BTC $77,254.53 -1.49%
ETH $2,406.73 -2.30%
BNB $685.17 -0.94%
XRP $1.34 -2.24%
SOL $99.86 -3.34%
TRX $0.3217 -3.04%
DOGE $0.0810 -2.15%
ADA $0.1971 -1.63%
BCH $246.77 -0.21%
LINK $11.17 -1.68%
HYPE $82.96 -1.12%
AAVE $127.45 +2.60%
SUI $0.7205 -1.17%
XLM $0.1750 -1.38%
ZEC $832.02 -2.23%
BTC $77,254.53 -1.49%
ETH $2,406.73 -2.30%
BNB $685.17 -0.94%
XRP $1.34 -2.24%
SOL $99.86 -3.34%
TRX $0.3217 -3.04%
DOGE $0.0810 -2.15%
ADA $0.1971 -1.63%
BCH $246.77 -0.21%
LINK $11.17 -1.68%
HYPE $82.96 -1.12%
AAVE $127.45 +2.60%
SUI $0.7205 -1.17%
XLM $0.1750 -1.38%
ZEC $832.02 -2.23%

re

All
Article
Flash

first_img Google claims that the cost of AI server memory has exceeded 75%, promoting a dual-track strategy for software and hardware

The SEMICON Taiwan 2026 Memory Summit took place on the 1st, where Nikhil Cherian, Senior Director of Supply Chain Infrastructure at Google Cloud under Alphabet, pointed out that with the popularity of multimodal and mixed expert architectures, AI computation has shifted from being power-limited to memory-limited, with high-performance memory accounting for over 75% of the cost of AI server hardware bill of materials. In the face of capacity, bandwidth, and power consumption bottlenecks, Google is breaking through the AI memory bottleneck through a dual-track strategy of hardware offloading for inference and training, and lossless quantization software algorithms.Google adopts an offloading strategy in hardware architecture, launching TPU 8i for low-latency inference and TPU 8t specialized for large-scale training. The TPU 8i is equipped with 288 GB of high-bandwidth memory, with SRAM capacity on the chip increased threefold to 384 MiB, placing dynamic conversation states and key-value caches on the chip itself to achieve zero chip-off latency. The TPU 8t forms a super-large computing cluster with 9600 chips, achieving a shared pool of HBM at a scale of 2 PB, eliminating chip-off data transfer bottlenecks, along with TPU Direct Storage technology.Google has developed the training-free TurboQuant lossless quantization algorithm, compressing the key-value cache of large models from 32 bits to 3 bits, reducing memory usage by six times without loss of accuracy, resulting in an eightfold acceleration in attention computation, and integrating old-generation DRAM technology to extend the lifecycle of components.

first_img The core ARR of China's open-source large models is approximately 6-8 billion USD

Robonomics author FD published the China Open-Source LLM Tracker on September 1, 2026, stating that the total ARR of China's open-source large models is approximately $10-15 billion, with core LLM revenue around $6-8 billion after excluding ByteDance / Seedance. Growth has not relied on price wars; after DeepSeek raised prices by about 3-12 times, usage still increased, and the average price of the Zhipu API rose by 101% while token usage exceeded 40 times within the year. The best estimate for overseas revenue is approximately $2 billion.Zhipu MaaS ARR increased from about $250 million in March to an annualized monthly rate of about $1.6 billion in August, with a weekly annualized rate of $2 billion. August revenue has already surpassed its API revenue for the first half of the year; the API gross margin is 24.6%, and inference costs per token decreased by 80%. MiniMax ARR rose from about $100 million in December 2025 to over $800 million in annualized weekly revenue by August 2026, with B2B accounting for about 80%. Kimi ARR increased from about $100 million in March to about $300 million by mid-June. DeepSeek's revenue from January to July was approximately $70 million, with the latest ARR estimated at about $1.2 billion.ByteDance's annualized AI revenue is approximately $4 billion, of which Seedance accounts for about $2 billion.

BitMine splurges $126 million to buy ETH, Eole increases holdings in HYPE

According to BBX data, yesterday global listed companies disclosed the latest developments in digital asset treasury allocation, financial performance, and compliance infrastructure expansion. The core information is as follows:BitMine purchased 51,000 ETH, valued at approximately $126 million: The latest transaction records show that the listed mining company BitMine made a significant purchase of 51,000 Ethereum (ETH) on September 2 through cryptocurrency broker FalconX and digital asset custodian BitGo. Based on the price at the time of the transaction, the total value of this purchase is approximately $126 million.Japanese listed company Eole increased its holdings by 8,709 HYPE: Japanese listed company Eole disclosed its latest treasury movements. Between July 30 and August 31, the company increased its holdings in the secondary market by approximately 8,709.57 HYPE, investing about $730,000. After this increase, its total holdings of HYPE have risen to approximately 9,787.82.OSL Group's total revenue exceeded HKD 55.8 billion in the first half of the year, with payment business dominating: The stablecoin payment and trading platform OSL Group (00863.HK) announced its interim results for the six months ending June 30. The financial report shows that the company's total revenue in the first half reached HKD 55.813 billion (a year-on-year increase of 65.8%). Among them, stablecoin and digital asset payment businesses have become the absolute core, contributing revenue as high as HKD 49.083 billion, accounting for 88% of total revenue.Coincheck Group partners with DFNS to enter the Japanese institutional custody market: Dutch-listed digital asset service provider Coincheck Group announced a strategic partnership with institutional financial infrastructure provider DFNS. The two parties will focus on in-depth cooperation around digital asset wallet technology and custody services in Japan. DFNS's underlying technology will be deployed to its Japanese subsidiary Coincheck, Inc., aiming to provide safe and reliable institutional-level digital asset custody and trading execution services while meeting local stringent regulatory requirements.
app_icon
ChainCatcher Building the Web3 world with innovations.