Huawei Ascend
PulseAugur coverage of Huawei Ascend — every cluster mentioning Huawei Ascend across labs, papers, and developer communities, ranked by signal.
- developed by Huawei 100%
- developed Cann 90%
- used by Meituan 90%
- instance of Ascend 950 90%
- used by LongCat-2.0 90%
- partners with University of Science and Technology of China 90%
- developed by Cann 70%
- developed Ascend 950 70%
- used by Shenzhen 70%
- instance of alphaXiv 70%
- used by Tsinghua University 70%
- used by Cann 70%
- 2026-06-24 product_launch Huawei will announce new Ascend AI chip solutions at the 2026 World Artificial Intelligence Conference. source
7 day(s) with sentiment data
-
Marvell pushes wafer production amid AI hardware demand; Huawei develops Ascend NPUs
Marvell Technology is urging GlobalFoundries to increase wafer production to meet anticipated demand, potentially for next-generation AI hardware. Meanwhile, Huawei is developing its Ascend NPUs, aiming to compete with …
-
Huawei's Ascend NPUs set to lead China's AI hardware race
Huawei is developing its next-generation Ascend NPUs, which are poised to become China's leading AI hardware option. The 960DT model is expected to offer performance significantly surpassing Western alternatives. This d…
-
US seeks fault-tolerant quantum computer by 2028 amid AI and security news
The US Department of Energy is seeking proposals for a fault-tolerant quantum computer by 2028, aiming to advance scientific research and national security. This initiative is part of a broader effort to develop advance…
-
Researchers find way to listen in on headphones from afar
Researchers have discovered a novel method to remotely eavesdrop on headphone audio signals. This technique exploits the electromagnetic emissions from audio playback devices, allowing for the reconstruction of sound wi…
-
Huawei Ascend CANN community leads China with 6,230+ active developers
Huawei's Ascend CANN open-source community has become a leader in China, boasting over 6,230 monthly active developers. The community has achieved more than 60 million repository downloads, with a significant portion of…
-
Cloud compute cost optimization: On-demand vs. dynamic scaling strategies
Optimizing cloud compute costs for AI workloads involves a strategic choice between on-demand allocation and dynamic scaling, depending on workload patterns and service level agreements (SLAs). On-demand allocation is b…
-
China's AI chip industry booms amid US sanctions, DeepSeek eyes Huawei chips
US sanctions aimed at hindering China's AI development appear to be inadvertently accelerating the growth of China's domestic chip industry. Companies like DeepSeek are reportedly planning significant purchases of Huawe…
-
DeepSeek builds Inner Mongolia data center, challenging Nvidia
DeepSeek is constructing a significant data center in Inner Mongolia, utilizing Huawei Ascend architecture. This move is positioned as a challenge to Nvidia's current dominance in the AI hardware market.
-
CogEvol models generate learning environments efficiently and reliably
A new family of models called CogEvol has been developed for generating learning environments, transforming course briefs into structured JSON slides or interactive HTML pages in a single pass. These models achieve high…
-
Huawei challenges AI competition with proprietary Ascend chips for drug discovery
Huawei is challenging Western competitors by deploying AI platforms built on its proprietary Ascend chips. These platforms are designed to significantly accelerate key stages in drug discovery, potentially shaving off o…
-
Xiaohongshu open-sources MoE large language model dots3-note
Xiaohongshu has open-sourced its large language model, dots3-note, a Mixture-of-Experts model with 280 billion total parameters and 16 billion activated parameters. The model boasts a 512K context window and supports te…
-
SMIC posts record quarter amid sanctions-driven AI demand and price hikes
SMIC, China's leading foundry, has reported a record quarter with $3 billion in revenue, driven by a significant increase in wafer prices and high utilization rates. This surge is largely attributed to U.S. sanctions th…
-
Huawei open-sources AscendNPU-IR for AI processor optimization
Huawei has released AscendNPU-IR, an open-source intermediate representation tool built on MLIR. This tool is designed to enhance the compilation and optimization of operators for Ascend AI processors, aiming to improve…
-
Nvidia H200 AI chips reach China, but Hong Kong deployment faces hurdles · 3 sources tracked
Nvidia has begun shipping its H200 AI accelerators to China, with major tech companies ByteDance and Tencent reportedly receiving initial batches of approximately 10,000 units each. Despite these shipments, Beijing is d…
-
New ASCEND framework optimizes federated learning with adaptive compression
Researchers have developed ASCEND, a novel framework for federated knowledge distillation (FedKD) that addresses communication overhead and model heterogeneity. This approach allows individual clients to select optimal …
-
Qwen3-8B model shows 15% token agreement on Ascend vs CUDA hardware
A technical comparison revealed significant differences in output between the Qwen3-8B model when run on Huawei Ascend hardware using DashScope versus NVIDIA GPUs with CUDA. Across 61 prompts, there was only a 15% agree…
-
Ruitong launches to measure LLM accuracy drift across hardware
A new initiative called Ruitong has been launched to address the issue of accuracy drift in large language models (LLMs) when migrated across different hardware. While latency and cost are well-understood, the impact on…
-
Virgin Galactic delays new spaceship flights to 2027
Virgin Galactic has delayed the first commercial flight of its new Delta-class spaceship to February 2027, citing numerous small installation issues during assembly. CEO Michael Colglazier stated that hundreds of minor …
-
New proxy models slash LLM debugging costs by up to 33x
Researchers have developed a method to create "proxy models" for large language models (LLMs) that significantly reduce the computational cost and time required for debugging and reproducing failures during reinforcemen…
-
openJiuwen launches enterprise AI agent swarm architecture, lands first financial deployment
openJiuwen has released an enterprise-grade distributed swarm architecture for its AI Agent platform, JiuwenSwarm. This new architecture addresses key challenges in scaling AI agents for enterprise use, including managi…