Reports
AI-generated structured vendor updates
Qualcomm获批AI芯片自动调优专利 应对制造差异
...
Huawei to Launch Ascend 950DT AI Chip with Native FP8, Doubling Compute Power
Huawei announced the upcoming Ascend 950DT AI chip, doubling compute power versus its predecessor with native FP8 support, improved memory bandwidth, and inter-chip interconnect. Huawei Cloud operates large AI clusters across China, with over 100,000 Ascend accelerators powering autonomous driving training, signaling a push to build a domestic AI ecosystem alternative to NVIDIA.
MediaTek Approves $5B for Data Center AI Chips, First Custom Accelerator to Tape Out in Q4
MediaTek's board has approved a $5 billion self-funded budget to expand into data center AI chips. Its first custom AI accelerator will tape out in Q4 2025, targeting over $2 billion revenue by 2026 and 15-20% market share by 2027. This signals a strategic pivot from mobile SoCs to AI infrastructure.
Intel's 18A Process Enters Mass Production with Strong Yield, $20B CapEx for AI Chips
Intel's 18A process has entered mass production with yields exceeding expectations. The company raised its 2026 CapEx to over $20B for AI chip manufacturing. Data center and AI revenue grew 59% YoY to $6.3B, now 70% of total revenue, signaling Intel's strategic pivot to AI infrastructure.
TSMC Develops EMIB-Like Packaging with Kinsus to Counter Intel's CoWoS Threat
TSMC, in response to CoWoS capacity constraints and potential customer loss, is developing an EMIB-like advanced packaging technology with Kinsus, aiming to simplify manufacturing steps and reduce costs, preventing clients from shifting to Intel's packaging platform, highlighting the strategic importance of advanced packaging in AI chip competition.
Google Guarantees Anthropic Data Center, Deploys TPUs to Challenge NVIDIA's GPU Dominance
Anthropic is building a $15B AI data center in Texas, backed by Google's financial guarantee in exchange for 20% equity. The facility will deploy Google's TPU chips, co-designed with Broadcom, signaling a shift in AI compute landscape and challenging NVIDIA's GPU dominance.
TSMC Accelerates 1.4nm Process, Partners with Unimicron on EMIB-Like Packaging to Counter Intel
TSMC accelerates its 1.4nm (A14) process to mid-2028 mass production, using second-gen GAA NanoFlex Pro transistors and existing EUV lithography. It also partners with Unimicron on EMIB-like packaging to alleviate CoWoS capacity constraints and prevent AI chip customer loss to Intel's packaging platform.
TSMC Develops EMIB-like Packaging to Challenge Intel and Ease CoWoS Bottleneck
TSMC is developing an EMIB-like advanced packaging technology that uses small silicon bridges instead of large silicon interposers, reducing cost and complexity while easing CoWoS capacity constraints. This move counters Intel's EMIB technology and addresses the growing packaging demands of AI accelerators, securing TSMC's dominance in advanced packaging.
NVIDIA Employee Detained in Taiwan B300 Smuggling Probe, Export Control Reaches Chipmaker
Taiwan prosecutors detain a NVIDIA employee for allegedly helping smuggle Super Micro servers with **GB300** chips to China. Seven people are held, involving $21 million. This marks the first time export control enforcement directly hits a chipmaker, shifting compliance risk upstream.
NVIDIA GPU套装全线涨价涉及GDDR7与GDDR6显存
...
Alibaba Cloud Unveils Agent-Native Suite and Open-Source SAIL Stack to Rival CUDA
At WAIC 2026, Alibaba Cloud launched its Agent-Native cloud suite (AgentLoop, AgentTeams, AgentRun, TokenWorks), open-sourced the T-Head SAIL AI software stack, and unveiled the 2.4T-parameter Qwen 3.8-Max-Preview model and Zhenwu M890 supernode, marking a comprehensive push into agent-native cloud and open-source AI ecosystem.
NVIDIA and SK Group Lock HBM4 Supply and Launch Sovereign AI Factory Model with $500B+ Deal
NVIDIA and SK Group announced a $500B+ AI partnership including a 2GW AI factory using Vera Rubin and HBM4, long-term HBM4 supply lock, and a $1B NVIDIA investment in Naver (with $9B from Brookfield). Samsung and Broadcom signed a $200B deal. This signals a new era of sovereign AI infrastructure and supply chain deep-locking.
NVIDIA Prepay $1.5B to Amkor for US 2nm/3nm/HBM4 Advanced Packaging
NVIDIA prepays $1.5B to Amkor to expand advanced packaging capacity in Arizona, covering 2nm/3nm/HBM4. This onshores AI chip packaging, complementing Wistron's system integration, to reduce reliance on Taiwan and secure supply for Vera Rubin and Blackwell Ultra.
Intel Accelerates 14A to 2027H2 Risk Production, 18A Yield Exceeds Target by 25%, Capex Raised to $20B
Intel announced 14A process acceleration to risk production in 2027H2 with HVM in 2028, 18A yield exceeding target by 25% supporting Panther Lake volume, and raising 2026 capex to $20B, signaling an aggressive foundry push. Advanced packaging EMIB-T becomes a profit pillar.
Tiered AI Chip Market Emerges as US Allows H200 Exports to China with 25% Levy
The US Commerce Department approved NVIDIA H200 exports to China with a 25% sales tax, while maintaining a ban on Blackwell. This formalizes a tiered AI chip market, making H200 the best available imported chip for China, but the performance gap and added tax burden increase deployment costs and complexity for Chinese AI infrastructure.
三星电子追加46万亿韩元投资扩建温阳HBM内存生产设施
...
AMD Invests $5B in Anthropic, Secures 2GW MI450 Deployment, Reshaping AI Compute Ecosystem
AMD and Anthropic announce a strategic partnership: Anthropic will deploy up to 2GW of AMD Instinct MI450 GPUs, with AMD investing up to $5B in Anthropic. They will collaborate on ROCm optimization and Claude workload tuning, marking AMD's transition from chip vendor to AI ecosystem investor and accelerating multi-sourcing in AI compute.
Apple自研AI服务器芯片Baltra遇性能瓶颈可能推迟至2027年
...
华为昇腾950超节点亮相WAIC 2026，单柜64卡支持8192卡高速互联
...
Google's Frozen v2 Chip Hardwires Gemini Architecture for 6-10x TPU Efficiency, Set for 2028
Google is developing Frozen v2, a dedicated AI chip that hardwires the Gemini model architecture into silicon for 6-10x energy efficiency per token over current TPUs. It is a new product line, planned for 2028, with weight update flexibility but a frozen architecture. This validates the industry shift from general-purpose GPUs to dedicated ASICs for AI inference.