Image credit: pixabay
[Industrial Internet Weekly is a featured product released by TMTpost of Titanium Media, which will integrate the cutting-edge trends, blockbuster policies and industry research reports in the fields of enterprise-level services, cloud computing and big data this week. 】
Domestic information
Alibaba Cloud announced that it will shut down data center services in Australia and India
On June 28, Alibaba Cloud announced on its official website that based on a careful assessment and review of the global infrastructure investment layout plan, Alibaba Cloud has decided to shut down data center services in Australia and India while increasing investment in data centers in Southeast Asia, Mexico and other regions.
According to the above historical notice, Alibaba Cloud's India regional data center will be suspended after 24:00 Mumbai time on July 15, and the Australian regional data center will be suspended after 24:00 Sydney time on September 30, Australia.
DingTalk big move: Announcing the open bottom model
On June 26, DingTalk announced that it would be open to all large model manufacturers to build the most open AI ecosystem in China. Seven large model manufacturers, including Tongyi Large Model, MiniMax, Dark Side of the Moon, Zhipu AI, Orion Starry Sky, Zero One Everything, and Baichuan Intelligence, have reached cooperation with DingTalk. At present, the total number of DingTalk ecosystem partners exceeds 5,600, of which more than 100 are AI ecosystem partners. DingTalk AI is called more than 10 million times a day.
DingTalk 7.6 was officially released. According to Ye Jun, president of DingTalk, after nearly a year of exploration, DingTalk management suite products have achieved revenue of nearly 100 million yuan, with 1 10 million revenue and 10 million revenue partners.
Baidu Intelligent Cloud Qianfan launched a large-scale inclusive plan, and OpenAI migrated users to give away Tokens packages
On June 25, Baidu Intelligent Cloud Qianfan announced the launch of a large model inclusive plan, and from now on, become a newly registered enterprise user of Baidu Intelligent Cloud, and will receive a number of free services such as 0 yuan call, 0 yuan training, 0 yuan migration and 0 yuan service.
Among them, the Wenxin flagship model is free for the first time, and 50 million Tokens will be given to newly registered users of the ERNIE3.5 flagship model, and the main model ERNIE Speed/ERNIE Lite and the lightweight model ERNIE Tiny will continue to be free. For OpenAI migration users, an additional ERNIE3.5 flagship model Tokens package with the same scale as OpenAI will be given.
Partnering with Broadcom to develop AI chips? ByteDance: The news is untrue
On June 24, according to Reuters, ByteDance is cooperating with Broadcom to develop advanced AI processors in order to cut procurement costs and ensure a stable supply of high-end AI chips. According to the news, the processor is a 5nm ASIC (application-specific integrated circuit), which is expected to comply with the new U.S. export control regulations, and the manufacturing work will be outsourced to TSMC.
In response, ByteDance responded that the news was untrue.
Meituan Enterprise released the Enterprise Consumption Panorama Solution
Meituan Enterprise is a multi-scenario, one-stop enterprise consumption management platform based on Meituan's supply chain, which focuses on the deep logic of cost reduction around corporate dining, travel, employee benefits and other scenarios, helping enterprises achieve cost leadership and build core competitiveness in the new economic situation.
Huang Haiqing, former vice president of Alibaba Group, joined the "national team" of Shanghai's AI model
On June 27, at the recent meeting on empowering state-owned enterprises with the large model of Shanghai State-owned Assets Innovation Exchange hosted by the Shanghai State-owned Assets Supervision and Administration Commission, Shanghai Information Investment Co., Ltd. announced the appointment of Huang Haiqing, former vice president of Alibaba Group and president of Alibaba Cloud China, as the CEO of Shanghai Coupas Technology Co., Ltd.
Ali Qwen-2 has become the world's No. 1 open-source large model
In the early morning of June 27, Clem, co-founder and CEO of huggingface, a world-renowned open-source platform, announced on social platforms that Alibaba's latest open-source Qwen2-72B instruction fine-tuned version has become the first open-source model in the list.
He said that in order to provide a new open source large model ranking, 300 H100 blocks were used to conduct a new evaluation on the benchmark test sets of BBH, MUSR, MMLU-PRO, GPQA, etc., on more than 100 mainstream open source large models in the world, such as Qwen2, Llama-3, mixtral, Phi-3, etc.
The results show that Ali's Qwen-2 72B is the new king of Ali's open source Qwen-2 72B, the social giant Meta's Llama-3, and the famous French large model platform Mistralai's Mixtral have become the new kings, and China is in a leading position in the field of global open source large models.
Overseas news
Nvidia's total market value evaporated by 4 trillion yuan in three days, and Huang Jenxun frantically reduced his holdings
After a short period of being the world's largest market capitalization, Nvidia's ultra-high market value began to fall, with a total of 557.1 billion US dollars (about 4.055 trillion yuan) evaporated in three days, and has now fallen 13% from its peak.
At the same time, the company's founder and CEO Jensen Huang took the lead in selling the company's shares, reducing his holdings of Nvidia shares by a total of 720,000 shares in the seven trading days since June 13, with a total cash amount of nearly 95 million US dollars (about 680 million yuan).
OpenAI's AI model sales revenue has surpassed Microsoft's
On June 27, according to foreign media reports, OpenAI's artificial intelligence model sales revenue has exceeded Microsoft's similar business revenue, and its annual revenue from AI models has reached $1 billion.
TikTok "ban if you don't sell"? Oracle: The U.S. government bill will hurt our performance
According to a report by the Financial Associated Press on June 25, Eastern time on Monday, the American software giant Oracle admitted in its annual report to the US Securities and Exchange Commission that the Biden administration's "ban or sell it" bill against TikTok may damage Oracle's financial performance.
On April 24 this year, U.S. President Joe Biden signed a bill that involves forcing ByteDance to divest its app TikTok's U.S. business. In the relevant clause, ByteDance is given a deadline of about nine months to divest its U.S. business or face a national ban. In its annual report, Oracle made it clear that the law signed by U.S. President Joe Biden in April "will make it illegal for him to provide internet hosting services to TikTok" and will adversely affect Oracle's "revenues and profits."
Oracle warned that its revenue and profits would be adversely affected if it could no longer provide internet hosting services to TikTok. TikTok is one of Oracle's largest customers in the cloud infrastructure business, and analysts estimate that Oracle's annual revenue from TikTok could be between $480 million and $800 million.
For the first time, OpenAI unilaterally announced that it would terminate its API services to China
On June 25, OpenAI sent an email to Chinese API users stating that it would take steps to block API traffic from regions on the list of countries and regions that are no longer supported. The email reads: "According to the data, your organization has API traffic from regions that are not currently supported by OpenAl. Starting July 9th, we will take additional steps to stop the use of APIs from countries and regions that are not on OpenAI's list of supported countries. ”
Currently, OpenAI's API is open to 161 countries and regions, but China is not included. This also means that OpenAI announced that it will terminate its API services to China.
Apple will set up a data center in Taiwan
On June 25, industry news that Apple will set up a data center in Taiwan. In response to this rumor, the "Ministry of Digital Development" in Taiwan said that it "could not be confirmed".
AMD将构建全球最大AI训练集群,集成1.2000片GPU
According to The Next Platform, Forrest Norrod, executive vice president and general manager of AMD's data center solutions group, said that AMD will help build the world's largest single artificial intelligence (AI) training cluster, which will integrate up to 1.2 million GPUs.
OpenAI API sales surpassed Microsoft's, with an annualized revenue of $1 billion
On June 28, according to industry sources, as of March, OpenAI's annualized revenue through the sale of model access was about $1 billion. In contrast, Microsoft's similar product, Azure OpenAI Service, only recently reached $1 billion in annual recurring revenue (ARR).
According to The Information, in March last year, Microsoft tried to convince businesses to buy OpenAI's technology through Azure instead of buying it directly from OpenAI, telling them that Azure was more private and secure. At the same time, OpenAI is also developing its API business. At the end of June 2023, OpenAI's ARR reached $333 million, accounting for one-third of its revenue at the time. At the same time, OpenAI has also begun to strengthen its sales business, and the team size has expanded from about 10 people a year ago to more than 200 people.
Google releases Gemma 2, the strongest open-source model
On June 27, local time, Google released the Gemma 2 model that was previewed at the I/O developer conference a month ago. According to Google, the new model has better performance and higher inference efficiency than the first-generation Gemma model. Gemma 2 includes two parameter sizes, 9B and 27B, of which the official claim is that the 27B model is comparable to models twice as large as it in terms of performance, and the 9B model is also better than Meta's Llama 3 8B and other open-source models of similar size.
Meta has released an LLM compiler that it says will change the way we program
On June 28, Meta announced the launch of LLM Compiler, a series of models built on Meta Code Llama, with code optimization and compiler functions. These models are able to simulate the compiler, predict the best route path for code size, and perform code disassembly. LLM Compiler is state-of-the-art in code size optimization and disassembly, demonstrating the potential of AI in the field of code optimization.
Financing mergers and acquisitions
Baiwangyun opened its IPO and plans to land on the Hong Kong Stock Exchange on July 9
Baiwangyun will open its IPO on June 28, 2024 and end on July 4, 2024, and the company plans to offer 9.262 million shares globally. It is expected to be officially listed on the main board of the Hong Kong Stock Exchange on July 9, 2024 under the stock code of "6657", and the sponsor is Haitong International Capital Co., Ltd.
Founded in 2015, Baiwang Cloud is a comprehensive enterprise digital solution provider, providing users with digital and data-driven intelligent solutions and products for finance and taxation by building an online, automated and intelligent enterprise business social network, processing all kinds of transaction vouchers (including but not limited to invoices, receipts, documents and other accounting vouchers, etc.) in the key processes of enterprise transactions, and realizing a full closed-loop digital business from procurement optimization, reconciliation and settlement, electronic invoices, smart finance and taxation to payment and financing.
Wang Yuan, the former vice president of NetEase, established Jiuji Technology, and has reserved sufficient funds for two to three years in the seed round of financing
Wang Yuan, former vice president of NetEase, executive director of Hangzhou Research Institute, and general manager of NetEase Shufan, recently established Hangzhou Jiuji Technology Co., Ltd. and served as CEO, and the first product is positioned as a "general office assistant centered on personal data". Jiuji Technology has completed a seed round of funding, mainly in US dollar funds, with sufficient funds for the initial team in reserve for two to three years. (Jiazi Lightyear)
OceanBase, a subsidiary of Ant Group, was restructured in its shareholding structure and was jointly held by multiple parties
Beijing Aoxing Bess Technology Co., Ltd., the main company of OceanBase, a former subsidiary of Ant Group, has completed the adjustment of its shareholding structure and accelerated its independent market expansion.
OceanBase is a native distributed database independently developed by Ant Group since 2010 and has been in commercial operation since 2020. In the course of 14 years of development, it has successively broken the world records of the "Database World Cup" TPC-C AND TPC-H, and currently serves more than 1,000 customers. In March this year, OceanBase established a board of directors and officially operated independently.
Tingyu Technology received hundreds of millions of yuan in Series B financing
Tingyu Technology, an edge cloud service provider, announced the completion of hundreds of millions of yuan in Series B financing, which was jointly led by Guohai Investment and Fosun Ruizheng, followed by Cornerstone Information Innovation, Zhongxiang Venture Capital, Guolian Shares, Sichuan Cultural Investment, Sanqi Mutual Entertainment, and Ingenuity Investment. Founded in 2019, Tingyu Technology's core product is a self-developed real-time interactive distributed edge computing cloud platform
Anke received tens of millions of yuan in financing, led by Jingya Capital
Cybersecurity manufacturer Anke announced the successful completion of tens of millions of yuan in Series A financing, which was led by Jingya Capital, followed by the old shareholder Shunwei Capital, and Crypto Capital served as the exclusive financial advisor.
Galaxy General raised 700 million yuan
Galaxy General Motors, an intelligent robot innovation company, announced the completion of an angel round of financing, totaling 700 million yuan, with investors including strategic and industrial investors such as Meituan Dianping War Investment, BAIC Industrial Investment, SenseTime Guoxiang Fund, and iFLYTEK Fund. Qiming Venture Capital, BlueRun Venture Capital, Matrix Partners, Source Code Capital, IDG Capital and other leading financial institutions, as well as Light Source Capital, also participated in the early investment.
DreamTech连续完成两轮数千万元融资
AI startup DreamTech has continuously completed tens of millions of yuan in angel rounds and angel+ rounds of financing, with the angel round led by Oriza Origin, followed by TusStar Venture Capital and Cloud Angel Fund, and the angel+ round is the exclusive investment of Chuxin Capital. DreamTech, an AI startup focused on native 3D generation, began official operations in December 2023.
Klarity Intelligence融资7000万美元
文档智能初创公司Klarity Intelligence宣布获得7000万美元融资,本轮融资由Nat Friedman和Daniel Gross领投,Scale Venture Partners、Tola Capital、Picus Capital、Invus Capital和Y Combinator也参与其中。 Klarity Intelligence成立于2017年6月5日,是一家利用自然语言处理、计算机视觉和通用机器学习技术为用户提供自动化文档管理、审查、报告和审计跟踪等服务的公司。
OpenAI acquires developer collaboration startup Multi
On June 25, OpenAI acquired Multi Software Co., a startup that provides a video conferencing platform for software development teams. Multi co-founder and CEO Alexander Embiricos announced the deal in a blog post.
Policies & Trends
2024Q1 Chinese mainland cloud service market: Alibaba Cloud first, HUAWEI CLOUD second
News on June 29, the latest data from Canalys shows that in the first quarter of 2024, Chinese mainland's cloud infrastructure service spending increased by 20% year-on-year to $9.2 billion. The three giants of China's cloud services market, Alibaba Cloud, Huawei Cloud and Tencent Cloud, continued to dominate the market, growing together by 22% and accounting for 72% of the overall market, down 2% from the previous quarter.
IDC Market Presentation: The State of China's Cybersecurity Hardware Market and Technology Development Trends Driven by Generative AI 2024
IDC's latest Worldwide Cybersecurity Spending Guide shows that the global cybersecurity hardware market will reach $176.4 billion by 2027. Among them, the scale of China's network security hardware market will reach 36.4 billion yuan in 2027, and the market will develop steadily. Specifically, the next-generation AI firewall, as a rigid product in the market, will continue to be the largest product category in the network security hardware market.
IDC: China's Hardware Video Conferencing Market Share and China's Cloud Conferencing Market Share
According to IDC data, in 2023, China's video conferencing market will reach US$920 million (about 6.49 billion yuan), a decrease of 2.7% from the same period in 2022. Among them, the market size of hardware video conferencing was 650 million US dollars (about 4.57 billion yuan), a year-on-year decrease of 7.6%, but the decline was narrower than that in the first half of 2023; The cloud conferencing market benefited from the acceleration of commercialization, with a year-on-year increase of 11.6% to US$270 million (about 1.91 billion yuan), maintaining a rapid growth trend.
![The group stage of Euro 2024 is coming to an end, and when watching live TV, you can see the advertisements of Chinese companies in almost every game. According to the reference Elimination[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)