火必周刊

火币资讯Pubblicato 2022-12-04Pubblicato ultima volta 2022-12-04

Introduzione

本周加密市场都发生了哪些事？一起来回顾吧。

1. HT将作为火必Huobi生态系统内的唯一通证；

2. 火必Huobi与波场各方合作推出首款国家代币DMC；

3. 欧盟委员会计划12月7日公布加密税收新提案；

4. 美国SEC在2022-2026财年战略计划中批露涉及加密资产的举措；

5. Animoca Brands拟推出规模高达20亿美元的元宇宙基金。

”“

#政策

1. 韩国对发行加密货币现状展开全面调查并起草《数字资产法》；

2. 新加坡政府本周面临议会对FTX后果的质询，包括淡马锡对FTX投资的尽职调查；

3. 阿根廷修改反洗钱法，拟议设立VASP注册管理机构；

4. 美众议院金融服务委员会将于12月13日举行FTX事件听证会；

5. 知情人士：欧盟委员会计划12月7日公布加密税收新提案；

6. 乌克兰考虑建立可以促进加密货币交易的CBDC；

7. CFTC主席：若CFTC获得对现货市场的监管权，将有益于保护普通投资者；

8. 巴西投票通过加密法案，承认比特币可用作支付手段和投资资产；

9. 美国SEC在2022-2026财年战略计划中批露涉及加密资产的举措；

10. 印度尼西亚央行发布数字货币白皮书；

11. 欧盟新法案将强制加密提供商向税务机关报告欧盟客户交易的详细信息；

12. 意大利在2023年预算提案中计划对数字资产收益征收26%的税；

13. 美CFTC主席：比特币是唯一应该被视为商品的加密货币；

14. 韩国金融服务委员会：将制定以投资者保护为中心的加密监管体系；

15. 乌兹别克斯坦批准加密新规，涉及资产注册、发行以及流通。

”“

#币种项目

1. 数据：Tether已封禁807个地址，总计超4.35亿枚USDT；

2. Telegram现支持在应用聊天界面直接交易BTC和TON；

3. 两个地址持有超过73%的FTT代币，FTX黑客为第二大持币者；

4. Alameda Research有超4860万枚SOL被锁定质押，占总供应量8.75%；

5. 约3000万枚USDC新铸币进入流通；

6. 数据：过去一个月价值超28亿美元的BTC流出交易所；

7. 数据：WBTC月供应量降幅达12.57%，总供应量回到2021年9月水平；

8. CoinShares：上周数字资产投资产品净流出2300万美元；

9. HT将作为火必Huobi生态系统内的唯一通证；

10. 火必Huobi与波场各方合作推出首款国家代币DMC；

11. 波卡首个平行链Statemint发布未来路线图，将支持DEX和NFT；

12. 报告：美国拥有45.3%以太坊节点，全球排名第一；

13. 以太坊Layer2 Gas使用量超1000亿gas，创历史新高；

14. 比特币挖矿难度或将迎来2022年内最大降幅；

15. Vitalik Buterin：以“治理权”作为衡量Token价值的指标是反常的；

16. 数据：自以太坊合并以来，总净发行量仅为2,317 ETH；

17. 火必Huobi与波场Poloniex达成战略合作伙伴关系；

18. Telegram已在TON上出售价值5000万美元的用户名；

19. 以太坊开发者：扩容解决方案EIP-4844的Devnet 3推迟到下周发布；

20. The Graph将增加对Polygon的支持，Polygon将参与The Graph的MIPs激励计划。

”“

#NFT

1. Avalanche链上NFT销售额超越Polygon；

2. OpenSea已集成BNB Chain；

3. 保时捷NFT官网更新项目细节，将于12月20日开放注册；

4. NFT市场Blur二次空投将于12月5日上线；

5. 加密支付平台MoonPay将于明年初推出灵魂绑定NFT忠诚度计划；

6. 杂志《花花公子》推出首个原创NFT系列“The Playboy Party People”。

”“

#元宇宙

1. Animoca Brands拟推出规模高达20亿美元的元宇宙基金；

2. 韩国新韩银行推出Shinamon元宇宙平台；

3. Meta敦促政策制定者暂缓制定管理元宇宙的新规则。

”“

#WEB3

1. 万事达卡已提交加密和Web3安全相关商标申请；

2. Jack Dorsey旗下比特币公司TBD拟申请Web 5商标以防止滥用；

3. 暹罗商业银行旗下SCB 10X开设Web3协作空间DistrictX。

”“

#DeFi

1. Aave v2停用YFI、CRV等17个低流动性资产池；

2. Across Protocol：ACX代币将在Balancer上建立流动性池；

3. Compound调整借贷上限提案获得通过；

4. Cardano生态DEX AdaSwap上线主网；

5. Kaiko推出市场数据产品用于跟踪Aave、Compound和MakerDAO；

6. Aave创始人：Aave社区正准备在以太坊上部署Aave V3；

7. 11月DEX交易额重返1000亿美元上方，系过去6个月以来首次。

”“

#融资

1. 数字银行Atom Bank计划在IPO前融资5000万英镑；

2. 机构质押平台Klin完成1700万欧元融资；

3. 数字资产做市商Keyrock完成7200万美元B轮融资，Ripple领投；

4. 645 Ventures第四支基金Fund IV和首支精选基金Select I完成3.47亿美元募资，将专注于Web3领域。

”“

#其他

1. 巴哈马总检察长：正对FTX进行民事和刑事调查；

2. LINE旗下加密交易所Bitfront宣布关闭，明年3月底全面结束运营；

3. BlockFi起诉SBF并要求没收其在Robinhood价值5.75亿美元的股份；

4. BlockFi已出售约2.39亿美元加密货币以支付预期破产费用，近250名员工或将被裁；

5. Jack Dorsey旗下加密支付公司Block起诉Bitcoin.com商标侵权；

6. 推特正恢复6.2万个封禁账号，包括1个500万粉丝账号；

7. 彭博社：FTX通过Alameda持有的银行账户处理部分客户银行电汇；

8. 知情人士：Genesis债权人正与重组律师合作，寻求使公司免于破产的选择；

9. SBF：CZ的质疑和保证金问题共同造成FTX破产；

10. 韩国检方已申请对Terra联创Daniel Shin等8人发出事前拘捕令；

11. 摩根士丹利：加密市场预计将进一步破产和去杠杆化；

12. 贝莱德CEO：向FTX投资了2400万美元，仍然相信加密货币背后的技术；

13. 报告：2018年以来，超过3300只基金对加密货币进行了风险投资；

14. 马斯克：Twitter已开始清除虚假账户；

15. 韩国法院驳回对Terra联合创始人Daniel Shin的逮捕令；

16. FTX旗下衍生品交易平台LedgerX正挂牌出售；

17.马斯克：Twitter使用法定货币和加密货币进行支付是非常容易的；

18. SBF：相比其他客户，Alameda Research在FTX拥有更高的借款额度；

19. SBF：FTX客户在Alameda账户存入超50亿美元，无法解释资金去向；

20. 马斯克：社交媒体平台希望让meme币交易变得更加容易。

Letture associate

After Tokenized U.S. Treasury Bonds, Tokenized Stocks Are Becoming the New Battleground for RWA

**Tokenized Stocks Emerge as the New RWA Battleground** Following the initial surge of tokenized U.S. Treasuries, the tokenized stock market is rapidly expanding in both scale and quality. Unlike the currently stagnating treasury tokenization market, tokenized stock offerings are seeing significant growth and attracting major players from traditional finance, fintech, crypto exchanges, and native Web3 platforms. The U.S. SEC has outlined a framework for tokenized securities, which applies to stocks: **Issuer-Sponsored Tokenized Securities** (direct tokenization by the issuer, inheriting all shareholder rights but with strict compliance, e.g., Securitize); **Custodial Tokenized Securities** (tokenization of custodied interests, e.g., DTCC, Ondo's recent IVV/MU tokens); **Linked Securities** (tokenized debt notes backed by the stock, offering price exposure and greater on-chain utility, e.g., Ondo, xStocks, Robinhood's new Stock Tokens); and **Security-Based Swaps** (tokenized derivatives contracts, e.g., Robinhood's earlier Classic Stock Tokens). Key platforms are pursuing different strategies: * **Securitize** leads with an issuer-sponsored model, ensuring full rights but limiting on-chain interactions via its compliance-enforcing DS Protocol. * **Ondo** and **xStocks** use the linked security structure for broad accessibility on CEXs and DeFi, though this fragments liquidity and excludes U.S. users. * **Robinhood** recently launched linked security-based Stock Tokens, leveraging its user base and new Robinhood Chain. * Traditional infrastructure giants like **DTCC**, the **NYSE**, and **Nasdaq** are actively developing pilots and platforms for tokenized settlement and trading. * **Coinbase** has announced plans for tokenized stocks, likely using a structure that offers on-chain utility while excluding U.S. customers. Despite different approaches, all players are converging on tokenized stocks as the next major catalyst for the RWA sector. The evolving regulatory landscape and market adoption will shape this competitive new battlefield.

marsbit3 min fa

After Tokenized U.S. Treasury Bonds, Tokenized Stocks Are Becoming the New Battleground for RWA

marsbit3 min fa

Nvidia Faces Collective Selling by Funds: What Are Private Equity Firms Sniffing Out?

Several prominent Chinese private equity funds, including Gaoyi Asset and Dantoo's Orient Harbor, made significant portfolio adjustments in Q2 2024, as revealed in their latest SEC 13F filings. The most notable move was a collective retreat from AI chip leader NVIDIA, with Gaoyi reducing its stake by over 70%, Jilin Asset selling out completely, and Orient Harbor trimming its holdings. This shift does not signal a loss of faith in the AI trend, but rather a change in investment focus from "expectation-driven" to "realization-driven" valuation. Capital is flowing from crowded, high-valuation names towards segments with clearer profitability visibility and better risk-reward profiles. The funds' new major conviction is Taiwan Semiconductor Manufacturing Company (TSMC), which Gaoyi built into its top holding. The thesis centers on TSMC's role as a critical bottleneck in the AI supply chain due to its dominance in advanced semiconductor manufacturing and, crucially, advanced CoWoS packaging capacity. As demand for AI chips explodes, TSMC's "must-pass" foundry services grant it significant pricing power, suggesting a potential migration of profits from chip designers to manufacturers. Simultaneously, both Gaoyi and Orient Harbor substantially increased positions in memory chip companies like Micron and SanDisk. The investment logic here is twofold: a cyclical recovery in the memory market combined with a new, structural growth driver from AI. High Bandwidth Memory (HBM), essential for AI processor performance, is supply-constrained and its production diverts capacity from standard DRAM and NAND, creating a broad-based pricing tailwind for the memory sector. The collective reduction in NVIDIA reflects a view that its massive prior gains have compressed its future upside potential ("lowered赔率"). The market now demands flawless execution and faces new risks like rising costs from TSMC and HBM suppliers, as well as competition from cloud companies' custom chips. The funds' reallocation highlights a broader thematic: as the AI boom matures, investment opportunities and excess profits are shifting along the supply chain—from design to the physical constraints of manufacturing and memory. The next phase of AI investing may be defined by pricing power derived from tangible bottlenecks like advanced packaging and HBM capacity.

marsbit36 min fa

Nvidia Faces Collective Selling by Funds: What Are Private Equity Firms Sniffing Out?

marsbit36 min fa

Reverse Turing Test: This 'Pure Handcrafted Large Model' Is Driving Netizens Crazy

An art project called ChatTJB, humorously billed as a "next-generation, single-operator Large Language Experience" (LLE), gained viral attention in San Francisco. Promoted via a billboard, its website mimicked a genuine AI service with a chat interface, documentation, and even a "Pro" subscription. The twist? It was entirely manual. A single person, named Tucker, personally read, thought about, and typed replies to thousands of user queries using his thumbs, even hand-drawing images. He coined terms like "human-powered reasoning layer" to describe this process, highlighting its inherent human limitations: speed depended on his alertness, capacity was strictly one conversation at a time, and errors were traceable to a specific individual. The project's popularity overwhelmed Tucker, forcing a pause on new users and leading to plans for a community version with volunteer "AIs" (Average Individuals). While initially a source of amusement, Tucker noted that some users began sharing genuine personal concerns, valuing the knowledge that a real person was attentively listening and responding. The project satirizes AI hype while underscoring the irreplaceable value of human connection and attention in an age of automated, scalable interactions. A similar platform, "Your AI Slop Bores Me," was mentioned, where users either submit requests or role-play as the "AI" to fulfill others' text or drawing prompts, further emphasizing the human element behind the interface. The core message: sometimes, knowing there's a real person on the other end is what truly matters.

marsbit39 min fa

Reverse Turing Test: This 'Pure Handcrafted Large Model' Is Driving Netizens Crazy

marsbit39 min fa

Overnight, GPT-5.6 Sol Was Accelerated 14x by OpenAI

OpenAI, in collaboration with chipmaker Cerebras, has unveiled a limited preview of an "Ultrafast Mode" for its flagship GPT-5.6 Sol model. This new service tier reportedly achieves output speeds of up to 750 tokens per second—a 14x increase over the standard mode's baseline of ~53 tokens/s—without any loss in quality. Key to this acceleration is Cerebras's wafer-scale architecture (WSE-3), which houses model parameters entirely in on-chip SRAM to eliminate the memory bandwidth bottlenecks typical of traditional GPU clusters. In benchmark testing on the challenging "Humanity's Last Exam" (HLE), GPT-5.6 Sol in Ultrafast Mode answered all 2500 questions in 11 hours and 11 minutes, compared to over 78 hours for a competitor model, while maintaining similar accuracy. The speed boost also translated to a 5.6x faster end-to-end performance on the GDP-Val benchmark for economically valuable knowledge work. OpenAI highlights several potential applications for such rapid inference, including real-time event response and reliability analysis, dynamic financial research and security, complex customer support, interactive shopping assistance, and accelerated research and experimentation workflows that enable multiple iterative cycles within a single workday. This advancement may allow users to deploy the highest-tier models for tasks previously requiring slower secondary models, significantly compressing multi-step agent workflows from hours to minutes.

marsbit43 min fa

Overnight, GPT-5.6 Sol Was Accelerated 14x by OpenAI

marsbit43 min fa

Just Now, DeepSeek Harness Shocks the Open Source World: Everything as a Plugin

DeepSeek Harness, an open-source AI agent framework, has been released. Built on a "everything is a plugin" architecture using Cordis, it provides a modular SDK for constructing, running, and extending agents. The framework separates capabilities—like file systems, shells, web access, and sub-agents—into independent, swappable packages, assembled via a configuration file (`cordis.yml`). This allows creation of diverse agent types, from web UIs to headless automation. Key features include a structured Agent Loop for managing model requests and tool execution with concurrency and safety controls, and an authoritative Session Log that records all events to ensure state can be perfectly reconstructed. It offers multiple interaction modes (Web UI, TUI, Headless) and four preset agent configurations (Standard, PTC, Minimalist, Creative), with the Creative mode allowing agents limited self-modification. Security is built-in with a default `workspace-write` sandbox and an "ask for approval" policy for privileged operations. The framework demonstrates DeepSeek's aim to provide not just another coding assistant, but a foundational, auditable, and extensible system for deploying AI agents into real-world environments, complementing their model ecosystem with robust engineering infrastructure.

marsbit48 min fa

Just Now, DeepSeek Harness Shocks the Open Source World: Everything as a Plugin

marsbit48 min fa

Trading

Spot
活动图片