Google adds two new specialized chips for AI agents and security to 8th-gen TPU lineup — TPU 8i Max puts two chips on one package for long-context work, TPU 8tx swaps Arm hosts for AMD EPYC
Google introduced TPU 8tx for confidential inference and TPU 8i Max for long-context agentic work. [[fact:fact-5esdbfebw]] TPU 8i Max combines two 8i chips in one package. [[fact:fact-m7px9u1v5]] Google says 8i Max doubles the original 8i's compute, SRAM, and HBM bandwidth. [[fact:fact-fq91aotpg]] TPU 8tx uses a fifth-generation AMD EPYC CPU instead of the Arm-based Axion hosts used by 8t and 8i. [[fact:fact-113vseqad]] The chip is optimized for secure and confidential inference as well as small-scale training. [[fact:fact-wxlr6sdw1]] Google Cloud's TPU page lists both systems as coming soon, while the article provides no availability date. [[fact:fact-cj88zxlyi]][[fact:fact-xkwkb71r4]]
ANR Opens Beta for Cross-Platform AI Frame Generation
Apex Neural Rendering (ANR) has launched a beta suite that includes AI frame generation, super resolution, and ray reconstruction. [[fact:fact-y0ytm8l40]]Only frame generation is currently open for testing; super resolution and ray reconstruction are expected to roll out later. [[fact:fact-27ghji03t]]ANR claims compatibility with multiple generations of Nvidia, AMD, and Intel graphics cards. [[fact:fact-rabtzcwya]]The ANR FG beta is available on GitHub and currently requires OptiScaler to run. [[fact:fact-xejqaqqgt]]Apex Compute's unverified test reported that, at 4K on an Nvidia GeForce RTX 5090, ANR 2x frame generation took 1.01 ms per frame versus 1.26 ms for Nvidia DLSS frame generation. [[fact:fact-4553dthuc]]The same test reported 0.61 ms for ANR and 0.82 ms for DLSS in 4x mode. [[fact:fact-alw2agfsf]]
Supermicro levert nu servers van de volgende generatie met AMD EPYC™ 9006-serieprocessoren van de 6e generatie voor cloud-, HPC- en agentische AI-workloads
Supermicro announced that its H15 server portfolio is now being delivered. [[fact:fact-a8jqy8dth]] The H15 portfolio is powered by 6th-generation AMD EPYC 9006-series processors, formerly codenamed Venice. [[fact:fact-86jtt7rtb]] The immediately available portfolio includes systems optimized for cloud, HPC, and agentic AI workloads. [[fact:fact-ql8nr1h82]] Supermicro says H15 systems on its 2U Hyper and CloudDC platforms have achieved multiple SPEC CPU world records. [[fact:fact-r3at5vqx7]]
Advanced Micro Devices vs. Marvell Technology: Which Artificial Intelligence Stock Is a Better Buy Going into 2027?
AMD designs high-performance computing products for data centers, gaming and embedded systems. [[fact:fact-zam55ohm7]] AMD secured a multiyear partnership with OpenAI to deploy its Instinct MI450 series AI accelerators in data centers. [[fact:fact-f49m6qtzv]] AMD fiscal 2025 revenue was $34.6 billion, up 34.3% from the prior fiscal year. [[fact:fact-jbx3nwmjk]] Marvell is a fabless semiconductor company that designs hardware for data centers and communications networks. [[fact:fact-tiw9tdrnp]] Marvell fiscal 2026 revenue was $8.2 billion, up 42.1% year over year. [[fact:fact-3qr6rvmue]] Marvell expanded its portfolio through the acquisition of Celestial AI and an agreement to acquire Lumentum Holdings. [[fact:fact-7kvgiqsir]]
PsiQuantum and AMD Broaden Collaboration Across Quantum, HPC and AI
PsiQuantum and AMD Broaden Collaboration Across Quantum, HPC and AI(摘要待生成)
Nvidia vs. AMD: Which Is the Better Artificial Intelligence (AI) Stock for 2027 After the Tech Sell-Off?
Nvidia vs. AMD: Which Is the Better Artificial Intelligence (AI) Stock for 2027 After the Tech Sell-Off?(摘要待生成)
Nvidia Turmoil Coincides With Gains in AMD and Broadcom Shares
Nvidia is experiencing leadership turmoil, and its shares have fallen sharply this week. [[fact-s07i7craf]] AMD rose 7.32% on Wednesday and gained more than 4.3% again in after-hours trading. [[fact-1kjckmyc7]] Broadcom rose nearly 4% on Wednesday and 2.36% after hours. [[fact-lloqmu3un]]
Baird Cuts Nvidia to Neutral, Slashes Price Target to $200
Nvidia shares fell 7.4% on Tradegate versus the previous day's close. [[fact-vxzb4s40v]] Nvidia's board met Wednesday to decide whether to remove CEO Jensen Huang. [[fact-g6zpemjd8]] Baird downgraded Nvidia to Neutral and cut its price target from $500 to $200, citing uncertainty over data-center spending, potentially weakening demand, and whether Huang would remain CEO next year. [[fact-xby3hsume]] At least three major banks, including Wells Fargo and Citi, also lowered their Nvidia price targets. [[fact-3df75s6hj]]
AMD Stock: What Does Its $2.5 Billion Anthropic Bet Buy? - Yahoo Finance
AMD Stock: What Does Its $2.5 Billion Anthropic Bet Buy? - Yahoo Finance(摘要待生成)
AMD CEO Lisa Su on data center backlash: Tech has to sell the vision
AMD CEO Lisa Su on data center backlash: Tech has to sell the vision Yahoo Finance
BofA Raises Its 2030 Server CPU Market Forecast
Bank of America raised its forecast for the potential global server CPU market in 2030 from about $210 billion to more than $240 billion. [[fact:fact-cq2t9a5r2]] BofA analysts expect the market to grow from about $35 billion in 2025 to $64.1 billion in 2026, reaching $240.2 billion in 2030. [[fact:fact-jc2edg5hv]] They project a 39.2% compound annual growth rate from 2026 to 2030, up from an earlier estimate of about 36%. [[fact:fact-tog9t1dox]] BofA sees AMD as a potential major beneficiary of the server CPU market's growth. [[fact:fact-9jnepca23]]
AMD 发布肾上腺素 26.11.1 可选驱动:修复 SSD 异常写入问题,公布多项已知游戏兼容问题
AMD 发布肾上腺素 26.11.1 可选驱动:修复 SSD 异常写入问题,公布多项已知游戏兼容问题(摘要待生成)
AMD与PsiQuantum扩大合作 计划对后者股权投资
AMD与PsiQuantum扩大合作 计划对后者股权投资(摘要待生成)