AI Weekly | OpenAI publicly opposes AI big model regulation bill; AMD plans to acquire server manufacturer for $4.9 billion
六月清晨搅
发表于 2024-8-25 11:10:56
182
0
0
AMD plans to acquire server manufacturer ZT Systems for $4.9 billion
Prior to the stock market on August 19th, AMD announced its intention to acquire server manufacturer ZT Systems for $4.9 billion through a combination of cash and stock. This is AMD's second major acquisition in recent times. AMD has just completed the acquisition of Silo AI, the largest private laboratory in Europe, for approximately $665 million. AMD stated that in the past 12 months, in addition to increasing research and development investment, the company has also invested over $1 billion to expand AMD's ecosystem and enhance its AI software capabilities.
Comment: The AI chip industry is sparking an investment or acquisition competition, and both AMD and Nvidia attach great importance to investment. According to publicly available information, Nvidia has invested in no less than 16 companies this year. During a conference call regarding AMD's acquisition of server manufacturer ZT Systems, AMD CEO Su Zifeng stated; quot; We must have the ability and expertise to optimize at the system rack and even data center level. Large scale clusters are becoming increasingly complex, which is why we are acquiring ZT Systems& amp;quot;
OpenAI publicly opposes AI big model regulation bill
OpenAI, an American artificial intelligence company, issued an open letter on Wednesday local time, opposing the draft of the Safe and Reliable Frontier Artificial Intelligence Model Innovation Act launched by California local legislators, saying that the draft may threaten California's future growth, slow down innovation, and drive out world-class engineers and entrepreneurs in California, and suggesting that the AI big model regulation act be launched at the federal level rather than the local level. OpenAI believes that if the bill is passed, it could have a "broad and significant" impact on the competitiveness of artificial intelligence and national security in the United States.
Comment: For new technologies, it is not easy for regulation to balance norms and incentives. Previously, AI godmother Li Feifei stated on social media that AI policies must encourage innovation, set appropriate restrictions, and mitigate the impact of these restrictions; Otherwise, the best outcome would be failure to achieve the goal, and in the worst case, it could lead to terrible or even unexpected consequences. Outside of California, legislation on artificial intelligence is also being pushed forward around the world. On August 1st of this year, the EU's Artificial Intelligence Act officially came into effect within the EU.
Tesla recruits high paying data collection operators to train Optimus
Recently, Tesla's latest recruitment has attracted widespread attention: Tesla is recruiting a "human teacher" for its Optimus humanoid robot, with the specific job title being Data Collection Operator. The job responsibilities of this position mainly involve wearing motion capture suits and VR helmets, performing specific actions such as standing and walking as required, in order to provide data support for the training of the Optimus robot. It is reported that the hourly wage for this position is as high as $48 (about 342 yuan), requiring employees to walk for more than 7 hours a day, carry devices weighing up to 30 pounds (about 13.61 kilograms), and wear VR headsets for long periods of time.
Comment: According to data from LinkedIn and other recruitment websites, Tesla has hired dozens of workers in the past year to train its humanoid robot Optimus using motion capture suits. In order to be similar to the Optimus, which is expected to be 5 feet 8 inches (about 1.72 meters) tall, applicants need to be between about 1.7 meters and 1.8 meters tall. Nvidia senior researcher Animesh Garg stated that Optimus may require millions of hours of data to be fully prepared to enter Tesla factories for work.
Australian Institute of Certified Public Accountants Survey: Mainland Chinese Enterprises Accelerate in Artificial Intelligence Applications
The latest survey results from CPA Australia show that the proportion of mainland Chinese companies applying artificial intelligence (AI) has steadily increased in the past three years. Meanwhile, the proportion of mainland Chinese companies expected to increase their AI applications in the next 12 months has also significantly increased. Nearly half (48%) of the surveyed accounting and finance professionals in mainland China expect their companies to increase their AI applications in the next 12 months, more than double the proportion of the same period last year (20%).
Comment: Jin Ke, Vice President of the East and Central China Committee of the Australian Institute of Certified Public Accountants and a senior registered accountant in Australia, stated that a series of supportive policies and measures, coupled with the rapid development of generative AI and big model technology (LLM), have created favorable conditions for Chinese mainland enterprises to apply AI. By applying AI, enterprises can better analyze data, improve decision-making, automate repetitive tasks to increase efficiency, and reduce costs. However, at the same time, the application of AI by mainland Chinese enterprises is still in its early stages. 72% of mainland Chinese respondents indicated that their companies have only applied AI to a limited or moderate extent. In addition, financial costs and low return on investment have been the main challenges faced by companies surveyed in mainland China and the Asia Pacific region in terms of technology applications over the past 12 months.
Meta cancels development plan for a high-end mixed reality headset
It is reported that Meta Platforms has cancelled a development plan for a high-end mixed reality headset aimed at competing with Apple Vision Pro. Meta employees stated that after a product review meeting this week, the company requested its Reality Labs department employees to stop developing this device. The internal code name of this discontinued device is La Jolla, originally planned to be released in 2027.
Comment: AR and VR devices are considered to be on the eve of commercialization. Meta has previously partnered with Ray Ban to launch the Ray Ban Meta smart glasses, which have sold over a million units. AR, VR, and even mixed reality headsets are ideal devices equipped with AI. Meta CEO Zuckerberg previously stated in a conversation with Nvidia CEO Renxun Huang that AI driven smart glasses will replace smartphones as the mobile version of the next generation computing platform. Meta's cancellation of its high-end mixed reality headset plan may be due to the lack of optimistic commercial returns for this high-end mixed reality headset in its product lineup.
Midjourney web version is open to all users
On August 23rd, Midjourney, an AI painting tool, announced on its official channel that in order to promote and popularize its powerful AI painting technology more widely, Midjourney has decided to officially open its web version usage rights to users worldwide for free from today. Regarding this free and open initiative, Midjourney officials stated that they hope to enable more people to easily experience the charm of AI painting, and inspire more creative inspiration and enthusiasm. At the same time, they also believe that this measure will help promote the popularization and development of AI painting technology, injecting new vitality into the entire industry.
Comment: The industry believes that Midjourney's move is also due to the threat of competition from new market forces. At the beginning of the month, the original team of Stable Diffusion, the leader of open-source image generation models, announced the launch of a new image generation model FLUX. 1, which performed exceptionally well. In addition, at the end of the month, Ideogram launched Ideogram 2.0, which flooded the AI painting circle. Midjourney's free web version may be aimed at further consolidating its market position and attracting more new users to join.
World Robot Conference opens, 27 humanoid robots showcase their skills
On August 21, 2024, the World Robot Conference (WRC 2024) was held in Beijing, with a duration of 5 days. As the largest robotics exhibition in China, this exhibition attracted 169 companies in the robotics field to participate. The most eye-catching event at WRC 2024 was the explosion of humanoid robot products, with a total of 27 humanoid robot manufacturers and over 30 upstream and downstream companies participating. In order to demonstrate their technological capabilities, various domestic humanoid robot companies are showcasing their robots by attempting to walk upright, kick a ball, and even being kicked from behind by humans.
Comment: From an industry perspective, humanoid robots are a relatively hot track this year, receiving a lot of attention from capital. Some attendees told reporters that this year is the busiest year since their participation, especially on weekends when venues are packed. However, humanoid robots are still in their early stages and there are still some problems. For example, some humanoid robots need to rest for a long time after display, and some robots need to be connected to long wires for display, which restricts their range of motion and may still have a long way to go before landing.
Kunlun Wanwei releases AI short drama platform SkyReels
On August 19th, Kunlun Wanwei released SkyReels, an AI short drama platform that integrates video and 3D models. According to the official introduction, SkyReels platform integrates script generation, character customization, storyboarding, plot, dialogue/BGM, and film synthesis. Creators can "create a drama with one click", and users can input a simple idea, and this tool can complete the entire process from script to finished short drama production. This platform integrates Kunlun Wanwei's self-developed script model SkyScript, self-developed storyboard model StoryboardGen, self-developed 3D generated model Sky3DGen, and innovative platform WorldEngine.
Comment: Short dramas have become one of the fastest-growing film and television fields in recent years, as well as a testing ground for new technologies. There have been many attempts at one click translation and AI face swapping technology, and several online literature companies have also released content generation models that can assist authors in writing. With the help of AI's large model video generation capability, the production time of short dramas that used to last three months now only takes half a month, and the short drama market will usher in a revolution.
Amazon plans to acquire Perceive, an edge AI inference company
Recently, Xperi signed an agreement with Amazon to acquire most of the assets and some of the debts of Xperi's subsidiary Perceive for $80 million in cash. Perceive focuses on edge AI inference hardware and software technology. This transaction is expected to be completed before the end of this year.
Comment: Amazon has made multiple investments in AI related companies this year and the investment is still ongoing. Previously, Amazon invested in basic model manufacturer Anthropic, AI data platform Scale AI, and biomedical artificial intelligence model company Evolutionary Scale. Amazon founder Bezos also invested in humanoid robot companies Figure AI and Skill AI, as well as AI search engine company Perplexity.
SK Hynix may collaborate with TSMC and Nvidia to develop the next generation HBM
It is reported that SK Hynix may announce a closer cooperation plan with TSMC and NVIDIA at the Semicon Taiwan in Taiwan, China, China, in September, and focus on the development of the next generation HBM (high bandwidth memory) technology to further master the key components of AI servers.
Comment: HBM is crucial for GPU chips used in the AI field. It has been reported that SK Hynix and TSMC have signed a memorandum of understanding to collaborate on the development of the sixth generation HBM product HBM4, which is expected to be put into production in 2026. SK Hynix, Samsung, and Micron are competing in the HBM field, and recently there have been reports that Samsung plans to launch a new generation of HBM memory chip by the end of this year.
CandyLake.com 系信息发布平台,仅提供信息存储空间服务。
声明:该文观点仅代表作者本人,本文不代表CandyLake.com立场,且不构成建议,请谨慎对待。
声明:该文观点仅代表作者本人,本文不代表CandyLake.com立场,且不构成建议,请谨慎对待。
猜你喜欢
- AI Weekly | AI search startup Perplexity valuation surges to $9 billion; Nvidia's market value exceeds Apple's
- Robin Lee: The average daily adjustment amount of Wenxin Model exceeded 1.5 billion, 30 times more than that of a year ago
- Will DeepMind's open-source biomolecule prediction model win the Nobel Prize and ignite a wave of AI pharmaceuticals?
- "AI new generation" big model manufacturer Qi "roll" agent, Robin Lee said that it will usher in an era of "making money by thinking"
- Robin Lee said that the illusion of the big model has basically eliminated the actual measurement of ERNIE Bot?
- AI Weekly | Yang Zhilin claims that Kimi has over 36 million monthly active users; Robin Lee: The illusion of big model is basically eliminated
- ERNIE Bot has more than 400 million users, Baidu Wu Tian: the big model is reshaping the industrial intelligence engine
- In October of this year, Tesla Model Y won the sales championship for first tier and new first tier city models
- Alibaba CEO Wu Yongming: AI development requires a batch of open-source models of different scales and fields
- Baidu's Q3 core net profit increased by 17%, exceeding expectations. Wenxin's large model daily usage reached 1.5 billion
-
知名做空机构香橼研究(Citron Research)周四(11月21日)在社交媒体平台X上发布消息称,该公司已决定做空“比特币大户”微策略(Microstrategy)这家公司,并认为该公司已经将自己变身成为一家比特币投资基金 ...
- caffycat
- 5 小时前
- 支持
- 反对
- 回复
- 收藏
-
每经AI快讯,11月20日,文远知行宣布旗下自动驾驶环卫车S6与无人扫路机S1分别在新加坡滨海湾海岸大道与滨海艺术中心正式投入运营。据介绍,这是新加坡首个商业化运营的自动驾驶环卫项目。 ...
- star8699
- 前天 19:48
- 支持
- 反对
- 回复
- 收藏
-
上证报中国证券网讯(记者王子霖)11月20日,斗鱼发布2024年第三季度未经审计的财务报告。本季度斗鱼依托丰富的游戏内容生态,充分发挥主播资源和新业务潜力,持续为用户提供高质量的直播内容及游戏服务,进一步 ...
- goodfriendboy
- 前天 20:09
- 支持
- 反对
- 回复
- 收藏
-
百度创始人李彦宏19日在2024中国5G+工业互联网大会上发表演讲时表示,多智能体协作应用无代码工具"秒哒"发布不到3天,超过5000家企业排队申请测试。此外,L4级端到端的自动驾驶大模型发布后,搭载百度这一大模 ...
- laozhucn
- 3 天前
- 支持
- 反对
- 回复
- 收藏