重点快讯
52 Posts
重点快讯
Open Trade
Frequent Trader
5 Years
630 Following
359 Followers
130 Liked
Posts
Portfolio
$NVDA OpenAi’s weak and ineffectual counterattack against the low-price strategy used to deal with China’s domestic open-source large language models.
With the performance gap in reasoning ability only 10–15%,
the API call price—
DeepSeek V4 flash—
needs only $0.4 per million tokens, whereas
GPT 5.6, across its various versions,
requires $7–35.
Under these circumstances, OpenAI’s future expected revenue will drop sharply,
and the existing OpenAI user base is also rushing to embrace China’s domestic LLMs.
With the performance gap in reasoning ability only 10–15%,
the API call price—
DeepSeek V4 flash—
needs only $0.4 per million tokens, whereas
GPT 5.6, across its various versions,
requires $7–35.
Under these circumstances, OpenAI’s future expected revenue will drop sharply,
and the existing OpenAI user base is also rushing to embrace China’s domestic LLMs.
$NVDA NVIDIA's AI myth is being shattered one after another by the successive releases of China's domestic open-source large models DeepSeek-V4-Pro and Kimi-K3.
When open-source model performance rivals GPT-5.5, yet API prices are only 2% of American competitors', why would enterprises still spend a fortune rushing to buy H100s? With training and inference costs plummeting off a cliff, the logic of brute-force computing power accumulation has already collapsed. Even Microsoft is researching replacing OpenAI with DeepSeek—who can NVIDIA still count on for its trillion-dollar orders?
More critically, the early development costs of closed-source models, which once ran into tens of billions of dollars, have instantly become unrecoverable sunk costs. Valuations of U.S. AI companies have evaporated by hundreds of billions, and capital expenditure cuts are imminent. And the CUDA ecosystem moat? The open-source community is bypassing it with optimizations and alternative solutions.
With valuations at 100 times earnings, the market narrative is shifting from "compute scarcity" to "cost efficiency." Once expectations reverse, a double blow is inevitable.
The AI dividend has peaked, and the bubble should burst.
When open-source model performance rivals GPT-5.5, yet API prices are only 2% of American competitors', why would enterprises still spend a fortune rushing to buy H100s? With training and inference costs plummeting off a cliff, the logic of brute-force computing power accumulation has already collapsed. Even Microsoft is researching replacing OpenAI with DeepSeek—who can NVIDIA still count on for its trillion-dollar orders?
More critically, the early development costs of closed-source models, which once ran into tens of billions of dollars, have instantly become unrecoverable sunk costs. Valuations of U.S. AI companies have evaporated by hundreds of billions, and capital expenditure cuts are imminent. And the CUDA ecosystem moat? The open-source community is bypassing it with optimizations and alternative solutions.
With valuations at 100 times earnings, the market narrative is shifting from "compute scarcity" to "cost efficiency." Once expectations reverse, a double blow is inevitable.
The AI dividend has peaked, and the bubble should burst.
$NVDA can we mount it with a jumper like this? It looks like it's about to make a big waterfall
$KOMA Can Shigou Zhuang go back for a round? With a market value of 6 million, how low can it fall?
$ACT now the moment this situation is taken down, once the delisting notice comes, it will crash and lure shorts with a fakeout, then surge violently, liquidating all the short positions—this dog of a broker is just too stingy with its shares.
$LIT doesn't look like it can rush up; the position is good, with a bit of room in the middle, in his hands
$ACT Hurry up and dump the order, you half-dead, half-alive. Air Force brothers, increase your efforts. The market cap is too high—take a look and drop to 3 million.
$ACT held on, brothers—at most let it drop to a 5 million market cap to cut losses. If it’s above 15 million market cap, find a spot to take profit.
$ACT hehe😁 time again for my entry price—going from a market cap of 8 million to 30 million isn’t too much to ask, right?
$ACT can’t even manage a 10 million market cap coin that’s listed on more than 20 exchanges—shouldn’t it be getting delisted soon?
$ACT has entered the observation period. At minimum, by the end of next month it has to be pulled down to 0.05 so we can continue to feed on the platform’s users. Otherwise, it’ll be taken down.
$POL Just wait until I hit the jackpot with this 10POL World Cup lottery ticket, then I'm going to dump the market. Brace yourselves, noobs!
$POL Why is it still dipping during the World Cup? The transaction fees for betting using POL are so cheap.
$ESPORTS has dropped below a $20 million market cap, it's about to get delisted, this year many platforms are likely to purge a hefty batch of shitcoins promoting US stock tokens.
Log in to explore more content