[#Huawei](<https://hardwarebrief.com/hashtag/huawei>) #Ascend960 [#AIChips](<https://hardwarebrief.com/hashtag/aichips>) [#Supercomputing](<https://hardwarebrief.com/hashtag/supercomputing>) [#Semiconductors](<https://hardwarebrief.com/hashtag/semiconductors>)



# Huawei Unveils Ascend AI Chip Roadmap Through 2029

September 18, 2026 at 08:52 AM

> Huawei announces the Ascend 960 and 980 AI chips, focusing on HBM capacity and optical interconnects for supercomputing architectures.

**Huawei** unveiled a new AI chip roadmap at the Huawei Connect 2026 trade fair in Shanghai, introducing the Ascend 960 and its successors. The Ascend 960, featuring 288 GByte of HBM, is set for Q1 2027, followed by the FP4-focused Ascend 960PR in Q3 2027.

The roadmap extends into 2029 with the newly announced Ascend 980, which aims to double the performance of the 2028 Ascend 970. Development for the 980 focuses on advanced interconnects for data exchange, utilizing optical connections for hardware scaling.

These advancements support Huawei's PoD and SuperPoD rack architectures for supercomputing. By accelerating its release schedule, Huawei aims to enhance accelerator performance through 2029, while a 2026 blind test recently evaluated nine mobile devices.



## Related news

[#Nvidia](<https://hardwarebrief.com/hashtag/nvidia>) [#HPC](<https://hardwarebrief.com/hashtag/hpc>) #GraceBlackwell [#DGXSpark](<https://hardwarebrief.com/hashtag/dgxspark>) [#Supercomputing](<https://hardwarebrief.com/hashtag/supercomputing>)

November 17, 2026 at 11:18 AM

## [NVIDIA Unveils Multi-Node HPC Developer Kit](<https://hardwarebrief.com/article/nvidia-unveils-multi-node-hpc-developer-kit-29a5fc50>)

> The new desk-side cluster links four DGX Spark systems with Grace Blackwell Superchips to enable local testing of complex HPC workloads.

[#Nvidia](<https://hardwarebrief.com/hashtag/nvidia>) #RubinGPU #SC26 [#HPC](<https://hardwarebrief.com/hashtag/hpc>) [#Supercomputing](<https://hardwarebrief.com/hashtag/supercomputing>)

November 16, 2026 at 01:57 PM

## [NVIDIA Unveils Rubin HPC GPU with 50,000 Cores](<https://hardwarebrief.com/article/nvidia-unveils-rubin-hpc-gpu-with-50-000-cores-2df2a9d9>)

> NVIDIA debuts the Rubin architecture at SC26, featuring a high-performance GPU designed for massive science and engineering workloads.

[#Intel](<https://hardwarebrief.com/hashtag/intel>) #Lawsuit #TechStocks [#Semiconductors](<https://hardwarebrief.com/hashtag/semiconductors>) #Shareholders

November 16, 2026 at 11:01 AM

## [Intel Executives Sued Over PC Sales Slowdown](<https://hardwarebrief.com/article/intel-executives-sued-over-pc-sales-slowdown-07e3868c>)

> Shareholders filed a class-action lawsuit against Intel's CEO and CFO, alleging the company concealed a market slump and inventory glut.

[#AIChips](<https://hardwarebrief.com/hashtag/aichips>) [#Nvidia](<https://hardwarebrief.com/hashtag/nvidia>) [#ExportControls](<https://hardwarebrief.com/hashtag/exportcontrols>) #NationalSecurity #USChina

November 16, 2026 at 10:58 AM

## [US Congress Tightens AI Chip Export Controls](<https://hardwarebrief.com/article/us-congress-tightens-ai-chip-export-controls-4ef24a42>)

> Lawmakers investigate Singapore-based Megaspeed for illegal chip transfers to China while proposing new foreign data center restrictions.

#NVIDIABlackwell #ChineseAILabs [#ExportControls](<https://hardwarebrief.com/hashtag/exportcontrols>) #DeepSeek [#AIChips](<https://hardwarebrief.com/hashtag/aichips>)

November 14, 2026 at 11:12 AM

## [Report: Chinese AI Labs Used NVIDIA Blackwell GPUs](<https://hardwarebrief.com/article/report-chinese-ai-labs-used-nvidia-blackwell-gpus-71e93872>)

> Uemployment Labs forensic analysis claims Chinese AI labs used thousands of export-controlled NVIDIA Blackwell GPUs for model training.

[#TSMC](<https://hardwarebrief.com/hashtag/tsmc>) [#CyberSecurity](<https://hardwarebrief.com/hashtag/cybersecurity>) #DataBreach #Ransomware [#Semiconductors](<https://hardwarebrief.com/hashtag/semiconductors>)

November 13, 2026 at 12:33 PM

## [TSMC Cloud Provider Hack Exposes Employee Data](<https://hardwarebrief.com/article/tsmc-cloud-provider-hack-exposes-employee-data-3e90058d>)

> TSMC confirms a cloud provider breach exposed 6.9GB of data on 46,000 employees as hackers demand a three-Bitcoin ransom.

![Rapidus Targets 2-nm Chip Mass Production by 2027](<https://hardwarebrief.com/images/2027_ben_indul_a_japan_2_nanometeres_chipgyartas_85344ab0_997693fb8e.webp>)

#Rapidus [#Semiconductors](<https://hardwarebrief.com/hashtag/semiconductors>) #JapanTech [#IBM](<https://hardwarebrief.com/hashtag/ibm>) #ChipManufacturing

November 12, 2026 at 04:11 PM

## [Rapidus Targets 2-nm Chip Mass Production by 2027](<https://hardwarebrief.com/article/rapidus-targets-2-nm-chip-mass-production-by-2027-a95a4e3d>)

> Japanese startup Rapidus aims to revitalize domestic silicon manufacturing by producing advanced 2-nm chips using IBM technology by 2027.

#Infineon #SiliconCarbide [#Semiconductors](<https://hardwarebrief.com/hashtag/semiconductors>) #PowerElectronics [#TechInnovation](<https://hardwarebrief.com/hashtag/techinnovation>)

November 9, 2026 at 03:45 PM

## [Infineon Produces First 300 mm SiC Power Wafers](<https://hardwarebrief.com/article/infineon-produces-first-300-mm-sic-power-wafers-835bd598>)

> Infineon Technologies achieves a milestone by producing functional power semiconductors on 300 mm silicon carbide wafers in Austria.

![Terafab Targets 1M Monthly Wafers by 2032](<https://hardwarebrief.com/images/terafab_targets_1m_monthly_wafers_by_2032_c965686b_5824da2b1a.webp>)

[#Terafab](<https://hardwarebrief.com/hashtag/terafab>) [#ElonMusk](<https://hardwarebrief.com/hashtag/elonmusk>) [#Semiconductors](<https://hardwarebrief.com/hashtag/semiconductors>) [#AIChips](<https://hardwarebrief.com/hashtag/aichips>) [#Intel](<https://hardwarebrief.com/hashtag/intel>)

November 9, 2026 at 01:44 PM

## [Terafab Targets 1M Monthly Wafers by 2032](<https://hardwarebrief.com/article/terafab-targets-1m-monthly-wafers-by-2032-c965686b>)

> SpaceX and Tesla's $16.8 billion Texas venture aims to produce 1 terawatt of AI compute annually using Intel's advanced 14A process.

[#Nvidia](<https://hardwarebrief.com/hashtag/nvidia>) [#JensenHuang](<https://hardwarebrief.com/hashtag/jensenhuang>) [#AIChips](<https://hardwarebrief.com/hashtag/aichips>) #NationalSecurity #DOJ

November 8, 2026 at 05:09 PM

## [Nvidia CEO Jensen Huang Faces House Briefing](<https://hardwarebrief.com/article/nvidia-ceo-jensen-huang-faces-house-briefing-48b5f847>)

> House Republicans have invited Jensen Huang to address national security concerns following a leaked memo alleging ties to a Chinese asset.
