CONNECT WITH US
NEWS TAGGED AI INFERENCE
Wednesday 1 July 2026
OpenAI engineers claim to discover way to cut inference costs in half
OpenAI engineers claim to have figured out a way to halve the costs of inference using its models, according to The Information. The development comes as AI model developers...
Wednesday 1 July 2026
AI chip startup Rebellions' acquisition of SqueezeBits signals push beyond hardware

South Korean AI chip designer Rebellions said on June 30 that it is acquiring AI inference optimization company SqueezeBits, as part...

Monday 29 June 2026
Qualcomm stirs AI data center competition with CPUs, ASICs, and accelerators
Qualcomm officially unveiled its Dragonfly data center platform at this week's annual investor day, laying out a four-pronged push into cloud AI that spans SerDes, PAM4 DSP, and other...
Thursday 25 June 2026
SambaNova targets US$10B valuation as demand rises for cheaper AI inference
AI chipmaker SambaNova could raise between US$800 million and US$1 billion in a new funding round, according to its executive chairman and Intel CEO Lip-Bu Tan. This would raise SambaNova's...
Thursday 25 June 2026
ByteDance's reported AI chip orders mark breakthrough for Chinese GPU maker Iluvatar CoreX
ByteDance's reported plan to purchase at least 50,000 AI inference chips from Shanghai-based GPU developer Iluvatar CoreX could become one of the most significant commercial wins yet...
Thursday 25 June 2026
Nvidia and AWS deepen push to simplify AI infrastructure at scale
Nvidia and Amazon Web Services (AWS) are expanding tools that could make it easier for companies worldwide to build and run large-scale AI systems. The changes aim to improve speed,...
Wednesday 24 June 2026
Cerebras bets on speed as a product, racing to add data center capacity through 2027
Cerebras Systems used its first earnings call on June 26 to argue that speed is its core advantage and that the entire AI inference market is addressable. It also detailed an aggressive...
Friday 19 June 2026
Kioxia keeps NAND capex in check as AI inference reshapes SSD demand

Kioxia, one of the world's major NAND flash memory suppliers, is prioritizing long-term agreements (LTAs), BiCS migration, and domestic...

Tuesday 16 June 2026
Nvidia reportedly tightens grip on AI inference market despite growing competition

Nvidia's dominance in AI is extending beyond model training and deeper into inference—the fast-growing segment of the AI market responsible...

Tuesday 9 June 2026
Tencent takes dual-track AI chip route with Canghai V2 and domestic partnerships
Tencent is sharpening a dual-track AI chip strategy, combining self-developed semiconductors for its own business workloads with deeper partnerships across China's domestic AI computing...
Friday 5 June 2026
Sambanova challenges GPU dominance in AI inference at Computex
SambaNova used a Computex 2026 session on June 4 to make its most public case yet that the GPU-only approach to AI inference is hitting a fundamental wall — and to demonstrate,...
Friday 5 June 2026
Foxconn deepens AI push with Intel on inference racks
Intel and Foxconn have signed a memorandum of understanding (MoU) to cooperate on AI rack infrastructure, edge AI, physical AI platforms, and custom chip design services. The deal...
Wednesday 3 June 2026
AI inference and agents push data centers to secure on-site power and storage
At COMPUTEX 2026, energy and data center executives warned that the industry shift from AI training to inference and agentic AI has driven a sharp rise in electricity demand and tightened...
Tuesday 2 June 2026
Intel unveils Xeon 6+ to power agentic AI inference, challenges GPU-centric infrastructure
Intel is expanding its data center portfolio with new Xeon 6+ processors, Ethernet E835 networking products, and fresh details on its Crescent Island AI accelerator, positioning the...
Monday 1 June 2026
ByteDance reportedly developing Groq-style chip with InnoStar
ByteDance is creating a new chip similar to those made by Nvidia partner Groq to help the Chinese creator of TikTok handle its AI inference loads, according to The Information...