At the opening ceremony of the 2024 Yunqi Conference, Giant Network made its debut and showcased its latest achievements in the "Game + AI" field. The company launched two self-developed large model applications—GiantGPT and BaiLing-TTS—while also demonstrating new technologies such as AI Digital Humans and the AI painting platform Giant Mimicry.

Recently, foreign media reported that ByteDance is accelerating its efforts to self-develop and manufacture AI chips to reduce dependence on expensive chips from NVIDIA. The report claims that ByteDance plans to collaborate with TSMC and expects to mass-produce two self-developed AI chips by 2026, having already reserved the production of hundreds of thousands of chips. More notably, the report pointed out that ByteDance aims for its self-developed chip clusters to achieve four times the computing performance of a single H100 chip at a similar cost. In response, ByteDance quickly addressed this report.

At the 2024 Inclusion · Bund Conference, Ant Group shared its latest progress in building knowledge-enhanced professional intelligent agents and launched the research and development achievement of combining knowledge graphs with large models—the Knowledge-Enhanced Large Model Service Framework KAG. This framework, introduced by Liang Lei, head of the knowledge graph at Ant Group, aims to guide decision-making and retrieval through graph logical symbols, significantly improving the precision and logical rigor of decisions in vertical domains.

Sichuan Changhong Electric Co., Ltd. is set to launch the world's first AI television globally, which will be officially unveiled on September 6th in Berlin, Germany. This innovative product integrates industry-leading large model technology and is supported by a national supercomputing center. It is equipped with Changhong's self-developed Yunfan AI platform, aiming to achieve autonomous thinking, communication, and growth in televisions, thus providing a new experience that transitions from passively receiving commands to actively serving and autonomously evolving.

This afternoon, Nubia officially unveiled its self-developed Nebula Large Model at a new product launch event, marking a significant breakthrough for the company in the field of AI technology. In the rapidly evolving landscape of AI technology, Nubia has solidified its "AI+" product strategy, successfully developing the Nebula Large Model through close collaboration with professional AI research teams. This model encompasses advanced AI capabilities across multiple domains, including natural lan

["NetEase released its 2023 financial report, with total revenue of 103.5 billion yuan", "Continued investment in self-developed technology, with annual R&D reaching 16.5 billion yuan", "AI technology deeply integrated into games, music, education, and other businesses", "Game business performed outstandingly, with multiple self-developed games breaking category records", "Youdao, Cloud Music, and other businesses also achieved significant results", "NetEase launched the 'Minor Mode' to prevent online fraud and donated materials to support disaster areas", "NetEase will continue to promote the application of AI technology in multiple fields, developing high-quality products and services"]

Microsoft has launched two self-developed AI chips: Maia 100 and CobAlt, which will be manufactured by TSMC using 5nm process technology. They are expected to be deployed in Microsoft's Azure data centers next year to enhance the energy efficiency of generative AI products and reduce costs, as Microsoft aims to gain an advantage in the AI and cloud computing fields.

Tencent has expanded the parameters of its advertising artificial intelligence model, enhancing targeting and attribution accuracy. Tencent upgraded its self-developed foundation model Tencent Hunyuan and deployed it to internal products. Tencent Cloud is applying the large model across multiple products by upgrading its MaaS capabilities. Tencent Hunyuan plays a role in advertising, gaming, and content sectors. Tencent Cloud promotes the application of large models across various industries.

Ant Group has released Kapacity V0.2, an intelligent cloud capacity technology tool that incorporates self-developed AI algorithms. It offers advantages in predictive elasticity scaling, allowing for proactive responses to traffic fluctuations. Kapacity uses a traffic-driven replica prediction algorithm and also provides support for custom metrics and workload dimension aggregation query capabilities.

["❗️vivo Vice President Zhou Wei stated that vivo invests 2 to 3 billion annually in large models.", "❗️vivo has released the self-developed general large model matrix Blue Heart Model, which includes parameter scales of billions, hundreds of billions, and trillions.", "❗️The 7B Blue Heart Model is designed for mobile devices and is capable of both cloud and edge computing, featuring abilities in dialogue, knowledge Q&A, and intent understanding.", "❗️The 70B model is the main model for cloud services, featuring abilities in role-playing, knowledge Q&A, and natural dialogue."]

Jovi Copilot, vivo's self-developed AI assistant, will make its debut in the OriginOS 4 system. Jovi Copilot features strong voice interaction, information retrieval, and text generation capabilities. Additionally, Jovi Copilot can create drawings and achieve artistic style transfer in images. It possesses problem analysis and conversational abilities, serving as a daily intelligent companion for users. The emergence of Jovi Copilot marks a significant breakthrough for vivo in the field of AI technology.

vivo's self-developed AI large model will be first applied in the OriginOS 4 system to be released on November 1, featuring five self-developed large models with different parameter scales of billions, tens of billions, and hundreds of billions. It ranks first in the C-Eval large model evaluation leaderboard. OriginOS 4 integrates a centralized intelligent assistant, providing more practical function scenarios.

OpenAI is considering self-developed AI chips to reduce hardware investment costs and overcome GPU shortages. The self-developed chips are seen as a signal of further independence from Microsoft. OpenAI is exploring the manufacture of consumer-level AI products in collaboration with Jony Ive to develop an 'AI iPhone.' The self-developed chips will help OpenAI achieve control similar to that of the Wintel alliance in the PC era. The ambition behind OpenAI's self-developed chips signifies the start of a new chess game.

OpenAI is exploring the development of its own AI chips to address the GPU shortage and high costs. The self-developed chips are seen as a signal of independence from Microsoft and Nvidia. OpenAI may aspire to lead the AI era by playing a role similar to that of Intel in the PC era. OpenAI is also collaborating with former Apple designers to lay the groundwork for the next generation of AI smart devices. The era of AI warfare has begun, and it will take time to see who will dominate.

OpenAI is exploring the self-development of AI chips and is currently evaluating a potential chip company acquisition target. The GPU supply shortage has prompted OpenAI to consider developing its own chips, as current AI computing resources primarily rely on NVIDIA GPUs. Developing self-made chips implies an average annual investment of hundreds of millions of dollars for OpenAI, and the entire process may take several years. Amidst the chip shortage, several tech giants like Microsoft and Meta are also exploring the development of their own AI chips, but results have been limited. If successful in developing its own chips, OpenAI will join the ranks of a few companies that possess autonomous AI chip technology.

Alexis Black Bjorlin will leave Meta by the end of the month. Bjorlin previously worked at Intel and Broadcom, leading Meta's self-developed AI chip team. Meta's Vice President of Engineering, Song, will succeed Bjorlin to continue advancing the self-developed chips. This week, Meta launched consumer-facing generative AI products that require substantial support from self-developed chips. Meta is redesigning its data centers to support AI operations using both self-developed chips and Nvidia chips.

Google is considering the independent production of Tensor Processing Units (TPUs) to reduce its reliance on Broadcom, aiming to save significant costs. This move aligns Google with giants like Amazon and Microsoft, who are also developing their own AI chips, potentially saving billions of dollars annually. A pricing strategy stalemate with Broadcom may lead to completely self-developed TPUs. Amazon and Microsoft are also actively advancing their own AI chip development to enhance AI capabilities. A Google spokesperson stated that there are currently no plans to change the partnership relationship with Broadcom.

vivo's self-developed AI large model has performed excellently on the global Chinese rankings. vivo will first apply the AI large model in the OriginOS 4 system. OriginOS 4 will include a centralized intelligent assistant, providing more practical functional scenarios. vivo has made significant progress in the field of AI. The release of OriginOS 4 will bring an outstanding user experience.

['Kunlun Wanwei's self-developed TianGong large model achieved an 80% accuracy rate in the Benchmark GSM8K test.', 'The inference capability of the TianGong large model exceeds that of GPT-3.5 and LLaMA2.', 'The TianGong large model performs exceptionally well across multiple datasets.', 'The problem-solving approach of the TianGong large model is simpler and clearer.', 'The TianGong large model will provide strong intelligent support for users and enterprises.']

{1: Samsung will release its self-developed generative AI tool on September 12, possibly named 'Simply Chat'. 2: The tool is aimed at Samsung's internal employees to enhance work efficiency. 3: The tool may begin internal testing in October, with a planned rollout for all Samsung employees by the end of 2023 or early 2024.}