aiqbench analysis
The article is mostly supplier messaging, and the firms quoted have an obvious stake in saying orders are strong, so the claims of "no cutback" deserve scrutiny. The more concrete signals are memory tightness, with capacity exits and continued shortages, and rising 400G/800G interconnect volumes, which point to sustained data-centre build-out. Readers should watch hyperscaler capex guidance and memory pricing data, which will say more about demand than supplier statements.
Tech Giants' AI Demand Slowing? Computing Power Supply Chain Firms Respond: No Sign of Cuts, Orders Rising in Price
Rumors suggest that major tech companies are scaling back AI demand, heightening market concerns about the scale and growth pace of tech AI spending. In response, several firms in the computing power supply chain said there are no signs of a cutback and that their existing order books are robust. Suzhou-based Cooperative Data (协创数据) said there are currently no signs of a cutback. Leading domestic internet companies are expected to keep capital expenditure at high levels next year. Investment in computing infrastructure is expected to remain highly prosperous. Xingyun Technology (行云科技) said that cumulative computing and storage-related orders announced by the company this year have reached about 20.641 billion yuan. GigaDevice (兆易创新) said that, driven by the rapid development of AI boosting demand for memory chips and related spillover effects, the storage industry's upturn continues. Major industry players are cutting or shifting capacity (exiting niche markets), and the overall niche memory market will remain in a state of tight supply. Ruikeda (瑞可达) said that its 400G/800G AEC products have gradually entered batch delivery, with capacity continuing to ramp up. (Securities Times)
Read the full article at 36KrRelated coverage
When to Think Deeply: Inhibitory Deliberation for LLM Reasoning
arXiv · Jun 8, 2026

£5 Million Fuels Imperagen’s Quantum-AI Enzyme Engineering
Quantum Zeitgeist · May 23, 2026
Physics-Informed Broad Learning System: An Efficient Backpropagation-Free Framework for Solving Partial Differential Equations
arXiv · Jul 29, 2026
Multimodal Dataset Normalization and Perceptual Validation for Music-Taste Correspondences
arXiv · Sep 15, 2026