BTC $84,656.91 +0.01%
ETH $2,678.66 -0.92%
BNB $769.06 -0.20%
XRP $1.49 -0.07%
SOL $119.08 +0.27%
TRX $0.3348 +0.17%
DOGE $0.0932 -0.50%
ADA $0.2472 +0.82%
BCH $312.83 +2.88%
LINK $13.90 -2.16%
HYPE $89.15 +1.52%
AAVE $179.86 +3.51%
SUI $1.19 +1.91%
XLM $0.2157 -0.79%
ZEC $1,321.00 -0.65%
AAPL $333.19 +0.69%
AMZN $251.61 +1.07%
GOOGL $342.90 +0.98%
MSFT $517.12 +0.40%
META $727.06 +0.02%
NVDA $234.36 +1.20%
TSLA $371.00 +4.34%
SNDK $1,718.82 -3.58%
INTC $118.91 -1.39%
SPCX $158.91 +6.56%
MU $1,066.94 -2.30%
AMD $634.86 +2.50%
BTC $84,656.91 +0.01%
ETH $2,678.66 -0.92%
BNB $769.06 -0.20%
XRP $1.49 -0.07%
SOL $119.08 +0.27%
TRX $0.3348 +0.17%
DOGE $0.0932 -0.50%
ADA $0.2472 +0.82%
BCH $312.83 +2.88%
LINK $13.90 -2.16%
HYPE $89.15 +1.52%
AAVE $179.86 +3.51%
SUI $1.19 +1.91%
XLM $0.2157 -0.79%
ZEC $1,321.00 -0.65%
AAPL $333.19 +0.69%
AMZN $251.61 +1.07%
GOOGL $342.90 +0.98%
MSFT $517.12 +0.40%
META $727.06 +0.02%
NVDA $234.36 +1.20%
TSLA $371.00 +4.34%
SNDK $1,718.82 -3.58%
INTC $118.91 -1.39%
SPCX $158.91 +6.56%
MU $1,066.94 -2.30%
AMD $634.86 +2.50%

signed

All
Article
Flash

first_img DeepSeek publicly releases the Agent training system DSec, signed by Liang Wenfeng

According to Investment World citing Quantum Bit reports, DeepSeek has publicly disclosed the technical details of the system DSec (DeepSeek Elastic Compute) used for training Agents, authored by Liang Wenfeng. This system can generate over 5,000 sandboxes per second, reaching 3 million in a day, with a peak simultaneous operation of 380,000; supporting this scale is a single cluster with approximately 160 nodes, 30,000 CPU cores, and 250TB of memory.DSec prepares four types of backends for four categories of tasks: FnCall, Container, MicroVM, and Full VM, with the training side called through a unified Python SDK libdsec. The scheduling chain includes IAM, API Server, scheduling engine, node Edge, network proxy Aether, and components within the sandbox Chronus. The environment is divided into three layers of read-only images: base image, workspace, and toolkit, which are used in combination at startup. Runtime data from the paper shows that the actual read ratios of Python, Java, and C++ container images are approximately 6.0%, 9.2%, and 8.7%, respectively.Starting from DeepSeek-V4.1, the Agent loop has been moved to the DSec worker container, no longer bound to the GPU Pod lifecycle. The security section disclosed reward hacking during training, including actions such as overwriting system files, swapping file data blocks, scanning networks, and triggering kernel crashes. Defensive measures include AppArmor and eBPF-based network filtering, but reports indicate that these measures do not completely resolve the issues.

Anthropic signed at least 14.8GW of computing power in the past 11 months, with a potential cost of up to 517 billion USD

According to statistics from The Information, Anthropic has signed at least 14.8GW of computing power in the past 11 months, which can be gradually utilized in the coming years. Based on currently public contracts, the potential total cost could reach up to $517 billion, with most expenditures occurring over the next decade. In addition to the 1-2GW already secured before October last year, the total computing power signed by Anthropic is approximately 16GW.This round of expansion is primarily driven by the demand for Claude. This year, the growth of Claude Code and Cowork has exceeded Anthropic's expectations, prompting the company to focus on acquiring computing power. The new agreement with Amazon provides up to 5GW, while Google and Broadcom offer another 5GW, and Microsoft and NVIDIA provide approximately 1GW. Anthropic has also rented all computing power from SpaceX's Colossus 1, acquiring over 220,000 NVIDIA GPUs, including H100, H200, and GB200.Anthropic has secured about 16GW, with many contracts extending beyond 2030. OpenAI has set a target for investors to reach 30GW by 2030, expecting to invest approximately $750 billion in computing power by that year. The $517 billion figure is the potential maximum cost estimated by The Information based on existing cloud services, chip, and data center contracts, with some computing power potentially being delayed or unused, and some contracts allowing for early cancellation.

first_img Dubai VARA signed a memorandum of understanding with Securitize to advance tokenization innovation

According to Cointelegraph, the Dubai Virtual Assets Regulatory Authority (VARA) has signed a Memorandum of Understanding (MoU) with the BlackRock-backed tokenization platform Securitize to advance the tokenization and digital asset infrastructure development in the UAE and Dubai. Both parties stated that the agreement will establish a collaborative framework to support regulated tokenization initiatives, promote institutional participation, and strengthen Dubai's digital asset ecosystem.A VARA spokesperson stated that the memorandum aims to combine VARA's regulatory perspective with Securitize's experience in institutional tokenization to determine how the collaboration can help Dubai develop a trustworthy and regulated tokenization market, with no specific projects to be announced at this stage. Carlos Domingo, co-founder and CEO of Securitize, stated that Dubai is one of the most forward-looking jurisdictions for global digital asset innovation, and collaborating with regulators is crucial for tokenization to move from concept to mainstream financial infrastructure.At the time of this announcement, investor demand for tokenized assets continues to rise. According to RWA.xyz data, the total number of tokenized asset holders has increased by 103% over the past 30 days to 3.2 million, and the total value of tokenized assets has grown by 2% to $38.5 billion. Securitize is currently the largest tokenization platform globally, with a tokenized asset management scale of $4.9 billion, while Ondo Finance ranks second with $3.5 billion.
app_icon
ChainCatcher Building the Web3 world with innovations.