Multiple sources revealed that OpenAI's annualized revenue has reached 1.3 billion US dollars. His remarks suggest that the company's monthly revenue exceeds 100 million US dollars, compared to last year's revenue of only 28 million US dollars. OpenAI's CEO believes that those who rely on ChatGPT for quick profits will not last long, and blind fundraising is very unwise.
Image Source: Generated by Wujie AI
Financing
Lambda Labs Completes Nearly 300 Million US Dollars in Financing
According to the "Science and Technology Innovation Board Daily" citing The Information, two informed sources revealed that cloud service startup Lambda Labs is competing with Amazon Web Services (AWS) to rent servers equipped with NVIDIA chips to AI developers. The company recently completed nearly 300 million US dollars in equity financing. One of the sources stated that billionaire Thomas Tull's investment fund plans to provide most of the funds for this round of financing. It is reported that,
Orbem Completes 30 Million Euros in Series A Financing, Using MRI+AI Technology to Determine Egg Gender
According to the webmaster's home on October 11th, Munich startup Orbem uses magnetic resonance imaging (MRI) + AI technology to scan and determine egg gender within 1 second. The company recently completed a round of 30 million euros in Series A financing, led by 83 North, with La Famiglia and early investors The Venture Collective, Possible Ventures participating. It is reported that Germany, France, and Italy legislated last year to ban the killing of male chicks and encourage technology companies to develop methods to determine the gender of chicks.
AI Startup Cleanlab Raises 25 Million US Dollars
According to Forbes on October 10th, the startup company Cleanlab, co-founded by three PhDs from MIT, announced the completion of a 25 million US dollars Series A financing, co-led by Menlo Ventures and TQ Ventures. Cleanlab is valued at 1 billion US dollars. Also joining this round of financing and collaborating with them is Databricks. Earlier this year, Databricks used Cleanlab to fine-tune the OpenAI Davinci model provided by the API and found that the error rate was reduced by 37% without adding any resources, and the test accuracy increased from 65% to 78%.
AI Startup FluidAI Medical Completes 15 Million US Dollars in Series A Financing
According to PRNewswire, on October 10th, artificial intelligence (AI) startup FluidAI Medical (FluidAI) announced that it has completed a 15 million US dollars Series A financing, led by SOSV and Graphene Ventures, with participation from Boutique Venture Partners, Threshold Impact, Cur8 Capital, Leva Capital, Kyto Technology & Life Science Inc., Juno Pharmaceuticals, and others. This round of financing will be used for the development of AI-driven postoperative monitors.
"Jinjingyunhua" Completes Nearly 100 Million Yuan in Series B Financing
According to 36Kr on October 16th, Beijing Jinjingyunhua Technology Co., Ltd. recently completed nearly 100 million yuan in Series B financing, with investment from Qian Investment and Yongxin Zhicheng. The funds from this round of financing will mainly be used to develop large-scale products in the network security industry, expand the research and development team, and expand the product market.
AI Recruitment Solution Provider Moonhub AI Completes 10 Million US Dollars in Seed Financing
According to IT Orange, the AI recruitment solution provider Moonhub AI recently received 10 million US dollars in seed funding from Khosla Ventures and GV, with other investors including Time Ventures, Day One Ventures, as well as angel investors such as Susan Wojcicki (former CEO of YouTube) and Mike Volpi (general partner of Index Ventures). It is reported that Moonhub AI mainly combines artificial intelligence and human resources recruitment expertise to help employers hire and manage high-quality talents, and is committed to providing recruitment management solutions for users.
Anysphere Receives 8 Million US Dollars Investment from OpenAI
According to TechCrunch, on October 12th, AI startup Anysphere announced that it has received an 8 million US dollars seed fund from OpenAI's Startup Fund, with angel investors including Nat Friedman (former CEO of GitHub), Arash Ferdowsi (co-founder of Dropbox), and others also participating in this financing.
Shanmu, a Shenzhen AI Health Management Robot Developer, Completes Tens of Millions in Strategic Financing
According to Babite, Shenzhen Shanmu Biotechnology Co., Ltd. (referred to as "Shanmu") recently announced the completion of tens of millions in strategic financing, with the current round of investment from Yunjing Intelligent. "Shanmu" was established in 2020 and is a micro-family AI health management robot developer. Its main product is a micro medical robot developed based on self-developed digital microfluidic chipsets and AI algorithms.
Large Models
Insiders: OpenAI's Revenue Calculated on an Annualized Basis Exceeds 1.3 Billion US Dollars
According to The Information on October 12th, multiple sources revealed that OpenAI CEO Sam Altman told employees this week that the company's annualized revenue has reached 1.3 billion US dollars. His remarks suggest that the company's monthly revenue exceeds 100 million US dollars, a 30% increase from this summer. Since the company launched the paid version of ChatGPT in February this year, this revenue growth (mainly from subscriptions to its conversational chatbots) has seen significant growth. Last year, the company's revenue was only 28 million US dollars for the entire year.
Moonshot AI Launches First Intelligent Assistant Product Kimi Chat Supporting Input of 200,000 Chinese Characters
According to Interface News on October 9th, large model startup Moonshot AI announced a breakthrough in the "long text" field, launching the first intelligent assistant product Kimi Chat that supports input of 200,000 Chinese characters. 200,000 Chinese characters is the longest context input length that can be productively used in the global market. Currently, the input length of most large language models worldwide is within 2K. Taking an input length of 2048 as an example, it means that the model accepts a token count of 2048 at a time, and the general token-to-word ratio is about 75%, meaning that a 2K model can generally only support an input of 2048*0.75=1500 English words.
Zhipu AI and Tsinghua KEG Release and Open-Source Multimodal Large Model CogVLM-17B
According to DoNews on October 12th, Zhipu AI and Tsinghua KEG recently released a new visual language base model called CogVLM. CogVLM achieves deep fusion of visual language features without sacrificing any NLP task performance. Zhipu AI has now open-sourced CogVLM-17B and provided fine-tuning code that can run on a single 3090 server for research and use. It is reported that CogVLM-17B is currently the top-performing model on the authoritative multimodal academic rankings, achieving state-of-the-art or second place results on 14 datasets.
Intelligence Research Institute: Wudao Tianying Aquila Large Language Model Series Fully Upgraded, Adds New 340 Billion Parameter Aquila2-34B
According to Babite on October 12th, the Intelligence Research Institute announced that the Wudao Tianying Aquila large language model series has been fully upgraded to Aquila2, and a new 340 billion parameter model, Aquila2-34B, has been added. It is reported that this new model has performed well in inference, generalization, and has achieved a series of results in intelligent agents, code generation, literature retrieval, and other scenarios.
SenseTime Upgrades and Releases Medical Health Large Model "Big Doctor," Supports High-Quality Training and Low-Threshold Deployment
According to the webmaster's home on October 16th, SenseTime Technology has released an upgraded version of the medical health large language model "Big Doctor" and is officially providing services to upstream and downstream institutions in the medical health industry. The model includes 13 preset scenarios, supports personalized functions, high-quality training, and low-threshold deployment.
Meitu Corporation Releases Meitu AI Vision Large Model MiracleVision 3.0
According to Babite on October 9th, Meitu Corporation announced the release of the Meitu AI Vision large model MiracleVision 3.0 at its 15th anniversary celebration. It is claimed that MiracleVision 3.0 will be fully applied to Meitu's imaging and design products and will empower the five major industries of e-commerce, advertising, gaming, animation, and film and television.
Applications --
Adobe Releases Multiple Generative AI Innovations, Including Three New Firefly Models
According to Babite, on October 10th, software giant Adobe announced multiple generative AI innovations at the Adobe MAX 2023 conference, including three new Firefly models (Adobe Firefly Image 2 model, Adobe Firefly Vector model, and Adobe Firefly Design model), which are designed to generate content that can be safely used for commercial purposes. In addition, Adobe has released over 100 significant new artificial intelligence features and updates in its flagship Creative Cloud applications, including Adobe Illustrator, Adobe Photoshop, Adobe Lightroom, Adobe Premiere Pro, Adobe After Effects, and Adobe Stock. Adobe Express also offers more artificial intelligence features, including generated fill, text-to-template, and translation functions.
NetEase Youdao Launches Virtual Spoken English Private Tutor Hi Echo Based on Ziyue Education Large Model
According to Babite on October 11th, NetEase Youdao announced the official launch of the independent app and WeChat mini-program for the virtual spoken English private tutor Hi Echo, which is powered by the Youdao "Ziyue" education large model. As a one-on-one spoken English private tutor, Hi Echo has the ability to accompany users across multiple platforms 24/7, providing high-quality spoken English practice anytime, anywhere.
Chinese Online Introduces "Xiaoyao" Ten-Thousand-Character Creation Large Model, Capable of Generating Ten-Thousand-Character Novels with One Click
According to the webmaster's home on October 13th, Chinese Online has launched a ten-thousand-character creation large model called "Xiaoyao." This model has the function of generating ten-thousand-character novels with one click and provides authors with AI-assisted creation tools throughout the entire lifecycle.
Baidu to Launch First AI Native Map, Lane-Level Navigation Plan to Cover Over 100 Cities Within the Year
According to the Shanghai Securities News, Baidu Vice President Shang Guobin revealed on October 11th that Baidu is about to launch a brand new AI native map, completely reconstructing the product interaction of the map using the Wenxin large model, bringing new experiences to users. Users can awaken various functions and services hidden in the map through natural language dialogue, further improving travel and decision-making efficiency.
Vivo to Release Self-Developed AI Large Model Matrix on November 1st
According to Babite on October 16th, Vivo announced on its official Weibo account that it will release a self-developed AI large model matrix, including five self-developed large models of different parameter levels: one billion, ten billion, and one hundred billion. These models will comprehensively cover core application scenarios.
Black Shark S1 Pro Smartwatch Equipped with Baidu "Wenxin Yiyuan" Large Model, Providing Free AI Watch Faces
According to IT Home on October 16th, Black Shark recently announced the launch of the new Black Shark S1 Pro and S1 Classic smartwatches in the domestic market. The S1 Pro is equipped with the "Wenxin Yiyuan" large model, supporting AI question and answer functions, Baidu map navigation, and will also provide a large number of free AI watch faces.
CDPR Uses AI to Reproduce the Voice of Deceased Voice Actors in "Cyberpunk 2077"
According to IT Home on October 14th, according to Bloomberg, CDPR used AI technology to reproduce the voice of deceased voice actors in "Cyberpunk 2077." CDPR stated in a recent media interview that they have reproduced the voice of Miłogost Reczek (who played V's friend and ripperdoc Victor Victor) in the recently released DLC "Past Shadows."
Big Companies --
Google AI Search Function Upgraded to Allow Users to Create Images Directly Based on Text Prompts
According to TechCrunch on October 13th, Google is adding new features to its AI search function SGE (Search Generation Experience). The new feature introduces a conversational mode in the search, allowing users to generate images directly in SGE using prompts. Additionally, SGE now allows users to draft in SGE, enabling users to customize the length of the output content or change the writing tone to make it more serious or casual.
Apple AI/ML Team Develops Multimodal Large Model Ferret, Successfully Solves Google's Human-Machine CAPTCHA Challenge
According to the webmaster's home on October 12th, Apple's AI/ML team, in collaboration with a research team from Columbia University, developed a multimodal large model called "Ferret" that can accurately identify traffic lights in images, outperforming GPT-4V and improving the accuracy of large models in the "see-say-answer" task.
Microsoft PowerPoint and Other Applications to Integrate Copilot in November: Generate Presentations in Seconds
According to IT Home on October 13th, according to the Microsoft 365 roadmap, productivity applications including OneNote, Word, OneDrive, and PowerPoint will integrate the Copilot feature in November to further enhance productivity through AI. According to Microsoft's updated FAQ page, Copilot will support Simplified Chinese, English, Spanish, Japanese, French, German, Portuguese, and Italian, with the best performance in an English context, but localization will be further improved.
AMD to Acquire Open Source AI Software Expert Nod.ai
According to a report from East Money, AMD announced on Tuesday that it will acquire open-source AI software expert Nod.ai. As part of enhancing its software capabilities, Nod.ai will accelerate the deployment of optimized AI solutions on AMD's high-performance platforms and strengthen AMD's open-source software strategy.
TSMC's AI Orders Expected to Increase by About 6% Next Year
According to Cailian Press citing Taiwan Economic Daily on October 11th, it is reported that there is industry speculation that TSMC's six major AI customer groups will have increased demand for chip production next year. TSMC's AI orders are expected to grow by about 6% in 2024, reaching a historical high compared to this year. These six major AI customer groups include NVIDIA, AMD, Tesla, Apple, Intel, and international giants that develop AI chips and produce them at TSMC.
Opinions --
Intellibus Founder: AI is Accelerating the Fourth Industrial Revolution, More Blue-Collar Jobs Will Face Risks
According to Sina Finance on October 16th, Ed Watal, the founder and head of the IT strategy company Intellibus, stated that artificial intelligence is rapidly accelerating the fourth industrial revolution. The manufacturing industry is undergoing transformation through automation, predictive maintenance, and quality control, which is changing operations. Watal stated, "With the second generation of AI being applied to computer vision and becoming more diverse, allowing people to upload images or videos and ask questions about them, more blue-collar jobs will face risks."
OpenAI CEO: Shell ChatGPT Operators Won't Last Long, Blind Fundraising Is Very Unwise
According to Quantum Bit on October 9th, in response to OpenAI CEO Altman's entrepreneurial advice at the YC Alumni Reunion 2023, the article summarized three points: 1. Simply packaging OpenAI's company won't last long; 2. Optimistic about AI medical consultants, AI personalized one-on-one tutoring, and other directions; 3. Blind fundraising without a plan is very unwise.
Regarding simple packaging, Altman stated that some companies are addressing minor flaws in GPT models, especially those only focusing on the interface. He explained that firstly, OpenAI is already addressing most of the work, and secondly, these companies that only compete with OpenAI in terms of cost and speed will not have a long-lasting and sustainable competitive advantage. Companies must provide truly unique value.
Research Reports --
Analysis Agency: Generative AI is Overhyped, Related Industries Expected to "Cool Down" in 2024
According to IT Home on October 10th, analysis agency CCS Insight made a prediction today, suggesting that the field of generative AI may face a reality check next year. They believe that as the "hype around AI technology itself gradually diminishes," "the cost of running AI continues to rise," and "calls for AI regulation increase," the future AI industry may cool down, and there are signs of a gradual slowdown in technological development.
CCS Insight's main prediction for 2024 is that generative AI will be "doused with cold water" as the costs and risks associated with AI technology "replace the hype surrounding the technology." Ben Wood, Chief Analyst at CCS Insight, stated, "Everyone is talking about generative artificial intelligence, including Google, Amazon, Qualcomm, and Meta. We are very supportive of artificial intelligence, believing it will have a huge impact on the economy, society, and productivity. But the hype around generative artificial intelligence in 2023 is so huge that we believe AI is being excessively hyped, and the related technology needs to overcome many obstacles to bring it to the market."
"State of AI Report": GPT-4 Remains the Strongest Large Language Model Eight Months After Release
According to Synced, the annual "State of AI Report" has released its latest edition. The report indicates that OpenAI's GPT-4 remains the strongest large language model (LLM) eight months after its release, outperforming all other large models in classic benchmark tests and exams designed to evaluate human performance. However, the report points out that as cutting-edge AI systems become more powerful and flexible, comparing them will become increasingly difficult.
IDC: Local AI Chip Brands Ship Over 50,000 Units in the First Half of 2023
According to "Science and Technology Innovation Board Daily" on October 11th, an IDC report shows that the scale of the accelerated server market reached $3.1 billion in the first half of 2023, a year-on-year increase of 54%. GPU servers still dominate, accounting for 92% of the market share, reaching $3 billion. Meanwhile, non-GPU accelerated servers such as NPU, ASIC, and FPGA accounted for 8% of the market share, reaching $200 million with a year-on-year growth rate of 17%. From a brand perspective, local AI chip brands in China shipped over 50,000 units.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
