AI Chips

Twitter

Facebook

Copy Link

Key Facts

Key ChipsH100, B200, TPU v5, Trainium

Key PlayersNVIDIA, AMD, Broadcom, Cerebras

Top News

Google Unveils TPU 8tx and 8i Max: Specialized AI Chips for Secure and Long-Context Workloads

Google introduces two specialized Tensor Processing Units—TPU 8tx for confidential computing and TPU 8i Max for massive context windows. The eighth-generation AI chip variants target regulated industries and agentic AI custom silicon needs, marking the first time Google has released four TPU systems in a single year.

Technology6 sources

Google Unveils TPU 8tx and 8i Max: Specialized AI Chips for Secure and Long-Context Workloads

Technology6 sources

NVIDIA Partners with Rebellions and SK Telecom to Advance Korea's Sovereign AI Infrastructure

NVIDIA announced a collaboration with Rebellions and SK Telecom to build Korea's sovereign AI infrastructure using NVLink Fusion technology. The partnership connects Korean-designed neural processing units with NVIDIA's platform to power AI data centers and services tailored to Korea's language and industries.

Technology4 sources

NVIDIA Partners with Rebellions and SK Telecom to Advance Korea's Sovereign AI Infrastructure

Technology4 sources

Alphabet Invests $548 Million in Qualcomm as AI Data Centre Push Gains Momentum

Alphabet disclosed a $548 million stake in Qualcomm as the chipmaker accelerates its expansion into AI data centres. Qualcomm targets $15 billion in data centre revenue by 2029, backed by major deals with Amazon, Microsoft, and Meta as tech giants seek alternatives to Nvidia's expensive processors.

Business2 sources

Alphabet Invests $548 Million in Qualcomm as AI Data Centre Push Gains Momentum

Business2 sources

TSMC Denies Exploring 3D DRAM Production Despite Reports of Nvidia Request Amid AI Memory Crunch

Reports suggest TSMC is drafting plans for 3D DRAM manufacturing at Nvidia's request as AI-driven memory shortage intensifies. However, TSMC denies the claims, reaffirming its focus as a pure-play logic foundry. The move would pit the world's largest chipmaker against Samsung, SK Hynix, and Micron in advanced memory chips production.

Business3 sources

TSMC Denies Exploring 3D DRAM Production Despite Reports of Nvidia Request Amid AI Memory Crunch

Business3 sources

Microsoft Ends Oracle AI Computing Deal as It Reassesses $80B in Cloud Partnerships

Microsoft is winding down its 2023 deal to rent AI computing power from Oracle after declining to renew the contract. The rented servers ran older Nvidia chips that no longer matched Microsoft's evolving needs. This move is part of a broader reassessment of over $80 billion in external cloud contracts, signaling a strategic shift in how the tech giant manages its AI infrastructure investments.

Business3 sources

Microsoft Ends Oracle AI Computing Deal as It Reassesses $80B in Cloud Partnerships

Business3 sources

Nvidia eyes 130+ telecom buildings for AI data centers to deliver 1 gigawatt compute capacity

Nvidia is negotiating with Brightspeed to transform over 130 existing telecom buildings across 20 US states into compact AI data centers. The proposal could deliver 1 gigawatt of compute capacity and accommodate more than 2,200 Supermicro Vera Rubin NVL72 racks, offering a faster alternative to building new facilities from scratch.

Technology4 sources

Nvidia eyes 130+ telecom buildings for AI data centers to deliver 1 gigawatt compute capacity

Technology4 sources

NVIDIA Vera CPU Achieves 2x Faster Agent Speed and 1.6x More Agents Per Watt Than Leading x86 Processors

NVIDIA Vera CPU is disrupting the Agentic AI segment with industry-leading performance. The chip delivers 2x faster agent speed than AMD's 256-core EPYC Venice and 1.6x more agents per watt than the 96-core frequency-optimized variant, while early adopters report production gains of up to 2.2x on agentic orchestration workloads.

Technology2 sources

NVIDIA Vera CPU Achieves 2x Faster Agent Speed and 1.6x More Agents Per Watt Than Leading x86 Processors

Technology2 sources

AI Boom Turns Chip Packaging Into Critical Battleground as Companies Invest Billions

The global race to build AI infrastructure has transformed advanced chip packaging from an overlooked process into a critical bottleneck. With demand for AI chips outstripping capacity, companies like Nvidia and AMD are investing billions to expand production, while TSMC's dominance faces challenges from emerging alternatives.

Business3 sources

AI Boom Turns Chip Packaging Into Critical Battleground as Companies Invest Billions

Business3 sources

Nvidia Opens Investigation as Jensen Huang Faces Chinese Spy Allegations Linked to $2B Chip Smuggling

US intelligence believes Nvidia CEO Jensen Huang had a close relationship with Alice Huang, an executive at Megaspeed suspected of being recruited by China's Ministry of State Security. Nvidia has launched an internal investigation as the Singapore-based firm faces accusations of smuggling $2 billion in restricted AI chips to China, raising critical national security concerns.

Policy20 sources

Nvidia Opens Investigation as Jensen Huang Faces Chinese Spy Allegations Linked to $2B Chip Smuggling

Policy20 sources

AI Chip Boom Redirects Semiconductor Supply Chain, Leaving Smartphones and PCs Behind

The AI boom is fundamentally reshaping the semiconductor industry as manufacturers prioritize AI chip production for data centers over traditional markets like smartphones and PCs. Siltronic CEO reveals chip supply is being redirected based on profit margins, while specialty chemicals and industrial gas makers report double-digit growth from semiconductor manufacturers expansion.

Business2 sources

AI Chip Boom Redirects Semiconductor Supply Chain, Leaving Smartphones and PCs Behind

Business2 sources

Dell Overtakes HPE as On-Prem Server Sales for HPC and AI Surge 20% to $16.5 Billion

Dell has surpassed HPE to become the leading HPC vendor by revenue, capturing 24% market share with $4.2 billion in the first half of 2026. The on-premises HPC and AI-for-Science server market reached $16.5 billion, up 19.9% year-over-year, driven by AI processing demands and costly GPU components despite unit shipments rising only 4.4%.

Business2 sources

Dell Overtakes HPE as On-Prem Server Sales for HPC and AI Surge 20% to $16.5 Billion

Business2 sources

Arm Holdings Beats Revenue Estimates as AI Demand and Data Center Growth Offset Smartphone Weakness

Arm Holdings reported second-quarter revenue of $1.44 billion, surpassing analyst estimates as AI demand and data center deployments drove licensing revenue to $773 million. The company raised its AGI CPU sales target to $2 billion across fiscal 2027-2028, though royalty growth slowed to 8% amid smartphone market weakness.

Business7 sources

Arm Holdings Beats Revenue Estimates as AI Demand and Data Center Growth Offset Smartphone Weakness

Business7 sources

RTX Spark AI Laptops Deliver Real Local AI Power, But Snapdragon X2 Offers Budget Alternative

Nvidia's RTX Spark laptops bring serious local AI processing to Windows with up to 128GB unified memory and powerful Blackwell GPUs. Early tests show the Microsoft Surface Laptop Ultra and HP OmniBook Ultra excel at running large AI models locally, though the Snapdragon X2 Elite Extreme offers compelling performance at half the cost.

Technology4 sources

RTX Spark AI Laptops Deliver Real Local AI Power, But Snapdragon X2 Offers Budget Alternative

Technology4 sources

Navitas Semiconductor Stock Surges as AI Infrastructure Business Set to Triple in 2027

Navitas Semiconductor shares jumped after reporting Q3 revenue of $13.99 million, beating estimates, and forecasting AI infrastructure to account for over 40% of total revenue by end of 2026. CEO Chris Allexandre expects the AI infrastructure business to triple in 2027, driven by partnerships with NVIDIA and expanding high-power solutions.

Business2 sources

Navitas Semiconductor Stock Surges as AI Infrastructure Business Set to Triple in 2027

Business2 sources

DeepSeek to train V5 AI model on 160,000 Huawei Ascend 950DT chips in full domestic stack shift

DeepSeek plans to train its upcoming V5 AI model using 160,000 Huawei Ascend 950DT chips at a new data center in Ulanqab, Inner Mongolia. This marks China's first attempt at domestic full-stack AI model training, reducing dependence on Nvidia hardware while testing the TileLang compiler ecosystem.

Technology2 sources

DeepSeek to train V5 AI model on 160,000 Huawei Ascend 950DT chips in full domestic stack shift

Technology2 sources

Prev12 3 4…110 Next