 CHIP.
GPU is the capital-intensive foundation of the AI era, and USDai's yield from GPU financing spreads remains very robust.
Even after the Binance Wallet campaign ends, this underlying annualized yield is still well worth earning.
Catch the last few days of the campaign and get in on it now~