bytedance

According to Bloomberg, ByteDance is preparing to launch an artificial intelligence model for real-time spatial video generation, with applications covering fields such as robotics and autonomous systems, and will compete with companies like Meta and Alphabet.Insiders say that ByteDance founder Zhang Yiming is personally overseeing the development of the model, coordinating multiple business departments within the company, and investing in artificial intelligence resources and computing power support. The model may be released as early as next month, but the specific timing is still subject to change.

According to Blue Whale News, from Changxin Technology, ByteDance and ZTE have jointly developed a new generation of Doubao phone------Nubia NaviX Ultra, which will be equipped with Changxin's highest speed LPDDR5X memory at 10667Mbps. The official statement indicates that this is also the first time that domestically produced 10667Mbps high-speed LPDDR5X memory chips have achieved mass production and application.Previous news reported that the Nubia "Doubao phone" has completed network filing and will be mass-produced and launched in September.

ByteDance is set to secure a $29.6 billion syndicated loan. Initially, ByteDance planned to raise $20 billion, but due to strong demand from banks, the size was ultimately increased to $29.6 billion. The funds will primarily be used for general corporate purposes. The transaction has not yet been officially signed, and banks are still confirming allocation amounts. Once completed, this loan will become the second-largest dollar-denominated syndicated loan in Asia this year, following SoftBank Group's $40 billion bridge loan in March.ByteDance is accelerating its investment in AI, considering raising its capital expenditures to as much as $70 billion in 2026, which is more than double last year's amount, mainly for the expansion of data centers and other AI infrastructure. The last large-scale overseas loan by ByteDance occurred in 2024, raising approximately $10.8 billion.

ByteDance announces an adjustment to the option price, with the price for active employees adjusted from $229.5 per share to $241.35 per share, an increase of approximately 5.1%. This is the second adjustment to the option price by ByteDance since 2026.The last adjustment to the option price by ByteDance was in April of this year, when the price for active employees was adjusted from $200.41 per share to $229.5 per share.

According to reports from the major factory, ByteDance recently held a briefing on Doubao stocks and introduced new policies for relevant employees: the price of Doubao stocks was adjusted from $14.85 per share in June to $17.02 per share, an increase of about 14.6%; the issuance of Doubao stocks was increased, with some employees reaching a total package of about 5%; employees can choose to convert a certain proportion of cash in their total package into Doubao stocks, with the proportion varying by job level.This adjustment mainly targets the large model departments under Seed (pre-training, multimodal, VLM, etc.) and the Doubao main APP team under Flow. From $13.08 in April of this year to $17.02 in August, the Doubao stock has increased by about 30.1% in less than half a year, which is about twice the increase of ByteDance options during the same period (approximately 14.5%).

According to reports from Intelligent Emergence, ByteDance has completed the team integration of its office AI products, with the TRAE and Coze teams being fully merged into the Doubao system. Among them, TRAE Work and Coze will integrate their product capabilities in work scenarios with Doubao, while TRAE IDE and CLI will continue to develop as a programming product line under the Doubao brand. After the adjustment, the relevant product and operations teams will report to Zhao Qi, the product head of Doubao.TRAE and Coze were originally part of ByteDance's Product R&D and Engineering Architecture Department, positioned respectively as AI programming products and AI intelligent entity development platforms. Starting this year, productivity agents have become an important strategic direction for Doubao, and the experience accumulated by both parties will be reused in Doubao. Following this adjustment, ByteDance has further clarified the AI core business positioning of Doubao, focusing AI office products on Doubao, and will launch the independent AI office product "Doubao Work" as a unified product and brand for AI office scenarios as early as this week.ByteDance responded that the adjustment aims to better coordinate product and technical resources to provide users with a higher quality AI work experience, and existing user rights will not be affected. Recently, Doubao has intensively updated office functions such as remote computer control, Windows virtual desktop, cloud computer, and side workbench, and has launched over 200 skills and connectors; the independent "Doubao Work" will also be deeply integrated with Feishu, with the former Feishu "Aily Intelligent Partner" having been renamed to "Doubao Work Partner."

According to LatePost, ByteDance is discussing a large model with training parameters exceeding 50 trillion, surpassing Alibaba's Qwen 3.8-Max (24 trillion) and Moonlight K3 (28 trillion), making it the largest known plan in the country so far. This plan is still in its early stages and does not guarantee a final release. The new model is intended to be led by Xiang Liang, head of Seed Foundation, in collaboration with Shen Ke, who is responsible for the pre-training data of large language models. Seed is reorganizing, dividing responsibilities, and allocating resources based on this.Two weeks ago, ByteDance founder Zhang Yiming held a company-wide meeting with Seed head Wu Yonghui. Zhang reassured the team that training large models is inherently difficult and that it is acceptable to lag behind for a period of time, hoping to aim for the upper limits of intelligence and join the world's top tier. He acknowledged that programming is a key direction at present, advocating for the integration of Volcano Engine, Feishu, and Doubao resources to build computational power and data advantages, while reminding not to be led by a single hot topic. He praised Seedance's differentiated leadership and clearly opposed distillation, believing it is difficult to truly surpass and that AGI barriers should be built from a more fundamental level, stating that the company will continue to increase investment in AI.In the past six months, Seed's multimodal performance has been outstanding, with Seedance 2.0, Seedream, and others driving Volcano Engine MaaS, but the market response to the language model Seed 2.0 has been limited, and its lagging coding capabilities have affected the revenue structure. ByteDance has hired Guo Daye at a high salary to specialize in coding and has consolidated related resources. In the face of the industry's general trend of increasing model sizes, ByteDance hopes to achieve a leapfrog advantage with a larger scale while promoting the elimination of horse racing and breaking down departmental walls to concentrate efforts on tackling challenges.

China has begun to relax import restrictions on NVIDIA's H200. ByteDance and Tencent have each received about 10,000 H200 in recent weeks, and several other tech companies may soon be approved for similar quantities.The United States had previously approved several Chinese companies to purchase H200, but the transactions had been stuck in the Chinese approval process. Now, chips are finally starting to enter the mainland in batches, but it is far from a complete opening.Beijing currently only allows a small number of H200 to enter the mainland, hoping that companies will deploy more chips in overseas regions such as Hong Kong. Mainland procurement still requires separate approval. This approach allows leading AI companies to continue using NVIDIA to train cutting-edge models while also leaving market space for domestic chips like those from Huawei.

In response to rumors on social media about "Doubao recommending hotels for a fee," Liu Xing, the head of public affairs for Doubao at ByteDance, stated on the evening of August 10 that the related statements are inaccurate. He indicated that Doubao's life service business currently does not have paid promotions, and Doubao does not charge advertising fees for hotel recommendations; merchants cannot influence recommendations or rankings through payment, and only pay channel service fees after an order is completed.Liu Xing pointed out that some platforms adopt a "commission + advertising" charging model, while Doubao primarily charges based on completed transactions at this stage, making direct comparisons inappropriate. For merchants, the actual channel costs and the effects of new orders are more meaningful references. He also mentioned that Doubao's life service business is still in a very early exploratory stage.

According to the Financial Times, as Chinese companies continue to narrow the gap with the top AI labs in the United States, ByteDance is training an AI model that may be close in scale to Anthropic's advanced Mythos system. According to three insiders, this Chinese tech giant is currently in the early stages of training a large-scale model, which may have a parameter count reaching the trillions, approximately three times that of the largest model released in China, the Kimi K3 model under the Moonlight brand.One insider stated that ByteDance's model is currently undergoing pre-training, a phase that typically takes three to six months. If all goes well, it will then undergo fine-tuning and be released. The specific scale of the model will be determined later. Anthropic has not publicly disclosed the parameter scale of its models, but industry estimates suggest that its advanced Mythos 5 model has a parameter count of about 8 trillion, while Fable 5 has around 5 trillion.It is important to note that the number of parameters determines the model's basic capacity for storing information or memory limit, but the actual capabilities of the model also depend on other factors, including data quality and training methods.

According to a report by LatePost, ByteDance is discussing training a model with a parameter scale exceeding 50 trillion, surpassing Alibaba's Qwen3.8-Max (24 trillion parameters) and the K3 from Dark Moon (28 trillion parameters), making it currently the largest known model in terms of parameter scale in the country. Generally speaking, the larger the model scale, the higher the level of intelligence.However, this plan is still in the early stages and does not guarantee that the model will ultimately be released. The new model will be led by Xiang Liang, head of SeedFoundation, in collaboration with Shen Ke, who is responsible for the pre-training data of large language models. To this end, Seed is reorganizing, defining responsibilities, and allocating resources.

ByteDance founder Zhang Yiming clearly stated at the Seed team all-hands meeting last month that the company will not use model distillation as a shortcut to catch up with large models, even if it temporarily falls behind domestic competitors. He urged the team to be willing to sacrifice short-term gains for long-term goals.Distillation refers to allowing a new model to learn from the outputs of a stronger model, which can save training time and computational power. Internally, ByteDance believes that this choice will make it more difficult for Seed's language model to catch up with domestic peers like DeepSeek, Kimi, and Qianwen in the short term. However, ByteDance is more concerned about political risks: the company has long faced scrutiny from the U.S. government due to TikTok, and if it is accused of massively extracting the capabilities of cutting-edge U.S. models, Washington may apply pressure on TikTok again. Previously, Anthropic accused DeepSeek, Dark Side of the Moon, MiniMax, Zhiyu, and Alibaba of massively extracting Claude's capabilities, but did not name ByteDance.

According to The Information, ByteDance founder Zhang Yiming told employees at a meeting last month that the company would not use distillation technology to advance its artificial intelligence model capabilities, even if it means the company temporarily falls behind domestic competitors.

According to a report by the South China Morning Post citing informed sources, ByteDance's AI business has achieved an annual revenue of $4 billion, the highest known level in the domestic market, but still far behind leading AI companies in the United States. Previously, ByteDance announced through an internal email the integration of the Feishu and Doubao product teams and GTM system, with Doubao head Zhao Qi leading the product line and Volcano Engine head Tan Dai responsible for the new ToB GTM organization "Creativity Service Platform," aimed at building "stronger AI product competitiveness and customer service capabilities."This integration is the largest adjustment to the ToB business since ByteDance's business segmentation in 2021. The Doubao enterprise version, developed with the participation of the Feishu team, has already entered internal testing with some clients. The report pointed out that ByteDance's AI revenue mainly comes from enterprise API calls of the Doubao large model, AI cloud services, and enterprise-level AI solutions.

According to exclusive reports from Titanium Media, ByteDance released an internal email on July 30, announcing the integration of the Feishu product team and the Doubao product team to form a new Doubao product team, which will be unified under the leadership of Doubao head Zhao Qi, while Feishu head Xie Xin will report to Zhao Qi. In terms of the GTM system, the Feishu GTM team will integrate with the Volcano Engine team to establish a new ToB GTM organization called the "Creativity Service Platform," which will be responsible for the marketing, sales, and customer service of ByteDance's MaaS and SaaS cloud services, led by Volcano Engine head Tan Dai, with Feishu sales head Lin Chan and Feishu strategy and marketing head Shi Zhijun reporting to Tan Dai.This integration is the largest adjustment to ByteDance's ToB business since it implemented business segmentation in 2021. After the reorganization, Feishu's products and services will remain unchanged, and Feishu will collaborate more deeply with Doubao on productivity scenarios. The Doubao enterprise version, developed with deep involvement from the Feishu product team, has already begun internal testing among some Feishu clients. Feishu currently covers over 10 million enterprise users, and this integration will further enhance the integrated collaborative capabilities of the ToB business.

According to IT Home, ByteDance announced today the official launch of the Seed STEM Scientist Program, inviting 100 outstanding scholars from the forefront of scientific fields to collaborate on real research problems and promote AI-accelerated scientific discovery.This program is initiated by the Seed Edge research team, providing ample computing resources and competitive salaries. Participation is divided into two categories: scientific advisors and doctoral interns. The first phase of the program is expected to last about 6 months, with the work location in Haidian District, Beijing. The application deadline is September 30, 2026.

According to Yicai, there were previous reports that Kunlun Core, a subsidiary of Baidu, plans to go public in Hong Kong, and Tencent has become its client, while ByteDance is also considering using its AI chips. In response to this, sources close to the matter told Yicai today that ByteDance currently has no intention of cooperating with Kunlun Core.

According to Jinshi, Qualcomm's stock in the U.S. night market once rose over 4%, and the current increase has narrowed to 3%. The news states that Qualcomm is in talks with ByteDance to provide customized chip design services.