7*24

Hot topic

Market

Activities

On June 7, the Korea News Agency reported that South Korean retail investors sold more than 1 trillion won worth of overseas stocks in the first week of June, continuing a trend of net selling that began in April. This scale far exceeds the 469 million dollars sold in April and is nearing the record of 940 million dollars for the entire month of May. If the selling continues until the end of the month, it will mark the first instance of three consecutive months of net selling since April to July 2023. Notably, this sell-off occurs against the backdrop of the three major US stock indices reaching new highs, with market expectations that retail funds may flow back to South Korea. The KOSPI index broke the historical 8000-point barrier at the end of May, led by semiconductor stocks. Korean retail investors are cashing out at high levels in the US stock market and returning to reallocate their assets domestically.

[Korean Retail Investors 'Escape the Peak' of US Stocks, Selling Over 1 Trillion Won in Overseas Stocks This Week]On June 7, the Korea News Agency reported that South Korean retail investors sold more than 1 trillion won worth of overseas stocks in the first week of June, continuing a trend of net selling that began in April. This scale far exceeds the 469 million dollars sold in April and is nearing the record of 940 million dollars for the entire month of May. If the selling continues until the end of the month, it will mark the first instance of three consecutive months of net selling since April to July 2023. Notably, this sell-off occurs against the backdrop of the three major US stock indices reaching new highs, with market expectations that retail funds may flow back to South Korea. The KOSPI index broke the historical 8000-point barrier at the end of May, led by semiconductor stocks. Korean retail investors are cashing out at high levels in the US stock market and returning to reallocate their assets domestically.

Korean Retail Investors 'Escape the Peak' of US Stocks, Selling Over 1 Trillion Won in Overseas Stocks This Week

According to monitoring by Dongcha Beating, following the concentrated release of 25 open-source weight models, including Nvidia's Nemotron 3 Ultra and Google's Gemma 4 12B, tech investor Chamath Palihapitiya pointed out that the capability gap between open-source and closed-source cutting-edge models is rapidly narrowing, but there remains a significant disparity in usage costs. For enterprises consuming 1 billion input and 1 billion output tokens monthly, the cost of GPT-5.5 Pro reaches $105,000, Claude Opus 4.8 costs $30,000, while DeepSeek R1 only requires $2,740, making the premium for GPT-5.5 Pro nearly 40 times higher. Chamath warned that most CEOs are unaware that their technical teams are defaulting to the most expensive large model APIs without proper governance and auditing, leading to budget overruns. As model routing controls, such as those from Software Factory, an AI-native software development lifecycle orchestration platform co-founded by him, become more prevalent, enterprises will shift towards model-agnostic architectures, defaulting large-scale inference to DeepSeek, directing high-end proxies to Claude Opus, and only calling GPT-5.5 Pro on demand when clear high incremental value is generated. Chamath predicts that refined routing will significantly slow the revenue growth of APIs from cutting-edge labs like OpenAI and Anthropic, while revenues from the open-source and low-cost inference ecosystem will surge.

[Open Source vs. Closed Source 'Money Grab': Price Gap Reaches 40 Times, Open Source Models Diverting Billions in Revenue from Industry Giants]According to monitoring by Dongcha Beating, following the concentrated release of 25 open-source weight models, including Nvidia's Nemotron 3 Ultra and Google's Gemma 4 12B, tech investor Chamath Palihapitiya pointed out that the capability gap between open-source and closed-source cutting-edge models is rapidly narrowing, but there remains a significant disparity in usage costs. For enterprises consuming 1 billion input and 1 billion output tokens monthly, the cost of GPT-5.5 Pro reaches $105,000, Claude Opus 4.8 costs $30,000, while DeepSeek R1 only requires $2,740, making the premium for GPT-5.5 Pro nearly 40 times higher. Chamath warned that most CEOs are unaware that their technical teams are defaulting to the most expensive large model APIs without proper governance and auditing, leading to budget overruns. As model routing controls, such as those from Software Factory, an AI-native software development lifecycle orchestration platform co-founded by him, become more prevalent, enterprises will shift towards model-agnostic architectures, defaulting large-scale inference to DeepSeek, directing high-end proxies to Claude Opus, and only calling GPT-5.5 Pro on demand when clear high incremental value is generated. Chamath predicts that refined routing will significantly slow the revenue growth of APIs from cutting-edge labs like OpenAI and Anthropic, while revenues from the open-source and low-cost inference ecosystem will surge.

Open Source vs. Closed Source 'Money Grab': Price Gap Reaches 40 Times, Open Source Models Diverting Billions in Revenue from Industry Giants

On June 7, according to the Financial Times, OpenAI is preparing for the largest upgrade of ChatGPT since its launch. The $850 billion company is looking for new growth engines ahead of its planned IPO this year. The company intends to transform the chatbot into a 'super app' by integrating programming tools with AI, adding several new products that executives believe will generate higher revenue. According to more than a dozen current and former employees, these adjustments are part of OpenAI's overall restructuring plan. OpenAI is reallocating resources to aggressively pursue high-profit enterprise clients and is gearing up for intensified competition with rival Anthropic. This adjustment will elevate the status of OpenAI's programming product Codex and increase resource investment, reflecting a growing consensus within the company that the future of AI is not just chatbots that answer questions, but intelligent agents that can perform tasks for users.

[OpenAI Plans Major Upgrade for ChatGPT to Generate More Revenue]On June 7, according to the Financial Times, OpenAI is preparing for the largest upgrade of ChatGPT since its launch. The $850 billion company is looking for new growth engines ahead of its planned IPO this year. The company intends to transform the chatbot into a 'super app' by integrating programming tools with AI, adding several new products that executives believe will generate higher revenue. According to more than a dozen current and former employees, these adjustments are part of OpenAI's overall restructuring plan. OpenAI is reallocating resources to aggressively pursue high-profit enterprise clients and is gearing up for intensified competition with rival Anthropic. This adjustment will elevate the status of OpenAI's programming product Codex and increase resource investment, reflecting a growing consensus within the company that the future of AI is not just chatbots that answer questions, but intelligent agents that can perform tasks for users.

OpenAI Plans Major Upgrade for ChatGPT to Generate More Revenue

HTX data monitoring shows that BTC has broken through 62000 USDT, currently reported at 62055.61 USDT, with a 24H increase of 3.27%.

【BTC breaks through 62000 USDT】HTX data monitoring shows that BTC has broken through 62000 USDT, currently reported at 62055.61 USDT, with a 24H increase of 3.27%.

BTC breaks through 62000 USDT

On June 7, China Galaxy's macro research report stated that strong non-farm data does not mean the Federal Reserve will raise interest rates this year. Overall, the labor market data for May indicates that employment conditions are still better than market expectations, and there is a lack of reasons and data to support a rate cut in the short term. However, the structure still shows that there is no clear risk of acceleration in the labor market, nor will it lead to a spiral of inflation, so there is no need to excessively price in the risk of a rate hike this year. Although the non-farm data has been relatively strong since March, in the short term, it only raises the threshold for a rate cut by the Fed this year and does not create pressure for a rate hike. Overall, market concerns about rate hikes have become the main theme of trading. While China Galaxy Macro tends to believe that the market has overly priced in the risk of rate hikes, the short-term lagging economic data cannot disprove the expectations for rate hikes, and the Fed can only maintain a 'data-dependent' approach. Currently, the market faces a period of correction risk following extreme pricing in AI, and a recovery in liquidity expectations is expected to occur within the year, but it requires waiting.

[China Galaxy Macro: Strong Non-Farm Data Does Not Imply Fed Rate Hike This Year]On June 7, China Galaxy's macro research report stated that strong non-farm data does not mean the Federal Reserve will raise interest rates this year. Overall, the labor market data for May indicates that employment conditions are still better than market expectations, and there is a lack of reasons and data to support a rate cut in the short term. However, the structure still shows that there is no clear risk of acceleration in the labor market, nor will it lead to a spiral of inflation, so there is no need to excessively price in the risk of a rate hike this year. Although the non-farm data has been relatively strong since March, in the short term, it only raises the threshold for a rate cut by the Fed this year and does not create pressure for a rate hike. Overall, market concerns about rate hikes have become the main theme of trading. While China Galaxy Macro tends to believe that the market has overly priced in the risk of rate hikes, the short-term lagging economic data cannot disprove the expectations for rate hikes, and the Fed can only maintain a 'data-dependent' approach. Currently, the market faces a period of correction risk following extreme pricing in AI, and a recovery in liquidity expectations is expected to occur within the year, but it requires waiting.

China Galaxy Macro: Strong Non-Farm Data Does Not Imply Fed Rate Hike This Year

On June 7, analyst @alicharts noted that the best risk-reward opportunities for Bitcoin typically occur when the MVRV (Market Value to Realized Value ratio) pricing bands drop to levels of 1.0 and 0.8. Currently, these two key pricing bands correspond to $53,900 and $43,130, respectively, indicating that if BTC retraces to the aforementioned range, it may provide a favorable entry window.

[Analysis: MVRV Indicator Shows Optimal BTC Entry Point Below $53,900]On June 7, analyst @alicharts noted that the best risk-reward opportunities for Bitcoin typically occur when the MVRV (Market Value to Realized Value ratio) pricing bands drop to levels of 1.0 and 0.8. Currently, these two key pricing bands correspond to $53,900 and $43,130, respectively, indicating that if BTC retraces to the aforementioned range, it may provide a favorable entry window.

Analysis: MVRV Indicator Shows Optimal BTC Entry Point Below $53,900

On June 7, 'New Investment Guru' Serenity responded to how beginners should learn to invest and systematically elaborated on his investment framework. Serenity admitted that his personal style is somewhat different, essentially making independent judgments based on information not yet grasped by the market, combined with the accumulation of life experiences. 'A lot of things are about guessing those unstructured relationships and then waiting to see if you're right. This ability is hard to teach through courses; it's more about accumulating life skills and then applying them to the market.' Serenity reviewed this method through two classic cases. The first is Raspberry Pi: while the market generally viewed RPI as an educational toy, he observed that after the rise of OpenClaw (an open-source AI agent framework), many friends began purchasing Raspberry Pis and Mac Minis to deploy AI applications, and numerous AI orchestration video tutorials appeared online. He concluded that AI would become an ideal growth engine for RPI—his private modeling predicted a revenue growth rate of about 55%, and the final financial report revealed 58%, far exceeding the market's expectation of 14%. 'At that time, the media called it a meme stock because public channels couldn't see the revenue growth brought by AI.' The second case is AXT: he was mocked for buying in at around $12, with large language model illusions even asserting that large-scale cloud vendors and governments should have discovered and fixed the indium phosphide substrate flaw. Analysts estimated AXT was overvalued using steady-state TAM estimates. 'But AXT controls about 40% of the InP supply chain; without it, the entire chain would break. What you need to guess is—if it becomes a bottleneck like NAND, what market value can it achieve based on its control, and what price will buyers assign to it?' Goldman Sachs' current research conclusions and the financial reports of substrate epitaxy companies confirmed his AXT analysis afterward. Serenity also acknowledged that not all investments need to be this complex, 'Many stocks can be evaluated using standard methods.' For example, if AAOI's annualized revenue guidance for the first half of 2027 reaches $471 million, while its current market value is only $12 billion, it may be undervalued; Samsung Electronics is even simpler, looking at the market's modeling of operating profits for 2027 and 2028 to judge whether the current valuation is reasonable. For the harder-to-value JBL, its 1.6T LRO (Linear Receiving Optical) currently lacks clear sales data, 'We can only guess how widespread it is and then deduce its impact on the current market value.' Serenity concluded that what he essentially does is piece together high-confidence inferences from unrelated fragments, 'Of course, it could be wrong at any time.'

[New Investment Guru Serenity Discusses Personal Investment Framework]On June 7, 'New Investment Guru' Serenity responded to how beginners should learn to invest and systematically elaborated on his investment framework. Serenity admitted that his personal style is somewhat different, essentially making independent judgments based on information not yet grasped by the market, combined with the accumulation of life experiences. 'A lot of things are about guessing those unstructured relationships and then waiting to see if you're right. This ability is hard to teach through courses; it's more about accumulating life skills and then applying them to the market.' Serenity reviewed this method through two classic cases. The first is Raspberry Pi: while the market generally viewed RPI as an educational toy, he observed that after the rise of OpenClaw (an open-source AI agent framework), many friends began purchasing Raspberry Pis and Mac Minis to deploy AI applications, and numerous AI orchestration video tutorials appeared online. He concluded that AI would become an ideal growth engine for RPI—his private modeling predicted a revenue growth rate of about 55%, and the final financial report revealed 58%, far exceeding the market's expectation of 14%. 'At that time, the media called it a meme stock because public channels couldn't see the revenue growth brought by AI.' The second case is AXT: he was mocked for buying in at around $12, with large language model illusions even asserting that large-scale cloud vendors and governments should have discovered and fixed the indium phosphide substrate flaw. Analysts estimated AXT was overvalued using steady-state TAM estimates. 'But AXT controls about 40% of the InP supply chain; without it, the entire chain would break. What you need to guess is—if it becomes a bottleneck like NAND, what market value can it achieve based on its control, and what price will buyers assign to it?' Goldman Sachs' current research conclusions and the financial reports of substrate epitaxy companies confirmed his AXT analysis afterward. Serenity also acknowledged that not all investments need to be this complex, 'Many stocks can be evaluated using standard methods.' For example, if AAOI's annualized revenue guidance for the first half of 2027 reaches $471 million, while its current market value is only $12 billion, it may be undervalued; Samsung Electronics is even simpler, looking at the market's modeling of operating profits for 2027 and 2028 to judge whether the current valuation is reasonable. For the harder-to-value JBL, its 1.6T LRO (Linear Receiving Optical) currently lacks clear sales data, 'We can only guess how widespread it is and then deduce its impact on the current market value.' Serenity concluded that what he essentially does is piece together high-confidence inferences from unrelated fragments, 'Of course, it could be wrong at any time.'

New Investment Guru Serenity Discusses Personal Investment Framework

On June 7, OpenAI is preparing for the largest overhaul of ChatGPT since its launch. The company plans to transform ChatGPT into a 'super app' that combines coding tools and AI capabilities, adding products that executives believe can generate more revenue. (Financial Times)

[OpenAI Prepares Largest Overhaul of ChatGPT Since Launch]On June 7, OpenAI is preparing for the largest overhaul of ChatGPT since its launch. The company plans to transform ChatGPT into a 'super app' that combines coding tools and AI capabilities, adding products that executives believe can generate more revenue. (Financial Times)

OpenAI Prepares Largest Overhaul of ChatGPT Since Launch

According to monitoring by Dongcha Beating, Xiaohongshu's hi lab has open-sourced the end-to-end autoregressive text-to-speech (TTS) model dots.tts, which has 2 billion parameters, and has publicly released the complete inference and fine-tuning code under the Apache 2.0 license. The released weights include a base pre-trained version, a self-correcting alignment (SCA) fine-tuned version, and a low-latency inference distilled version. Unlike traditional TTS architectures that rely on discrete audio codec tokens (such as VALL-E, CosyVoice, ChatTTS, etc.), dots.tts implements a fully continuous, end-to-end autoregressive flow matching architecture that does not use any discrete tokens throughout the pipeline. Dots.tts combines continuous features extracted from AudioVAE at a 48 kHz sampling rate with a semantic encoder, a backbone language model (initialized from Qwen2.5-1.5B-Base, directly processing BPE text without the need for pinyin input), and an autoregressive flow matching acoustic head to predict continuous latent variables, which are reconstructed into audio by the generator. By directly predicting continuous features, dots.tts avoids the quality loss caused by discrete quantization, preserving pronunciation details, timbre similarity, and emotional expressiveness. Dots.tts is pre-trained on approximately 1.5 million hours of speech data. In the Seed-TTS-Eval evaluation, dots.tts achieved word error rates (WER) of 0.94% / 1.30% / 6.60% on Chinese, English, and Chinese difficult test sets, respectively, with similarity scores (SIM) of 81.0 / 77.1 / 79.5, reaching open-source SOTA levels. In the MiniMax Multilingual benchmark across 24 languages, the average speaker similarity reached 83.9. Xiaohongshu has provided a Gradio experience space on Hugging Face for users to test zero-shot voice cloning online.

[Xiaohongshu Releases Open Source Continuous Autoregressive TTS Model dots.tts Supporting Zero-Shot Voice Cloning]According to monitoring by Dongcha Beating, Xiaohongshu's hi lab has open-sourced the end-to-end autoregressive text-to-speech (TTS) model dots.tts, which has 2 billion parameters, and has publicly released the complete inference and fine-tuning code under the Apache 2.0 license. The released weights include a base pre-trained version, a self-correcting alignment (SCA) fine-tuned version, and a low-latency inference distilled version. Unlike traditional TTS architectures that rely on discrete audio codec tokens (such as VALL-E, CosyVoice, ChatTTS, etc.), dots.tts implements a fully continuous, end-to-end autoregressive flow matching architecture that does not use any discrete tokens throughout the pipeline. Dots.tts combines continuous features extracted from AudioVAE at a 48 kHz sampling rate with a semantic encoder, a backbone language model (initialized from Qwen2.5-1.5B-Base, directly processing BPE text without the need for pinyin input), and an autoregressive flow matching acoustic head to predict continuous latent variables, which are reconstructed into audio by the generator. By directly predicting continuous features, dots.tts avoids the quality loss caused by discrete quantization, preserving pronunciation details, timbre similarity, and emotional expressiveness. Dots.tts is pre-trained on approximately 1.5 million hours of speech data. In the Seed-TTS-Eval evaluation, dots.tts achieved word error rates (WER) of 0.94% / 1.30% / 6.60% on Chinese, English, and Chinese difficult test sets, respectively, with similarity scores (SIM) of 81.0 / 77.1 / 79.5, reaching open-source SOTA levels. In the MiniMax Multilingual benchmark across 24 languages, the average speaker similarity reached 83.9. Xiaohongshu has provided a Gradio experience space on Hugging Face for users to test zero-shot voice cloning online.

Xiaohongshu Releases Open Source Continuous Autoregressive TTS Model dots.tts Supporting Zero-Shot Voice Cloning

Today marks the 14th anniversary of Satoshi Nakamoto's first Bitcoin transaction. Bitcoin creator Satoshi Nakamoto executed the first Bitcoin transaction on January 12, 2009, sending about 10 BTC to early BTC user Hal Finney. Additionally, the first Bitcoin-related tweet was posted on January 11, 2009, written by Hal Finney, and read: Running Bitcoin.

Detail

【Exactly 14 years since Satoshi Nakamoto's first Bitcoin transaction】Today marks the 14th anniversary of Satoshi Nakamoto's first Bitcoin transaction. Bitcoin creator Satoshi Nakamoto executed the first Bitcoin transaction on January 12, 2009, sending about 10 BTC to early BTC user Hal Finney. Additionally, the first Bitcoin-related tweet was posted on January 11, 2009, written by Hal Finney, and read: Running Bitcoin.

Exactly 14 years since Satoshi Nakamoto's first Bitcoin transaction

A $300 Used Mining Rig Mined a $230,000 Bitcoin Block

The End of the 'Bitcoin Hoarding Myth': MicroStrategy Sells BTC for the First Time in 3 Years

Selling 32 BTC to Pay Dividends: Strategy Breaks the 'Never Sell' Vow

Another Bitcoin Buy Ahead? Michael Saylor’s Latest Post Fuels Rumors

Strategy Ends Bitcoin Buying Streak With Rare Sale—How Much Did It Sell?

AMD Stock Price Forecast 2026 to 2030: Wall Street Signals a Major $657 Target!

Over $1 Billion in Cryptocurrencies Forced to Liquidate, Bitcoin Plunges 6% Below $67,000

How To Avoid The Major Trap That Bitcoin Is Setting Up For Traders

Bitcoin Whales Most Active In Six Weeks As BTC Drops Under $67,000

Bulls Approach the Ceiling

Article

Bitwise CEO: A New Cycle Has Begun, the Crypto Market is Experiencing 'Renewal' This Year

Exactly 14 years since Satoshi Nakamoto's first Bitcoin transaction

Penafian：Konten diatas tidak merepresentasikan posisi HTX.，HTX tidak memberikan rekomendasi perdagangan apa pun.。

Artikel Terkait

Semua Komentar0TerkiniHangat

Topik Trending

Artikel Terkait