CONNECT WITH US
NEWS TAGGED TRAINING
Tuesday 31 March 2026
Chinese GPU maker Moore Threads wins US$91 million AI cluster order

Chinese GPU developer Moore Threads has secured a CNY660 million (approx. US$95.5 million) contract to supply its KUAE intelligent computing...

Tuesday 31 March 2026
AI computing shifts from training to inference; heterogeneous architectures go mainstream
As generative AI continues to advance, its capabilities and application scenarios are rapidly expanding, driving structural changes in computing infrastructure. At AI EXPO Taiwan 2026,...
Monday 30 March 2026
Kioxia outlines supply strategy, stresses new 2027 capacity won't disrupt NAND market balance
As generative AI (GenAI) shifts from large-model training to large-scale inference, Japanese NAND giant Kioxia forecasts continued price increases for NAND and SSD products. Despite...
Monday 30 March 2026
AI meets reality: why physical data is the real constraint

Artificial intelligence is entering a new phase centered on agentic systems, with momentum shifting from digital environments to real-world...

Monday 30 March 2026
AI compute shifts to inference, reshaping data center bottlenecks

The focus of artificial intelligence computing is set to shift from training to inference beyond 2025, a transition that will also redefine...

Thursday 26 March 2026
AI Expo 2026: Surging costs, memory crunch drive shift toward inference, researcher says

Artificial intelligence is entering a new phase in which inference, rather than training, is becoming the dominant driver of computing...

Thursday 26 March 2026
MediaTek advances Taiwanese language AI amid linguistic and security challenges
MediaTek highlighted at AI EXPO Taiwan 2026 how language-specific challenges complicate the deployment of AI globally, as Taiwanese tonal variety, mixed writing systems, and local...
Thursday 26 March 2026
Tescan Opens Korea Demo Lab for APAC Semiconductor and Academia

Tescan  announced the official opening of its upgraded Tescan Korea Demo Lab & Office in Seoul, bringing together its relocated Korea office and enhanced demo...

Tuesday 24 March 2026
Techman's product shift and semiconductor focus could sustain growth despite macro uncertainty
Techman Robot plans to release a dual-arm robot and deploy an upgraded humanoid model at customer sites in the second half of 2026, moves the company says will broaden its addressable...
Tuesday 24 March 2026
Huawei's Ascend 950PR debuts with nearly 3x H20 performance, targets China's AI compute gap
Huawei has officially launched its Ascend 950PR processor and introduced the Atlas 350 AI accelerator card equipped with this chip, marking the commercial debut of its next-generation...
Monday 23 March 2026
Explainer: Why Nvidia's Groq LPU runs on Samsung silicon— Groq's scale and inference strategy
Nvidia CEO Jensen Huang highlighted at GTC 2026 that AI has shifted from early model training to an era defined by inference and agent computing. To meet growing inference demands,...
Thursday 19 March 2026
Nvidia positions Groq 3 LPUs alongside Vera Rubin for an inference-first era
The 2026 Nvidia GTC keynote signaled a clear industry shift toward inference. CEO Jensen Huang declared that "training is just the beginning — inference is the core battleground...
Thursday 19 March 2026
Nvidia GTC 2026: Techman Robot showcases motion training tech
Techman Robot demonstrated its humanoid robot TM Xplore I, and advanced motion training technology at Nvidia GTC 2026. The company announced a partnership with motion capture specialist...
Thursday 19 March 2026
Column: Cheap tokens are masking an expensive problem
The cost of running an AI query has fallen by roughly 99% over the past two years. That should be a story about savings. Instead, it's a story about demand.
Wednesday 18 March 2026
Analysis: How Nvidia is leveraging DeepSeek R1 surge to cement hardware–model domination
On March 11 (US time), Nvidia announced a US$26 billion five-year investment in open-source large language models and introduced Nemotron 3 Super, its most powerful hybrid mixture-of-experts...