BTC $73,949.21 -0.43%
ETH $2,054.67 -1.70%
BNB $680.81 -1.00%
XRP $1.05 -2.08%
SOL $83.85 -0.84%
TRX $0.3171 -0.53%
DOGE $0.0680 -1.76%
ADA $0.1761 -0.97%
BCH $229.46 -2.67%
LINK $10.36 -0.78%
HYPE $72.36 -1.87%
AAVE $123.31 -0.92%
SUI $0.8293 -0.09%
XLM $0.1477 +1.19%
ZEC $990.66 -13.42%
AAPL $346.86 -1.30%
AMZN $271.52 +1.23%
GOOGL $291.08 +0.14%
MSFT $535.44 -1.55%
META $647.24 +0.24%
NVDA $175.99 -1.70%
TSLA $386.40 +2.75%
SNDK $928.28 +5.79%
INTC $61.68 +5.71%
SPCX $147.56 +4.59%
MU $771.95 +5.96%
AMD $548.50 +8.81%
BTC $73,949.21 -0.43%
ETH $2,054.67 -1.70%
BNB $680.81 -1.00%
XRP $1.05 -2.08%
SOL $83.85 -0.84%
TRX $0.3171 -0.53%
DOGE $0.0680 -1.76%
ADA $0.1761 -0.97%
BCH $229.46 -2.67%
LINK $10.36 -0.78%
HYPE $72.36 -1.87%
AAVE $123.31 -0.92%
SUI $0.8293 -0.09%
XLM $0.1477 +1.19%
ZEC $990.66 -13.42%
AAPL $346.86 -1.30%
AMZN $271.52 +1.23%
GOOGL $291.08 +0.14%
MSFT $535.44 -1.55%
META $647.24 +0.24%
NVDA $175.99 -1.70%
TSLA $386.40 +2.75%
SNDK $928.28 +5.79%
INTC $61.68 +5.71%
SPCX $147.56 +4.59%
MU $771.95 +5.96%
AMD $548.50 +8.81%

ppi

All
Article
Flash

Serenity: Investors are indiscriminately selling off assets in the AI industry chain, with some storage stocks' valuations dropping to an annualized price-to-earnings ratio of 1.6 times

Investor Serenity (@aleabitoreddit) posted that the current market deleveraging has become quite severe, with investors indiscriminately selling stocks related to the AI industry chain. Some companies' stock prices have even fallen back to levels before significant orders and performance expectations were realized.Serenity cited the storage chip company ESMT as an example, noting that the company has a market capitalization of about $2 billion, achieving a gross profit of $526.2 million and a net profit of $330 million in the third quarter, with net assets of approximately $937 million, of which net cash is about $740 million. Based on current performance annualized, its price-to-earnings ratio is only about 1.6 times. Despite ongoing market concerns about the cycle peaking, unsustainable profits, and lack of competitive barriers, the trend of DDR2 and DDR3 production lines exiting is tightening supply, and ESMT expects to hedge against declining consumer demand in the fourth quarter by raising product average prices.Serenity also pointed out that manufacturers like Samsung and Micron are gradually exiting DDR2 and DDR3 production lines, and Nanya Technology has also stopped related production lines, while ESMT's revenue mainly comes from these two types of storage products. It believes that the market underestimates the price support brought by the contraction of traditional storage supply and ignores the situation where Micron, Samsung, SK Hynix, and SanDisk are locking in profitability through long-term supply agreements and price floors. Serenity criticized that the market is selling off AI and storage industry chain companies that actually generate substantial cash flow while shifting funds to companies like Walmart, which has a price-to-earnings ratio of about 47 times, and CrowdStrike, which has a forward price-to-earnings ratio of about 180 times. The current focus is not on the actual free cash flow generated by companies but rather on indiscriminate selling.

Anthropic applies zero-knowledge proof technology to AI reasoning to prevent model swapping

Anthropic has open-sourced FlashProve, bringing the commonly used zkSNARK (zero-knowledge proof) technology from the cryptocurrency field into AI model inference.This technology was originally designed to prove the validity of blockchain transactions without disclosing transaction details. Anthropic aims to use it to solve another problem: when users pay a high price for an AI model's API, how can they know if the service provider has secretly switched to a cheaper model?FlashProve allows service providers to first generate encrypted fingerprints for the model weights and then provide mathematical proof for the inference results. Users do not need to obtain the model weights or rerun the model; they can check whether the specified output comes from the previously committed set of weights. Claude was also involved in the development and optimization of this system.Anthropic has already completed experiments on Meta's 400 billion parameter model Llama 4 Maverick. For every 1024 tokens of context processed, the generated proof is about 19 MB, and verification on a regular CPU takes about 17 seconds.However, currently, each proof can only verify one output token and cannot validate an entire response. The GPU computation time required to generate the proof is also about 70 to 95 times that of regular inference. Whether it can be used to identify API service providers that secretly switch models in the future will depend on when the cost of proof generation can be reduced.
app_icon
ChainCatcher Building the Web3 world with innovations.