BTC $68,526.03 -6.94%
ETH $1,924.32 -6.04%
BNB $631.62 -6.57%
XRP $1.00 -3.86%
SOL $78.44 -5.61%
TRX $0.3106 -2.06%
DOGE $0.0633 -5.80%
ADA $0.1674 -2.89%
BCH $214.93 -6.28%
LINK $9.69 -4.20%
HYPE $70.56 -2.03%
AAVE $109.80 -7.22%
SUI $0.7687 -6.05%
XLM $0.1424 -2.78%
ZEC $899.44 -8.54%
AAPL $356.03 -0.02%
AMZN $267.97 +0.47%
GOOGL $299.60 +1.47%
MSFT $520.14 -0.22%
META $624.24 -0.83%
NVDA $164.45 +2.28%
TSLA $398.43 +3.03%
SNDK $922.09 -2.83%
INTC $60.84 +0.65%
SPCX $140.33 +0.57%
MU $793.04 +0.91%
AMD $574.48 +2.31%
BTC $68,526.03 -6.94%
ETH $1,924.32 -6.04%
BNB $631.62 -6.57%
XRP $1.00 -3.86%
SOL $78.44 -5.61%
TRX $0.3106 -2.06%
DOGE $0.0633 -5.80%
ADA $0.1674 -2.89%
BCH $214.93 -6.28%
LINK $9.69 -4.20%
HYPE $70.56 -2.03%
AAVE $109.80 -7.22%
SUI $0.7687 -6.05%
XLM $0.1424 -2.78%
ZEC $899.44 -8.54%
AAPL $356.03 -0.02%
AMZN $267.97 +0.47%
GOOGL $299.60 +1.47%
MSFT $520.14 -0.22%
META $624.24 -0.83%
NVDA $164.45 +2.28%
TSLA $398.43 +3.03%
SNDK $922.09 -2.83%
INTC $60.84 +0.65%
SPCX $140.33 +0.57%
MU $793.04 +0.91%
AMD $574.48 +2.31%

sea

All
Article
Flash

first_img Chinese AI company seeks to rent overseas NVIDIA computing power

Reuters reported on November 12, citing three informed sources, that Chinese artificial intelligence companies are attempting to rent Nvidia chip computing power from overseas data centers, concerned that the controversy surrounding Nvidia CEO Jensen Huang may prompt Washington to tighten restrictions on the remote use of advanced processors. The United States has restricted the export of Nvidia's most advanced artificial intelligence chips to China, and Chinese companies can currently use such processors remotely by renting computing power from overseas data centers. A large Chinese cloud service provider has internally discussed whether to secure overseas Nvidia chip capacity, and some companies have contacted data center operators to negotiate reservations. The informed sources declined to be named due to the sensitivity of the matter and did not specify the cloud service provider.Two informed sources stated that smaller artificial intelligence labs are also considering similar arrangements, including reaching out to major computing power rental service providers in Southeast Asia. These companies are concerned that the controversy surrounding Jensen Huang may provide political momentum for the U.S. Remote Access Security Act (RASA). This bill aims to authorize the U.S. Department of Commerce to regulate remote and cloud access to restricted technologies. The U.S. House of Representatives passed the bill in January with a vote of 369 in favor and 22 against, and it is currently awaiting Senate review. Reuters stated that Nvidia's advanced processors are used to train and run the large models behind generative artificial intelligence products like chatbots, and restricting overseas cloud access will make it more difficult for Chinese developers to obtain high-end computing power without directly importing chips.Jensen Huang stated on Sunday that he was unfaithful to his wife. The New York Times reported that the individual involved is Alice Huang, a former executive of Singapore cloud computing company Megaspeed, who U.S. authorities have identified as a Chinese intelligence operative. Reuters stated it was unable to independently verify the allegation, and the Chinese side has denied it. Two informed sources said that Alice Huang has business dealings and investment relationships with Chinese tech companies such as Tencent and Alibaba, which did not immediately respond to requests for comment. Business filings in Singapore reviewed by Reuters show that Alibaba subsidiary Apex Enterprise Solutions had provided financing to Malaysian company Speedmatrix, which is associated with Megaspeed. Megaspeed denied illegally transshipping Nvidia chips subject to U.S. export controls to China. Two other informed sources stated that traders and intermediaries have begun hoarding restricted Nvidia artificial intelligence chips in China's gray market, with buyers concerned that the controversy may prompt the U.S. to take broader action against the smuggling of advanced chips and artificial intelligence servers.

Binance Research: The cryptocurrency market lacks an actively managed discovery layer, with neutral strategies siphoning off over 90% of the funds

Binance Research's latest report indicates that while institutional funds have significantly entered the cryptocurrency market, they are primarily flowing into passive Beta products (such as ETFs), and the actively managed sector has not kept pace.The core reason is not a lack of talent or returns, but rather the absence of a middle layer that connects funds with managers for discovery and verification. Currently, the funding sources for active cryptocurrency strategies still heavily rely on personal networks and conferences, lacking a standardized performance verification and benchmarking system.Research shows that 80% of asset allocators prioritize asset safety over historical returns, and verified trading platform data can significantly enhance due diligence. Managers believe that unverifiable performance and lengthy compliance reviews are the biggest obstacles to fundraising.Capital Connect data shows that directional strategies account for 44.3% of the total number of funds but hold only 6.7% of the capital; market-neutral, statistical arbitrage, and traditional multi-strategy funds make up less than half but attract 92.1% of the capital. The average single investment size from institutions in neutral strategies is about 11 times that of directional strategies.In October, BTC fell by 7.2%, with directional funds averaging a positive return of 1.9%, while neutral strategies recorded 1.0%. Binance Research concludes that institutional preferences and the introduction of traditional structures are driving the market towards maturity, but allocators remain highly selective. Only a comprehensive verification system and operational infrastructure can truly unlock the inflow of actively managed funds.

Tencent launched an independent AI search website Yuanbao Search

Tencent officially launched an independent search website "Yuanbao Search" (so.qq.com), competing with Baidu and Google. The website defaults to AI search, allowing users to receive complete answers and reference sources after inputting questions, and supports follow-up inquiries. It also retains traditional web search and provides search entry points for public accounts, video accounts, images, and academic content.Currently, there have been instances where inputting keywords in Sogou Search directly redirects to the Yuanbao Search results page, and the latest version of QQ Browser also supports setting Yuanbao Search as the search engine. Tencent is gradually integrating existing search entry points into Yuanbao.The last self-developed search by Tencent dates back to "Soso," launched in 2006, when it brought in technical executives like Wu Jun from Google. In 2013, Tencent merged Soso's search business into Sogou, investing $448 million to acquire a 36.5% stake in Sogou; in 2021, it completed a full acquisition of Sogou. Starting in 2024, Tencent will promote Yuanbao AI Search, integrating WeChat Search and Sogou Search, and will open search capabilities to enterprises and developers in 2025. In October of this year, the service was renamed "Yuanbao Search," supporting API, MCP, and Skill integration. Official data shows that Yuanbao Search has an index of over 100 billion, with an average daily call volume of hundreds of millions, supporting more than 50 internal businesses such as Tencent Video, ima, and WorkBuddy.

Trip promotes the DRI system across the company to implement a major integration of production and research, breaking down functional boundaries

According to exclusive reports from Big Factory Daily, Trip.com held an all-hands meeting last week to announce the promotion of the "Integration of Production and Research" system across the company. This system is centered around DRI (Directly Responsible Individual) and merges personnel from product, operations, front-end, back-end, algorithms, testing, and other areas into project-based teams, breaking down traditional functional team boundaries with the goal of achieving "everyone is a full-stack developer, everyone is a product manager."In terms of specific execution, DRI leaders are not limited by function and can be original product, operations, or technical leaders. Team members within the project group collaborate across functions, which may lead to situations where products evaluate technical performance, and technology reports to operations. Front-end and back-end teams can also propose requirements and implement them independently. Ideas from product and operations that can be realized by AI can also be delivered independently, ultimately decided by the DRI.Currently, departments such as Trip.com Travel, Flights, and Content Center have begun to respond, with different teams progressing at varying paces, and the company has not imposed a one-size-fits-all requirement. The report indicates that this move is partly influenced by Silicon Valley concepts and aligns with the trend of blurred job boundaries in the AI era. This model can effectively reduce buck-passing and disputes, shorten delivery chains, and accelerate decision-making, but the challenge lies in the high comprehensive ability requirements for DRI leaders, who need to possess both technical understanding and business insight, and such qualified leaders are currently relatively scarce.
app_icon
ChainCatcher Building the Web3 world with innovations.