Binance Square
几米2026
382 Posts

几米2026

Open Trade
ZK Holder
ZK Holder
Occasional Trader
8.6 Years
19 Following
53 Followers
94 Liked
Posts
Portfolio
·
--
See translation
白宫喊话民主党:接受加密限制,但民主党嫌太松? 白宫想让民主党别闹了——乖乖接受针对特朗普的加密资产限制。但民主党回呛:CLARITY 法案就是个半成品,远远不够。 来看这本法案到底写了啥: - 暂时禁止高级官员发行或赞助加密货币(包括总统、副总统、国会议员、联邦法官) - 不过,过去的活动不追溯,特朗普的 World Liberty Financial 持股也可能被豁免(等于白限制) - 执法权交给司法部,罚款上限 50 万美元 - 监管机构可以不让平台列出被禁资产 - 限制到 2029 年初自动失效 民主党还加码:州总检察长也该有执法权。 然而现实很骨感:法案在参议院还差 60 票。要是夏季 deadline 一过,2026 年通过的概率直接砍半。 所以,这到底是管制加密,还是管制特朗普的生意?你觉得谁在演谁?
白宫喊话民主党:接受加密限制,但民主党嫌太松?

白宫想让民主党别闹了——乖乖接受针对特朗普的加密资产限制。但民主党回呛:CLARITY 法案就是个半成品,远远不够。

来看这本法案到底写了啥:
- 暂时禁止高级官员发行或赞助加密货币(包括总统、副总统、国会议员、联邦法官)
- 不过,过去的活动不追溯,特朗普的 World Liberty Financial 持股也可能被豁免(等于白限制)
- 执法权交给司法部,罚款上限 50 万美元
- 监管机构可以不让平台列出被禁资产
- 限制到 2029 年初自动失效

民主党还加码:州总检察长也该有执法权。

然而现实很骨感:法案在参议院还差 60 票。要是夏季 deadline 一过,2026 年通过的概率直接砍半。

所以,这到底是管制加密,还是管制特朗普的生意?你觉得谁在演谁?
USDT Turns Into the “Money-Laundering King” of Scam Rings A new UN report pulls back the curtain: in 2025, cyber-scam losses in the Asia-Pacific region are projected to skyrocket to between US$8.83 billion and US$11.41 billion—three times the 2023 figure (US$18.0–37.0 billion). And on the TRON blockchain, USDT has become the laundering channel criminals love most. A crypto service provider in the Mekong River region processed US$49.0–64.0 billion worth of illicit transactions in just four years. The number is even higher than the GDP of several countries in Southeast Asia. Victim distribution is especially telling: East Asia bears 71% of the losses; Australia and New Zealand account for 16%–20%; while Southeast Asia’s own locals make up only 8%–13%. Why? Because Southeast Asia is the scam factory—the production lines are here, but the customers (victims) are all over the world. Just in the United States, in 2024 alone, it was swindled out of more than US$10 billion by Southeast Asian scam groups. In Thailand, online-crime cases jumped from fewer than 20,000 in 2023 to 380,000 in 2025, with losses exceeding US$700 million. Malaysia is even more intense: in 2025, losses reached US$730 million—doubling year over year. Singapore introduced legislative intervention, finally bringing losses down from US$822 million in 2024 to US$686 million in 2025. While blockchain is still loudly preaching decentralization and freedom, criminals have already used it to flee. So, will the next target be you?
USDT Turns Into the “Money-Laundering King” of Scam Rings

A new UN report pulls back the curtain: in 2025, cyber-scam losses in the Asia-Pacific region are projected to skyrocket to between US$8.83 billion and US$11.41 billion—three times the 2023 figure (US$18.0–37.0 billion). And on the TRON blockchain, USDT has become the laundering channel criminals love most.

A crypto service provider in the Mekong River region processed US$49.0–64.0 billion worth of illicit transactions in just four years. The number is even higher than the GDP of several countries in Southeast Asia.

Victim distribution is especially telling: East Asia bears 71% of the losses; Australia and New Zealand account for 16%–20%; while Southeast Asia’s own locals make up only 8%–13%. Why? Because Southeast Asia is the scam factory—the production lines are here, but the customers (victims) are all over the world.

Just in the United States, in 2024 alone, it was swindled out of more than US$10 billion by Southeast Asian scam groups.

In Thailand, online-crime cases jumped from fewer than 20,000 in 2023 to 380,000 in 2025, with losses exceeding US$700 million. Malaysia is even more intense: in 2025, losses reached US$730 million—doubling year over year. Singapore introduced legislative intervention, finally bringing losses down from US$822 million in 2024 to US$686 million in 2025.

While blockchain is still loudly preaching decentralization and freedom, criminals have already used it to flee.

So, will the next target be you?
DeFi Truth: Loans Are the Real Money Everyone focuses on TVL, but almost nobody looks at loans. Over the past 3 months, borrowers pulled $8.0 billion out of DeFi lending markets. Meanwhile, Spark Finance bucked the trend by lending an additional $0.8 billion—up 82%. During the same period, the entire market fell 25%. TVL can be “boosted” with points and rewards. Loans don’t work that way—no one will borrow real money just to rack up points and pay interest. People only borrow from trusted protocols. And right now, they’re doing it in a falling market. Data from DefiLlama.
DeFi Truth: Loans Are the Real Money

Everyone focuses on TVL, but almost nobody looks at loans.

Over the past 3 months, borrowers pulled $8.0 billion out of DeFi lending markets.

Meanwhile, Spark Finance bucked the trend by lending an additional $0.8 billion—up 82%.

During the same period, the entire market fell 25%.

TVL can be “boosted” with points and rewards.

Loans don’t work that way—no one will borrow real money just to rack up points and pay interest.

People only borrow from trusted protocols.

And right now, they’re doing it in a falling market.

Data from DefiLlama.
Article
A public letter from Huang Renxun that exposes the most honest interest map in the AI industryLate July, the U.S. government grew concerned about the open-source KIMI K3 model. Rather than cheering for technology to benefit everyone, the first reaction in Washington was to consider a ban. Twenty-five companies, including NVIDIA, IBM, Dell, Meta, and Microsoft, issued an open letter opposing restrictions on open-sourcing AI models. At first glance, it looks like a dispute over the path between openness and closure, but who signed and who didn’t reveals the truth. OpenAI, Anthropic, and Google—three closed-source tech giants—were all absent. They were not silent for long; they publicly accused Chinese companies of “stealing” capabilities through distillation, framing competition as a safety threat. Now that a ban policy is coming to light, their earlier remarks are essentially the raw material for the policy.

A public letter from Huang Renxun that exposes the most honest interest map in the AI industry

Late July, the U.S. government grew concerned about the open-source KIMI K3 model. Rather than cheering for technology to benefit everyone, the first reaction in Washington was to consider a ban.
Twenty-five companies, including NVIDIA, IBM, Dell, Meta, and Microsoft, issued an open letter opposing restrictions on open-sourcing AI models. At first glance, it looks like a dispute over the path between openness and closure, but who signed and who didn’t reveals the truth.
OpenAI, Anthropic, and Google—three closed-source tech giants—were all absent. They were not silent for long; they publicly accused Chinese companies of “stealing” capabilities through distillation, framing competition as a safety threat. Now that a ban policy is coming to light, their earlier remarks are essentially the raw material for the policy.
ICYMI: Injective has submitted an application for a Transfer Agent license to the SEC and released the MiCA whitepaper in Europe. It’s like getting a compliance “driver’s license.” Injective can now, on the regulatory tracks in the United States and the European Union, package RWA (real-world assets) and $INJ and launch them on a new route. The seasoned SEC is a strict examiner, and Europe’s MiCA is the updated traffic code—both sides recognize it, meaning the lawful marriage between traditional assets and the crypto world has an official notary. In one sentence: $INJ ’s compliance passport has been stamped, and the door to global liquidity is opening.
ICYMI: Injective has submitted an application for a Transfer Agent license to the SEC and released the MiCA whitepaper in Europe.

It’s like getting a compliance “driver’s license.” Injective can now, on the regulatory tracks in the United States and the European Union, package RWA (real-world assets) and $INJ and launch them on a new route.

The seasoned SEC is a strict examiner, and Europe’s MiCA is the updated traffic code—both sides recognize it, meaning the lawful marriage between traditional assets and the crypto world has an official notary.

In one sentence: $INJ ’s compliance passport has been stamped, and the door to global liquidity is opening.
The EU has fined Google $1 billion again, and this time the thunder is really loud On July 23 local time, the European Commission issued Google an €890 million fine (about $1 billion)—€460 million for its search business and €430 million for the Play Store. The reason was that Google manipulated search results, suppressed rivals, and blocked app developers from directing users to cheaper deals outside the Play Store. In addition to the fine, the EU also required Google to treat third-party services fairly in search results; and to allow developers to freely communicate outside the Play Store, promote offers, and sign contracts. Google must complete the整改 within 60 days. This is the first time Google has been fined under the EU’s Digital Markets Act (DMA), and it is also the EU’s third enforcement action under the law after Apple and Meta were penalized last year. The DMA states that if your platform has 45 million active users and 10,000 business users in Europe, and its revenue exceeded €7.5 billion or its market value exceeded $75 billion over the past three years, it will be designated as a “gatekeeper” and then subjected to strict rules—with fines of up to 10% of global annual revenue. The European Commission said Google’s search remediation plan has made “substantial progress,” and its Play Store plan has also made “good progress.” They are also discussing with Google how this decision will affect Google’s “AI Overviews” and “AI Mode.” Google is not convinced. Kent Walker, president of global affairs, said that to comply, they would have to remove real-time search features popular with European users—such as instant prices and direct booking information for hotels, flights, and restaurants—and also dismantle Play’s safety protections. He complained: “This is not fair competition; it is a small group of self-interested complainants forcing us to degrade our product.” This latest fine brings the EU’s cumulative penalties against Google to more than $12 billion. Last September, the EU also fined Google nearly €3 billion for ad-tech violations. The timing is sensitive: when the penalty was announced, it was only one day before the White House was expected to unveil a new round of tariffs. U.S. Trade Representative Jamieson Greer had previously warned the EU not to keep fining American tech companies. Recently, 25 Republican lawmakers in the U.S. wrote to Trump, urging action against the EU’s tech regulation and even proposing a trade investigation. The EU’s hammer, America’s tariffs—both sides are escalating. What do you think: in this transatlantic tech settling of accounts, who ends up paying the bill in the end?
The EU has fined Google $1 billion again, and this time the thunder is really loud

On July 23 local time, the European Commission issued Google an €890 million fine (about $1 billion)—€460 million for its search business and €430 million for the Play Store. The reason was that Google manipulated search results, suppressed rivals, and blocked app developers from directing users to cheaper deals outside the Play Store. In addition to the fine, the EU also required Google to treat third-party services fairly in search results; and to allow developers to freely communicate outside the Play Store, promote offers, and sign contracts. Google must complete the整改 within 60 days.

This is the first time Google has been fined under the EU’s Digital Markets Act (DMA), and it is also the EU’s third enforcement action under the law after Apple and Meta were penalized last year. The DMA states that if your platform has 45 million active users and 10,000 business users in Europe, and its revenue exceeded €7.5 billion or its market value exceeded $75 billion over the past three years, it will be designated as a “gatekeeper” and then subjected to strict rules—with fines of up to 10% of global annual revenue. The European Commission said Google’s search remediation plan has made “substantial progress,” and its Play Store plan has also made “good progress.” They are also discussing with Google how this decision will affect Google’s “AI Overviews” and “AI Mode.”

Google is not convinced. Kent Walker, president of global affairs, said that to comply, they would have to remove real-time search features popular with European users—such as instant prices and direct booking information for hotels, flights, and restaurants—and also dismantle Play’s safety protections. He complained: “This is not fair competition; it is a small group of self-interested complainants forcing us to degrade our product.”

This latest fine brings the EU’s cumulative penalties against Google to more than $12 billion. Last September, the EU also fined Google nearly €3 billion for ad-tech violations.

The timing is sensitive: when the penalty was announced, it was only one day before the White House was expected to unveil a new round of tariffs. U.S. Trade Representative Jamieson Greer had previously warned the EU not to keep fining American tech companies. Recently, 25 Republican lawmakers in the U.S. wrote to Trump, urging action against the EU’s tech regulation and even proposing a trade investigation.

The EU’s hammer, America’s tariffs—both sides are escalating. What do you think: in this transatlantic tech settling of accounts, who ends up paying the bill in the end?
TRON has moved into the crypto indices of S&P and Pantera—mainstream finance is finally starting to assess blockchain projects the way they “check IDs.” This index doesn’t care whether your coin price has gone up; it looks at three things instead: whether your chain is being used, whether your funds are actually moving, and whether your network is active. As a result, Bitcoin and XRP were knocked out, while TRX unexpectedly became one of the top five. How does TRON prove itself? 394 million user accounts, $90 billion worth of USDT as a steady main payment channel, and 4.5 trillion USD in transfer volume year-to-date—ranked first across all networks. It has also carried out a series of integrations with Anchorage Digital, Securitize, and Bitnomial, connecting to the U.S. compliant market. Founder Justin Sun puts it plainly: if digital assets are going to go mainstream, the criteria for scoring must shift from “trading/speculation” to “use.” On-chain activity, adoption rate, and network earnings—these are the new yardsticks. A longtime cross-industry playbook: what Wall Street used to score stocks has now landed on blockchain. Time is running out for speculators. Do you think this “pragmatist” index can filter out truly valuable networks—or will it miss the next Bitcoin?
TRON has moved into the crypto indices of S&P and Pantera—mainstream finance is finally starting to assess blockchain projects the way they “check IDs.”

This index doesn’t care whether your coin price has gone up; it looks at three things instead: whether your chain is being used, whether your funds are actually moving, and whether your network is active. As a result, Bitcoin and XRP were knocked out, while TRX unexpectedly became one of the top five.

How does TRON prove itself? 394 million user accounts, $90 billion worth of USDT as a steady main payment channel, and 4.5 trillion USD in transfer volume year-to-date—ranked first across all networks. It has also carried out a series of integrations with Anchorage Digital, Securitize, and Bitnomial, connecting to the U.S. compliant market.

Founder Justin Sun puts it plainly: if digital assets are going to go mainstream, the criteria for scoring must shift from “trading/speculation” to “use.” On-chain activity, adoption rate, and network earnings—these are the new yardsticks.

A longtime cross-industry playbook: what Wall Street used to score stocks has now landed on blockchain. Time is running out for speculators.

Do you think this “pragmatist” index can filter out truly valuable networks—or will it miss the next Bitcoin?
Collector_Crypt Acquires a 62.8% share of the tokenized collectibles market. Monthly trading volume surged from $97 million in January to $406 million in June over four months (up more than 3x), while also pushing Courtyard off the throne.
Collector_Crypt Acquires a 62.8% share of the tokenized collectibles market.

Monthly trading volume surged from $97 million in January to $406 million in June over four months (up more than 3x), while also pushing Courtyard off the throne.
$68,000, long-vs-short battle Bitcoin has been rising for four straight weeks, from $57,800 to $66,000—cumulatively nearing 15%. But what truly determines its fate is the $68,000 ahead. Why $68,000? Because three layers of pressure stack here: the average cost of buyers over the past five months, the technical resistance from the previous rebound, and the psychological level the market cares most about. With these three walls piled up together, it’s destined to be a tough fight. Once the price moves near $68,000, a large amount of positions currently sitting at a floating loss will come close to breaking even. Some will continue to bet, while others will decisively exit. When this level is tested the first time, the market likely won’t be calm—whether it breaks through or pulls back, it will trigger a chain reaction. More importantly, liquidity is currently on the weak side. CME Bitcoin futures open interest has fallen to the lowest level since 2023, and the 30-day trading volume is only 62% of the in-year average. The market may look calm, but in reality it’s more sensitive than usual. The good news is that selling pressure is easing. Flows out of Bitcoin ETFs have started to stabilize: this month, net outflows occurred on only one-third of trading days (up to 90% in June). Sellers aren’t as resolute, but buyers still haven’t returned in a big way.
$68,000, long-vs-short battle

Bitcoin has been rising for four straight weeks, from $57,800 to $66,000—cumulatively nearing 15%. But what truly determines its fate is the $68,000 ahead.

Why $68,000? Because three layers of pressure stack here: the average cost of buyers over the past five months, the technical resistance from the previous rebound, and the psychological level the market cares most about. With these three walls piled up together, it’s destined to be a tough fight.

Once the price moves near $68,000, a large amount of positions currently sitting at a floating loss will come close to breaking even. Some will continue to bet, while others will decisively exit. When this level is tested the first time, the market likely won’t be calm—whether it breaks through or pulls back, it will trigger a chain reaction.

More importantly, liquidity is currently on the weak side. CME Bitcoin futures open interest has fallen to the lowest level since 2023, and the 30-day trading volume is only 62% of the in-year average. The market may look calm, but in reality it’s more sensitive than usual.

The good news is that selling pressure is easing. Flows out of Bitcoin ETFs have started to stabilize: this month, net outflows occurred on only one-third of trading days (up to 90% in June). Sellers aren’t as resolute, but buyers still haven’t returned in a big way.
The supply of long-term Bitcoin holders has just hit a new all-time high. Data from @coinglass_com. More people are locking BTC away in “cold wallets” to let it sit and gather dust—steady holding, with no intention to move. What does this mean? Investors are continuing to accumulate, not just doing a quick short-term trade and then running. They believe this story still has a long way to go. When you hold a coin for a long time, it turns into faith.
The supply of long-term Bitcoin holders has just hit a new all-time high. Data from @coinglass_com.

More people are locking BTC away in “cold wallets” to let it sit and gather dust—steady holding, with no intention to move.

What does this mean? Investors are continuing to accumulate, not just doing a quick short-term trade and then running. They believe this story still has a long way to go.

When you hold a coin for a long time, it turns into faith.
Yesterday afternoon, a Chinese research team threw out a so-called “EEG bomb.” For the first time in the world, they achieved cross-regional, synchronized EEG collection with over a thousand people at the same time. The devices are small enough to carry around, yet their precision hasn’t been compromised. Network latency has also been dealt with—milliseconds-level alignment, bringing the brainwaves of a thousand people together for analysis. Two key technologies: miniaturization + high precision, and network latency + time synchronization. They conquered both. So what does this mean? AI will no longer rely only on “secondhand information” like text, images, and videos to learn. It can now directly read your brain state from neural signals. The “raw data” of human cognition—AI is about to start looking at it.
Yesterday afternoon, a Chinese research team threw out a so-called “EEG bomb.”

For the first time in the world, they achieved cross-regional, synchronized EEG collection with over a thousand people at the same time. The devices are small enough to carry around, yet their precision hasn’t been compromised. Network latency has also been dealt with—milliseconds-level alignment, bringing the brainwaves of a thousand people together for analysis.

Two key technologies: miniaturization + high precision, and network latency + time synchronization. They conquered both.

So what does this mean? AI will no longer rely only on “secondhand information” like text, images, and videos to learn. It can now directly read your brain state from neural signals.

The “raw data” of human cognition—AI is about to start looking at it.
DeepSeek Founder Liang Wenfeng: China’s AI chips will break the impasse within a year On July 23, details from an internal investor exchange meeting at DeepSeek leaked. Founder Liang Wenfeng said plainly that China’s AI chip replacement is entering a historic window. There are no adaptation barriers between hardware and the ecosystem, and NVIDIA can’t stop the replacement process. The only issue is insufficient production capacity—however, it’s expected that within a year it can reverse the bias of “can’t be used” and “not good to use.” He revealed that NVIDIA’s CUDA moat is rapidly eroding. Export controls combined with industry changes have opened a time window for Chinese chips. DeepSeek has already partnered with Huawei, securing 16,000 950 cards to run and validate the ecosystem. Huawei’s 950 supernode delivers performance and pricing that fully matches NVIDIA’s GB200 and GB300. Liang Wenfeng remains optimistic about the outlook for China’s computing power industry.
DeepSeek Founder Liang Wenfeng: China’s AI chips will break the impasse within a year

On July 23, details from an internal investor exchange meeting at DeepSeek leaked.
Founder Liang Wenfeng said plainly that China’s AI chip replacement is entering a historic window.
There are no adaptation barriers between hardware and the ecosystem, and NVIDIA can’t stop the replacement process.
The only issue is insufficient production capacity—however, it’s expected that within a year it can reverse the bias of “can’t be used” and “not good to use.”

He revealed that NVIDIA’s CUDA moat is rapidly eroding.
Export controls combined with industry changes have opened a time window for Chinese chips.
DeepSeek has already partnered with Huawei, securing 16,000 950 cards to run and validate the ecosystem.
Huawei’s 950 supernode delivers performance and pricing that fully matches NVIDIA’s GB200 and GB300.

Liang Wenfeng remains optimistic about the outlook for China’s computing power industry.
Is $1.6B in liquidity just “sleeping”? A study by 1inch and Dune found that currently $1.6 billion (about RMB 11 billion) worth of DeFi liquidity is sitting idle. It’s like buying a bunch of snacks and stuffing them in a cabinet, forgetting to eat them. As institutional capital and tokenized assets flow onto the chain, can you “wake up” these “inventory” and directly determine whether you grow or waste? Don’t let outdated liquidity designs hold you back.
Is $1.6B in liquidity just “sleeping”?

A study by 1inch and Dune found that currently $1.6 billion (about RMB 11 billion) worth of DeFi liquidity is sitting idle. It’s like buying a bunch of snacks and stuffing them in a cabinet, forgetting to eat them. As institutional capital and tokenized assets flow onto the chain, can you “wake up” these “inventory” and directly determine whether you grow or waste? Don’t let outdated liquidity designs hold you back.
DeepSeek raised $7.4 billion, bidding farewell to pure idealism Chinese AI company DeepSeek has completed the first round of external fundraising in its history, raising more than 50 billion RMB (about $7.4 billion). Pre-money valuation was 367.5 billion RMB (about $54.3 billion). That research-minded young prodigy who once said “no fundraising, no going public, no commercialization” has finally been invited into the capital party scene. At an investor communication event in May lasting nearly four hours, founder Liang Wenfeng explained that “restraint makes success easier,” while also showing his cards: API pricing will only generate a reasonable profit sufficient to recoup costs, and open-sourcing won’t get in the way of making money. He believes the main gap between China and the U.S. in AI lies in compute power, and that domestic chips are on the historical stage—when the industry reaches its endgame, there are likely only three or four large-model players left.
DeepSeek raised $7.4 billion, bidding farewell to pure idealism

Chinese AI company DeepSeek has completed the first round of external fundraising in its history, raising more than 50 billion RMB (about $7.4 billion). Pre-money valuation was 367.5 billion RMB (about $54.3 billion). That research-minded young prodigy who once said “no fundraising, no going public, no commercialization” has finally been invited into the capital party scene. At an investor communication event in May lasting nearly four hours, founder Liang Wenfeng explained that “restraint makes success easier,” while also showing his cards: API pricing will only generate a reasonable profit sufficient to recoup costs, and open-sourcing won’t get in the way of making money. He believes the main gap between China and the U.S. in AI lies in compute power, and that domestic chips are on the historical stage—when the industry reaches its endgame, there are likely only three or four large-model players left.
In the past 30 days, exchanges have become the stage for a massive transfer of funds. Bitget and OKX are the winners, with net inflows of as much as +677 million yuan and +420 million dollars, respectively (congratulations to both—treat yourselves!). On the other hand, Binance and Bybit have become major drains, with outflows of 1.65 billion yuan and 416 million dollars each. In the crypto world, there is no permanent throne—only continuously moving wallets.
In the past 30 days, exchanges have become the stage for a massive transfer of funds. Bitget and OKX are the winners, with net inflows of as much as +677 million yuan and +420 million dollars, respectively (congratulations to both—treat yourselves!). On the other hand, Binance and Bybit have become major drains, with outflows of 1.65 billion yuan and 416 million dollars each. In the crypto world, there is no permanent throne—only continuously moving wallets.
NEAR’s mainnet learning automatic scaling. Yesterday, the status size of a shard hit the threshold—like a supermarket checkout line getting backed up; the system simply opened 1 additional window: from 9 shards to 10. The split is deterministic, with no human intervention throughout. Dynamic sharding is live and is getting to work (and you don’t have to ask for any programmers).
NEAR’s mainnet learning automatic scaling. Yesterday, the status size of a shard hit the threshold—like a supermarket checkout line getting backed up; the system simply opened 1 additional window: from 9 shards to 10. The split is deterministic, with no human intervention throughout. Dynamic sharding is live and is getting to work (and you don’t have to ask for any programmers).
OpenAI’s Compute Spending Soars to $750 Billion On Wednesday, OpenAI raised its forecast for compute spending through 2030 from $600 billion to $750 billion. Of the additional $150 billion, half is due to newly signed cloud contracts, and half is because its “Stargate” project fell short—so they have to build data centers themselves. The latest move is investing $20 billion (about $2.9 billion) in Georgia, U.S., to launch the “Camellia Project”—OpenAI’s first time designing and developing a data center under its own leadership. The land has already been secured, and it has also signed a contract with Georgia’s power provider to guarantee access to 3.2 gigawatts of electricity from 2028 to 2032. The company’s vice president of compute strategy said that the reserved fee for buying electricity “isn’t cheap,” though he did not specify the exact amount (likely to give the other party confidence to begin expansion first). “Stargate” originally aimed to burn $500 billion together with SoftBank, Oracle, and Middle East investors, but progress has been difficult. OpenAI recently hired Brent Mayo, who previously built data centers for Musk, to focus specifically on overseeing the construction timeline. The biggest spending items still involve outsourced compute capacity: Oracle 6 gigawatts; Amazon AWS $138 billion over eight years (including its in-house chips); and Microsoft Azure an additional $250 billion (no timeline provided). The CEO once said it would invest $1.4 trillion by 2033, but the CFO later changed the figure to only spending $600 billion by 2030. The numbers don’t quite match up, but the money going out is real: this year, the revenue target is $13 billion, while cash burn reaches as high as $25 billion; next year, it’s expected to burn $57 billion. A hopeful follow-up: when a company burns money at twice the rate it earns, is it investing in the future—or betting its life?
OpenAI’s Compute Spending Soars to $750 Billion

On Wednesday, OpenAI raised its forecast for compute spending through 2030 from $600 billion to $750 billion. Of the additional $150 billion, half is due to newly signed cloud contracts, and half is because its “Stargate” project fell short—so they have to build data centers themselves.

The latest move is investing $20 billion (about $2.9 billion) in Georgia, U.S., to launch the “Camellia Project”—OpenAI’s first time designing and developing a data center under its own leadership. The land has already been secured, and it has also signed a contract with Georgia’s power provider to guarantee access to 3.2 gigawatts of electricity from 2028 to 2032. The company’s vice president of compute strategy said that the reserved fee for buying electricity “isn’t cheap,” though he did not specify the exact amount (likely to give the other party confidence to begin expansion first).

“Stargate” originally aimed to burn $500 billion together with SoftBank, Oracle, and Middle East investors, but progress has been difficult. OpenAI recently hired Brent Mayo, who previously built data centers for Musk, to focus specifically on overseeing the construction timeline.

The biggest spending items still involve outsourced compute capacity: Oracle 6 gigawatts; Amazon AWS $138 billion over eight years (including its in-house chips); and Microsoft Azure an additional $250 billion (no timeline provided).

The CEO once said it would invest $1.4 trillion by 2033, but the CFO later changed the figure to only spending $600 billion by 2030. The numbers don’t quite match up, but the money going out is real: this year, the revenue target is $13 billion, while cash burn reaches as high as $25 billion; next year, it’s expected to burn $57 billion.

A hopeful follow-up: when a company burns money at twice the rate it earns, is it investing in the future—or betting its life?
Hackers move $24.15 million—then the bridge collapses again? Tonight at 21:30 UTC, security firm Blockaid found that a USDC bridge used by AFX Trade on Arbitrum had been exploited, resulting in losses of about $24.15 million. The attacker quickly transferred the funds to the Ethereum chain, converting them into 12,467 ETH at an average price of $1,937. Arbitrum co-founder sgoldfed rushed to clarify that the problem was with AFX’s own bridge, not Arbitrum’s native bridge. AFX’s official account had gone silent 17 hours before the incident and still hasn’t responded. One thing to nitpick: the attackers’ average purchase price for ETH was $1,937—this level looks even more comfortable than many people’s cost basis. It’s not that the bridge doesn’t work—it’s that the bridge’s security code is held together like macaroni glued with glue. Defenders were asleep, while the hackers were doing the math. Before you lock the door next time, remember to hide the keys first.
Hackers move $24.15 million—then the bridge collapses again?

Tonight at 21:30 UTC, security firm Blockaid found that a USDC bridge used by AFX Trade on Arbitrum had been exploited, resulting in losses of about $24.15 million. The attacker quickly transferred the funds to the Ethereum chain, converting them into 12,467 ETH at an average price of $1,937. Arbitrum co-founder sgoldfed rushed to clarify that the problem was with AFX’s own bridge, not Arbitrum’s native bridge. AFX’s official account had gone silent 17 hours before the incident and still hasn’t responded.

One thing to nitpick: the attackers’ average purchase price for ETH was $1,937—this level looks even more comfortable than many people’s cost basis. It’s not that the bridge doesn’t work—it’s that the bridge’s security code is held together like macaroni glued with glue. Defenders were asleep, while the hackers were doing the math. Before you lock the door next time, remember to hide the keys first.
Aave gets ruthless—it’s even scary for itself Since the launch of V4, the numbers have been going through the roof: deposits have surpassed $300 million, and active loans have exceeded $100 million—both setting new all-time highs. Aave’s total active loans are now around $10.6 billion, more than all other EVM lending protocols combined. The first cross-chain move is live on Avalanche. New Stable Vaults were released too: a fixed-rate stablecoin yield infrastructure aimed at fintechs, wallets, and exchanges—driven by Chainlink CCIP and price feeds, targeting the $310 billion stablecoin market. On the security front: CertiK Skynet score 94.48 (AAA), and RootData transparency is Grade A—with a fundamentals score of 98, and the finance section gets a perfect 100. Growth, governance, security… Aave isn’t just filling in boxes—it’s drawing a flawless checkmark. Ask them one question: guess what Aave’s next step will be—keep pushing DeFi harder, or directly eat into the stablecoin track of traditional finance?
Aave gets ruthless—it’s even scary for itself

Since the launch of V4, the numbers have been going through the roof: deposits have surpassed $300 million, and active loans have exceeded $100 million—both setting new all-time highs. Aave’s total active loans are now around $10.6 billion, more than all other EVM lending protocols combined. The first cross-chain move is live on Avalanche.

New Stable Vaults were released too: a fixed-rate stablecoin yield infrastructure aimed at fintechs, wallets, and exchanges—driven by Chainlink CCIP and price feeds, targeting the $310 billion stablecoin market.

On the security front: CertiK Skynet score 94.48 (AAA), and RootData transparency is Grade A—with a fundamentals score of 98, and the finance section gets a perfect 100.

Growth, governance, security… Aave isn’t just filling in boxes—it’s drawing a flawless checkmark.

Ask them one question: guess what Aave’s next step will be—keep pushing DeFi harder, or directly eat into the stablecoin track of traditional finance?
The prediction market isn’t a casino—it’s a new playground for insider trading Bloomberg Businessweek’s latest investigation pulls back the curtain on Polymarket. The data comes from Dune, and each traceable wallet is linked to the exchange where its funding source transactions originated. Key findings: — About $21 billion (half of the trackable trading volume) comes from U.S.-regulated exchanges — Of that, 70% flows to the Iran war markets — Operations flagged as “insider trading” are 3 times more likely to be tied to U.S. funds — 38 associated wallets, with a 98% win rate, cashed out $1.6 million through a single Coinbase address In other words, you think it’s the global retail crowd gambling on the war—but it’s actually some people using U.S. accounts to play “I know it, but you don’t—yet.” If prediction markets are the future, then its first bug is already live. Do you consider this kind of trading—knowing the outcome in advance—insider trading, or information arbitrage?
The prediction market isn’t a casino—it’s a new playground for insider trading

Bloomberg Businessweek’s latest investigation pulls back the curtain on Polymarket.

The data comes from Dune, and each traceable wallet is linked to the exchange where its funding source transactions originated.

Key findings:
— About $21 billion (half of the trackable trading volume) comes from U.S.-regulated exchanges
— Of that, 70% flows to the Iran war markets
— Operations flagged as “insider trading” are 3 times more likely to be tied to U.S. funds
— 38 associated wallets, with a 98% win rate, cashed out $1.6 million through a single Coinbase address

In other words, you think it’s the global retail crowd gambling on the war—but it’s actually some people using U.S. accounts to play “I know it, but you don’t—yet.”

If prediction markets are the future, then its first bug is already live.

Do you consider this kind of trading—knowing the outcome in advance—insider trading, or information arbitrage?
Log in to explore more content
Join global crypto users on Binance Square
⚡️ Get latest and useful information about crypto.
💬 Trusted by the world’s largest crypto exchange.
👍 Discover real insights from verified creators.
Email / Phone number
Sitemap
Cookie Preferences
Platform T&Cs