This morning’s biggest development is on the policy front: the Ministry of Commerce and eight other departments issued a document promoting “commerce through science and technology,” incorporating “AI + Manufacturing” into consumer goods supply. On the industry side, inference computing is becoming the main driver of demand, and OpenAI is ramping up robotics hiring.
Top Story: Nine Departments Issue Policy to Advance Commerce Through Science and Technology and Deepen “AI + Manufacturing”
According to a National Business Daily news brief, on October 19, the Ministry of Commerce and eight other departments released the Opinions on Accelerating the Advancement of Commerce Through Science and Technology [1]. The document calls for further implementation of the “AI + Manufacturing” initiative, promoting deep integration of AI technology into product research and development, design, manufacturing, and marketing, and expanding the supply of smart products, including smartphones, computers, smart home products, wearables, and robots [1]. It also encourages the development and deployment of smart products based on secure, reliable open-source operating systems [1].
Based on the details disclosed so far, the document focuses on commerce, not just technology: AI is being integrated throughout the product development and marketing process, with the goal of bringing more AI-enabled consumer goods to market. The same document also covers non-AI topics, including upgrading traditional industries, building a “Chinese Consumer Brands Showcase,” technology-driven agriculture, and innovation among time-honored brands [1]. This indicates that AI is part of a broader effort to boost consumption and drive upgrading.
It is worth noting that only a news brief is currently available, with no budget, numerical targets, or timeline. Specific support measures and the division of responsibilities among departments will need to be checked against the full official text once it is released. For smartphone, home appliance, wearable, and robotics manufacturers, this signals a policy direction, but how many orders it might generate in the near term remains uncertain.
Key News
Inference Computing Becomes the Main Driver of Demand, With Its Share Projected to Reach 70.5% in 2026
At the 2026 National Conference on High Performance Computing, held on October 18, several industry participants said inference computing had become the main driver of demand, with market attention shifting toward value for money per token [2]. Citing a report released in September by the China Academy of Information and Communications Technology, The Paper reported that inference computing is projected to account for 70.5% of computing demand in 2026; a single 1.5-minute episode of an AI-generated short drama can consume 5 million to 10 million tokens [2]. A Huawei Kunpeng HPC expert also noted that the CPU-to-GPU ratio is moving from 1:8 in the training era toward a higher ratio [2]. Why it matters: computing procurement criteria are shifting from “how large a model can it train?” to “how low is the cost per token?” This will affect how servers, chips, and cloud services are priced.
Image: Xin Aisuan co-founder Yang Zhi says customers are now focused on token costs, as a new, largely untapped market worth trillions emerges. (Source: The Paper (republished by Sina Finance))
OpenAI Lists 19 Robotics Jobs on Its Website
According to a tally by The Paper, as of October 19, OpenAI had posted 19 robotics-related positions on its website, spanning control systems, electrical engineering, actuators, motor design, manufacturing, and data systems [3]. OpenAI says on its website that the team aims to develop general-purpose robots capable of operating in dynamic, real-world environments and explore a range of robot forms [3]. Why it matters: The roles cover hardware and manufacturing, suggesting that OpenAI’s robotics plans extend beyond models. For now, however, these are only job postings, with no product or timeline announced.
Image: Robotics positions posted on OpenAI’s website (Source: The Paper, republished by Sina Finance)
News Briefs
- NVIDIA: According to MyDrivers, the Vera CPU integrates 88 cores on a single die. Official figures claim single-core performance up to 1.5 times that of AMD EPYC Venice, based on the vendor’s own testing [4].
- Canton Fair: At the 140th Canton Fair, coffee robot company Haitun Technology said prospective orders on the first day were up about 15% from the spring session. The claim of a “four-month payback period” comes from the company [5].
References
- Nine Departments: Accelerate Commerce Through Science and Technology and Deepen Implementation of the “AI + Manufacturing” Special Initiative — National Business Daily (republished by Sina Finance), accessed 2026-10-19
- Demand for Inference Computing Power Surges: AI Servers Face Supply Shortages as a New Trillion-Yuan Blue-Ocean Market Emerges — The Paper (republished by Sina Finance), accessed 2026-10-19
- OpenAI Launches Major Hiring Push to Ramp Up Robotics R&D — The Paper (republished by Sina Finance), accessed 2026-10-19
- NVIDIA Admits Chip Design Misstep: Single-Core Performance Was Sacrificed for Years, and Agentic AI Is Forcing a Reversal — MyDrivers (republished by Sina), accessed 2026-10-19
- Chinese Companies Sell Coffee Robots to 75 Countries, Claiming Some Stores Recoup Their Investment in Just Four Months! Leading Manufacturers Share Their Strategies for Overseas Expansion — National Business Daily (republished by Sina Finance), accessed 2026-10-19
