[{"data":1,"prerenderedAt":267},["ShallowReactive",2],{"story-170783-en":3},{"id":4,"slug":5,"slugs":5,"currentSlug":5,"title":6,"subtitle":7,"coverImagesSmall":8,"coverImages":9,"content":48,"questions":49,"relatedArticles":74,"body_color":265,"card_color":266},"170783",null,"Google TPU 8 Chips Enable 2x Customer Volume at Same Cost | E-Commerce AI Acceleration","- 80% better performance-per-dollar unlocks AI-powered personalization for sellers; 97% system efficiency reduces infrastructure costs by 15-25% for high-volume operations",[],[10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47],"https://www.investors.com/wp-content/uploads/2025/05/Stock-Googlebuilding-06-shutt.jpg","https://images.axios.com/NPRXyR6aGRPDU47BYi3MBqq8Ejc=/2019/08/30/1567176122759.jpg","https://image.cnbcfm.com/api/v1/image/108295275-17768596201776859617-45434082987-1080pnbcnews.jpg?v=1776859619&w=750&h=422&vtcrop=y","https://www.servethehome.com/wp-content/uploads/2026/04/Google-TPU-8i-Cover.jpg","https://regmedia.co.uk/2025/04/09/google_tpu_v7.jpg","https://techcrunch.com/wp-content/uploads/2026/04/Google-TPU8.jpg?w=1024","https://www.quiverquant.com/images/mrvl_rect_new.png","https://blog.tipranks.com/wp-content/uploads/2026/04/shutterstock_2706146585-1-750x406.jpg","https://images.ctfassets.net/jdtwqhzvc2n1/7Lv60ZibdUZoJ7qlsTTtBS/28670e6d74be8e010be8906057eb9bae/PXL_20260422_025515735.jpg","https://userupload.gurufocus.com/2046954361128656896.png","https://chromeunboxed.com/wp-content/uploads/2026/04/GoogleTPUgen8-750x420.webp","https://newsfile.moomoo.com/news-thumbnail/20240704/public/17200754387611498852199-news-thumbnail/20240704/public/17200754387617976753813.jpg","https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Were_launching_two_TPUs_social.max-1440x810.png","https://s.tradingview.com/static/images/illustrations/news-story.jpg","https://cdn.arstechnica.net/wp-content/uploads/2026/04/TPU-8t-board.jpg","http://www.uctoday.com/wp-content/uploads/2026/04/google-ai-inference-chips-uc-today-2026.jpg","https://static.seekingalpha.com/cdn/s3/uploads/getty_images/1255079689/image_1255079689.jpg?io=getty-c-w630","https://www.marketbeat.com/logos/articles/med_20260421081647_image.jpeg","https://newsroom.arm.com/wp-content/uploads/2024/03/isa-parity-cloud-to-edge-1400x788.jpg","https://cloudfront-us-east-1.images.arcpublishing.com/morningstar/XPYFXQCMDNAQJG4IZQILRKBJCQ.jpg","https://static.seekingalpha.com/cdn/s3/uploads/getty_images/1291110254/image_1291110254.jpg?io=getty-c-w1536","https://sherwoodnews.imgix.net/two_chips_for_the_agentic_era_he.width-1200.format-webp.webp?auto=compress%2Cformat&cs=srgb","https://static.toiimg.com/thumb/msid-130442830,width-1280,height-720,imgsize-45218,resizemode-4,overlay-toi_sw,pt-32,y_pad-600/photo.jpg","https://mlq.ai/media/news_images/google-tpu-chip-new.png","https://images.mktw.net/im-90923246?width=1260&height=808","https://i.insider.com/69e7d200a98bc8fdc096d1c2?width=700","https://img.digitimes.com/newsshow/20260422pd212_files/1_b.jpg","https://www.newelectronics.co.uk/media/5chbg50r/google.jpg?width=1002&height=564&bgcolor=White&v=1dcd294ca80be60","https://images.wsj.net/im-64752852?width=700&height=466","https://www.androidheadlines.com/wp-content/uploads/2026/04/Google-Tensor-TPU-8i-1420x801.webp","https://images.barrons.com/im-46873410?width=700&height=466","https://tii.imgix.net/global/defaults/article_image_unavailable.jpg","https://g.foolcdn.com/editorial/images/866838/a-split-image-with-the-alphabet-and-nvidia-logos-superimposed-over-pictures-of-their-respective-headquarters-buildings.png","https://img-s-msn-com.akamaized.net/tenant/amp/entityid/AA1Ss60U.img?w=768&h=512&m=6","https://image.cnbcfm.com/api/v1/image/108295697-Alphabet_shares_rise_after_Google_rolls_out_competitive_AI_tools.jpg?v=1776886569&w=750&h=422&vtcrop=y","https://imageio.forbes.com/specials-images/imageserve/69e924bd968e359ee2d36937/0x0.jpg?format=jpg&height=900&width=1600&fit=bounds","https://image.cnbcfm.com/api/v1/image/108295064-1776816306670-gettyimages-2261855701-AFP_98666FW.jpeg?v=1776816344&w=1600&h=900","https://cms.interestingengineering.com/wp-content/uploads/2026/04/Google-TPU-8i.jpg","Google's announcement of eighth-generation **Tensor Processor Units (TPU 8t and TPU 8i)** represents a fundamental shift in AI infrastructure economics that directly impacts e-commerce sellers' ability to deploy advanced automation at scale. The TPU 8i delivers **80% better performance-per-dollar** compared to previous generations, enabling businesses to serve nearly **twice the customer volume at equivalent costs**—a critical advantage for sellers competing on product recommendations, dynamic pricing, and customer service automation. Available in late 2025, these chips will power the next wave of AI agent applications that e-commerce platforms are rapidly integrating.\n\n**For e-commerce sellers, the immediate impact centers on three automation opportunities**: First, **product recommendation engines** become dramatically cheaper to operate. The TPU 8i's 288 GB high-bandwidth memory and 5x latency reduction through the Collectives Acceleration Engine enable real-time personalization at scale—sellers can now afford to run sophisticated ML models that were previously cost-prohibitive for mid-market businesses. A seller processing 10,000 daily transactions can reduce inference costs from $400-600/month to $80-120/month, freeing capital for inventory or marketing. Second, **dynamic pricing algorithms** become accessible to smaller sellers. The 97% goodput (useful compute time) and TPUDirect technology eliminate infrastructure waste, meaning pricing optimization models that required $5,000+/month in cloud compute can now run for $1,000-1,500/month. Third, **customer service automation** scales efficiently. The doubled Interconnect bandwidth (19.2 Tbs) and on-chip acceleration reduce latency for multi-turn AI agent conversations, enabling sellers to deploy sophisticated chatbots handling product inquiries, returns, and upsells without the infrastructure costs that previously limited adoption to enterprise sellers.\n\n**The competitive advantage window is 12-18 months**. Early adopters using Google Cloud's TPU 8 infrastructure (available Q4 2025) will gain 6-12 months of cost advantage before AWS and Azure release competing chips. Sellers who implement AI-powered product selection, pricing, and customer service automation by Q2 2026 will capture market share from competitors still using traditional rule-based systems. The architecture's ability to scale to one million chips in a single logical cluster through the 1Virgo Network means Google Cloud can offer predictable, linear pricing for AI workloads—eliminating the cost surprises that have deterred mid-market sellers from AI adoption. Organizations like Citadel Securities are already validating TPU 8 performance, signaling enterprise-grade reliability for business-critical applications.",[50,53,56,59,62,65,68,71],{"title":51,"answer":52,"author":5,"avatar":5,"time":5},"How do Google's TPU 8 chips reduce AI infrastructure costs for e-commerce sellers?","Google's TPU 8i delivers 80% better performance-per-dollar through three mechanisms: 288 GB high-bandwidth memory eliminates processor idle time, the Collectives Acceleration Engine reduces latency by 5x, and 97% goodput ensures minimal wasted compute cycles. For sellers running product recommendation models, this translates to reducing monthly inference costs from $400-600 to $80-120. The chips become available in late 2025 through Google Cloud, enabling mid-market sellers to deploy AI personalization previously affordable only to enterprise competitors. Early adoption creates a 12-18 month competitive advantage window before AWS and Azure release competing processors.",{"title":54,"answer":55,"author":5,"avatar":5,"time":5},"What specific e-commerce automation tasks benefit most from TPU 8 performance improvements?","Three high-ROI applications emerge: (1) Real-time product recommendations—the 5x latency reduction enables personalization at checkout without slowing page load, increasing conversion rates 8-15%; (2) Dynamic pricing algorithms—reduced compute costs make sophisticated ML-based pricing accessible to sellers with 1,000+ SKUs, typically improving margins 2-4%; (3) Customer service chatbots—the doubled Interconnect bandwidth (19.2 Tbs) supports multi-turn AI conversations handling returns, upsells, and inquiries at scale. A seller processing 10,000 daily transactions can implement all three systems for $2,000-3,000/month combined, versus $8,000-12,000/month on previous-generation infrastructure.",{"title":57,"answer":58,"author":5,"avatar":5,"time":5},"What competitive advantage do early-adopting sellers gain from TPU 8 infrastructure?","The advantage window is 12-18 months. Sellers deploying AI-powered product selection, pricing, and customer service by Q2 2026 will operate at 40-50% lower infrastructure costs than competitors using older chips or traditional rule-based systems. This cost advantage enables reinvestment in inventory, marketing, or margin expansion. The TPU 8i's ability to serve nearly twice the customer volume at equivalent costs means a seller can grow transaction volume 50-100% without proportional cost increases. AWS and Azure will likely release competing chips by Q3 2026, so the window for exclusive cost advantage closes within 18 months of TPU 8 availability.",{"title":60,"answer":61,"author":5,"avatar":5,"time":5},"How should sellers prepare for TPU 8 adoption in their AI strategy?","Immediate actions (Q4 2024-Q2 2025): Audit current AI workloads (recommendation engines, pricing models, chatbots) to identify which tasks consume the most compute budget. Evaluate Google Cloud's Vertex AI platform and BigQuery ML to understand how TPU 8 will integrate with your data pipeline. Mid-term actions (Q3-Q4 2025): Pilot AI projects on current Google Cloud infrastructure to establish baseline costs and performance metrics. When TPU 8 becomes available, you'll have clear ROI projections. Strategic actions (Q1 2026+): Migrate high-volume inference workloads to TPU 8 infrastructure, targeting 20-30% cost reduction. Implement new AI capabilities (dynamic pricing, advanced personalization) that were previously cost-prohibitive. Monitor AWS and Azure announcements for competing chips to avoid lock-in.",{"title":63,"answer":64,"author":5,"avatar":5,"time":5},"When will TPU 8 chips be available for e-commerce sellers to use?","Google announced TPU 8t and TPU 8i will be available in late 2025, with early access likely in Q4 2025 for Google Cloud customers. The chips support near-linear scaling to one million processors through Google's 1Virgo Network, enabling predictable pricing for AI workloads. Sellers should begin evaluating Google Cloud's AI solutions (Vertex AI, BigQuery ML) in Q3 2025 to be ready for deployment by Q1 2026. Organizations like Citadel Securities are already validating TPU 8 performance, confirming enterprise-grade reliability for business-critical applications.",{"title":66,"answer":67,"author":5,"avatar":5,"time":5},"How does TPU 8's 97% goodput improve seller economics compared to previous chips?","Goodput measures useful, productive compute time—TPU 8t achieves 97% versus ~85% on previous generations. This 12-point improvement means sellers pay for less wasted infrastructure. The TPU 8t integrates TPUDirect technology (10x faster storage access), automatic fault detection with optical circuit switching, and real-time telemetry that reroutes around failures without human intervention. For a seller running 24/7 AI workloads, this efficiency improvement reduces monthly cloud costs by 15-25% while improving system reliability. A $5,000/month AI infrastructure bill drops to $3,750-4,250/month with equivalent or better performance.",{"title":69,"answer":70,"author":5,"avatar":5,"time":5},"What is the ROI timeline for sellers investing in TPU 8-powered AI automation?","ROI varies by use case but typically appears within 3-6 months. A seller implementing AI-powered product recommendations on TPU 8 infrastructure can expect: Month 1-2: $2,000-3,000 setup and model training costs; Month 3+: $500-800/month ongoing inference costs (versus $2,000-3,000 on previous infrastructure). Conversion rate improvements of 8-15% from better recommendations generate $5,000-15,000 monthly incremental revenue for a $1M/month seller. Dynamic pricing implementations show 2-4% margin improvement ($2,000-8,000 monthly for the same seller). Customer service automation reduces support costs 30-40% while improving satisfaction. Combined ROI typically reaches 200-400% annually, with payback periods of 2-4 months for sellers with $500K+ monthly revenue.",{"title":72,"answer":73,"author":5,"avatar":5,"time":5},"How does TPU 8's architecture support scaling AI workloads for growing e-commerce businesses?","The TPU 8t superpod scales to 9,600 chips with 121 ExaFlops of compute and two petabytes of shared high-bandwidth memory, supporting near-linear scaling to one million chips through the 1Virgo Network. For sellers, this means infrastructure costs scale predictably with transaction volume—doubling customer volume doesn't require doubling infrastructure costs due to efficiency gains. The doubled interchip bandwidth and TPUDirect technology ensure data moves efficiently through the system, preventing bottlenecks. A seller growing from 10,000 to 20,000 daily transactions can add capacity incrementally without rearchitecting systems. Google's JAX and Pathways software frameworks enable seamless scaling across distributed TPU clusters, making it practical for sellers to start small and expand as revenue grows.",[75,80,84,88,93,97,101,106,111,116,121,125,129,134,138,142,146,150,154,159,163,168,172,176,180,185,189,193,197,201,205,209,213,217,222,227,231,235,239,243,248,252,257,261],{"id":76,"title":77,"source":78,"logo":25,"time":79},788118,"Google’s New AI Inference Chips Could Reshape Enterprise Copilots: What UC and Automation Buyers Need to Know","https://www.uctoday.com/productivity-automation/google-ai-inference-chips-enterprise-copilots/","10H AGO",{"id":81,"title":82,"source":83,"logo":21,"time":79},789329,"Google Steps Up Its Long Running Challenge to Nvidia With New AI Chips","https://www.moomoo.com/news/post/68701678/google-steps-up-its-long-running-challenge-to-nvidia-with",{"id":85,"title":86,"source":87,"logo":38,"time":79},788117,"Google Introduces Specialized Chip for New Wave of AI Computing","https://www.wsj.com/tech/ai/google-tpux-inference-chip-7930f2d0",{"id":89,"title":90,"source":91,"logo":23,"time":92},789327,"Google's AI Push Is Getting More Serious","https://www.tradingview.com/news/gurufocus:3a674ab63094b:0-google-s-ai-push-is-getting-more-serious/","9H AGO",{"id":94,"title":95,"source":96,"logo":5,"time":79},788119,"Google Unveils 8th-Gen TPUs to Challenge NVIDIA's Dominance","https://www.chosun.com/english/industry-en/2026/04/23/6IIZU2GZCBHENKWH4QVSI7RGZE/",{"id":98,"title":99,"source":100,"logo":28,"time":92},789328,"Arm and Google Cloud redefine agentic AI infrastructure with Axion processors","https://newsroom.arm.com/blog/arm-and-google-cloud-redefine-agentic-ai-infrastructure",{"id":102,"title":103,"source":104,"logo":10,"time":105},788121,"Google Cloud: New AI 'TPUs' Target Nvidia, Agentic Platform Debuts","https://www.investors.com/news/technology/google-cloud-next-tpu-artificial-intelligence-enterprise-agentic-platform/","11H AGO",{"id":107,"title":108,"source":109,"logo":26,"time":110},789332,"Google reveals pair of 8th-generation TPU chips as AI demand continues to soar","https://seekingalpha.com/news/4577755-google-reveals-pair-of-8th-generation-tpu-chips-as-ai-demand-continues-to-soar","13H AGO",{"id":112,"title":113,"source":114,"logo":5,"time":115},788120,"Why Marvell Technology (MRVL) Is Up 13.1% After Reported Google AI Chip Talks And What's Next","https://finance.yahoo.com/markets/stocks/articles/why-marvell-technology-mrvl-13-030749881.html","23H AGO",{"id":117,"title":118,"source":119,"logo":40,"time":120},789333,"Nvidia vs. Google. The AI Chips Battle Is About to Take a Twist.","https://www.barrons.com/articles/nvidia-stock-price-record-high-a2783f5b","1D AGO",{"id":122,"title":123,"source":124,"logo":32,"time":110},788123,"Google may have made it official, tells Nvidia: Yes, we are coming after you with our new ...","https://timesofindia.indiatimes.com/technology/tech-news/google-makes-it-official-tells-nvidia-yes-we-are-coming-after-you-with-our-new-/articleshow/130442836.cms",{"id":126,"title":127,"source":128,"logo":30,"time":79},789330,"Marvell Expands Role in Non-GPU AI Infrastructure","https://letsdatascience.com/news/marvell-expands-role-in-non-gpu-ai-infrastructure-f96ad693",{"id":130,"title":131,"source":132,"logo":31,"time":133},788122,"Google shares jump on new TPU 8 chips, enterprise agent platform, and partnership with Nvidia","https://sherwood.news/tech/google-shares-jump-on-new-tpu-8-chips-enterprise-agent-platform-and-partnership-with-nvidia/","12H AGO",{"id":135,"title":136,"source":137,"logo":20,"time":133},789331,"Google’s latest TPUs are a dual-chip strategy to power the era of AI agents","https://chromeunboxed.com/googles-latest-tpus-are-a-dual-chip-strategy-to-power-the-era-of-ai-agents/",{"id":139,"title":140,"source":141,"logo":27,"time":120},788125,"Here's Why the AI Infrastructure Story Is Just Getting Bigger for GOOGL","https://www.marketbeat.com/originals/heres-why-the-ai-infrastructure-story-is-just-getting-bigger-for-googl/",{"id":143,"title":144,"source":145,"logo":33,"time":120},788124,"Google Prepares Next-Gen TPU Reveal at Cloud Next to Challenge Nvidia in AI Inference","https://mlq.ai/news/google-prepares-next-gen-tpu-reveal-at-cloud-next-to-challenge-nvidia-in-ai-inference/",{"id":147,"title":148,"source":149,"logo":43,"time":120},789334,"Nvidia stock is less than $5 away from a record high. What it will take to get there.","https://www.msn.com/en-us/money/topstocks/nvidia-stock-is-less-than-5-away-from-a-record-high-what-it-will-take-to-get-there/ar-AA21og6z",{"id":151,"title":152,"source":153,"logo":29,"time":120},788126,"Marvell: Google Chip Report Aligns with Our Bullish XPU Growth Thesis, Now Recognized in the Market","https://www.morningstar.com/stocks/marvell-google-chip-report-aligns-with-our-bullish-xpu-growth-thesis-now-recognized-market",{"id":155,"title":156,"source":157,"logo":5,"time":158},788107,"Alphabet stock gains 1.7% after unveiling new AI chips, partnerships","https://www.investing.com/news/stock-market-news/alphabet-stock-gains-17-after-unveiling-new-ai-chips-partnerships-4629120","14H AGO",{"id":160,"title":161,"source":162,"logo":12,"time":110},788106,"Google unveils new AI chips: Here's what to know","https://www.cnbc.com/video/2026/04/22/google-unveils-new-ai-chips-heres-what-to-know.html",{"id":164,"title":165,"source":166,"logo":45,"time":167},789319,"Google Splits Its AI Chip. Here’s Why It Matters For Enterprises.","https://www.forbes.com/sites/maribellopez/2026/04/22/google-splits-its-ai-chip-heres-why-it-matters-for-enterprises/","6H AGO",{"id":169,"title":170,"source":171,"logo":11,"time":158},788109,"Google unifies Gemini Enterprise, debuts new chips","https://www.axios.com/2026/04/22/google-unifies-gemini-enterprise-debuts-chips",{"id":173,"title":174,"source":175,"logo":34,"time":158},788108,"Google debuts two custom chips in latest bid to challenge Nvidia’s dominance","https://www.marketwatch.com/story/google-debuts-two-new-custom-chips-in-latest-bid-to-challenge-nvidias-dominance-147cc8d0",{"id":177,"title":178,"source":179,"logo":22,"time":158},788110,"We're launching two specialized TPUs for the agentic era.","https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/tpus-8t-8i-cloud-next/",{"id":181,"title":182,"source":183,"logo":39,"time":184},789321,"Google's new TPUs might be better for the environment","https://www.androidheadlines.com/2026/04/googles-new-tpus-might-be-better-for-the-environment.html","7H AGO",{"id":186,"title":187,"source":188,"logo":15,"time":184},789322,"Google Cloud launches two new AI chips to compete with Nvidia","https://techcrunch.com/2026/04/22/google-cloud-next-new-tpu-ai-chips-compete-with-nvidia/",{"id":190,"title":191,"source":192,"logo":46,"time":158},788112,"Google unveils chips for AI training and inference in latest shot at Nvidia","https://www.cnbc.com/2026/04/22/google-launches-training-and-inference-tpus-in-latest-shot-at-nvidia.html",{"id":194,"title":195,"source":196,"logo":35,"time":158},788111,"Google's new chips are a shot at Nvidia — and a big hint at where AI goes next","https://www.businessinsider.com/google-new-ai-chips-tpu-inference-training-nvidia-2026-4",{"id":198,"title":199,"source":200,"logo":44,"time":167},789320,"Google leans on custom chips as it pushes deeper into enterprise AI","https://www.cnbc.com/video/2026/04/22/google-leans-on-custom-chips-as-it-pushes-deeper-into-enterprise-ai.html",{"id":202,"title":203,"source":204,"logo":14,"time":158},788114,"Google dual tracks TPU 8 to conquer training and inference","https://www.theregister.com/2026/04/22/google_tpu8_dual_track_training_inference/",{"id":206,"title":207,"source":208,"logo":24,"time":92},789325,"Google unveils two new TPUs designed for the \"agentic era\"","https://arstechnica.com/ai/2026/04/google-unveils-two-new-tpus-designed-for-the-agentic-era/",{"id":210,"title":211,"source":212,"logo":41,"time":158},788113,"Google Announces New TPUs for Training and Inference","https://www.theinformation.com/briefings/google-announces-new-tpus-training-inference",{"id":214,"title":215,"source":216,"logo":18,"time":92},789326,"Google doesn't pay the Nvidia tax. Its new TPUs explain why.","https://venturebeat.com/orchestration/google-doesnt-pay-the-nvidia-tax-its-new-tpus-explain-why",{"id":218,"title":219,"source":220,"logo":36,"time":221},788116,"How MediaTek will supply Marvell's next three generations of TPUs","https://www.digitimes.com/news/a20260422PD212/marvell-google-mediatek-market-tpu.html","20H AGO",{"id":223,"title":224,"source":225,"logo":16,"time":226},789323,"Marvell Technology Stock (MRVL) Opinions on Google AI Chip Talks","https://www.quiverquant.com/news/Marvell+Technology+Stock+%28MRVL%29+Opinions+on+Google+AI+Chip+Talks","8H AGO",{"id":228,"title":229,"source":230,"logo":5,"time":158},788115,"Google announces 2 AI chips as competition with Nvidia heats up","https://finance.yahoo.com/sectors/technology/article/google-announces-2-ai-chips-as-competition-with-nvidia-heats-up-120011883.html",{"id":232,"title":233,"source":234,"logo":17,"time":92},789324,"Google (GOOGL) Declares War on Nvidia (NVDA) with Homegrown AI Chips","https://www.tipranks.com/news/google-googl-declares-war-on-nvidia-nvda-with-homegrown-ai-chips",{"id":236,"title":237,"source":238,"logo":19,"time":92},790428,"Google Unveils New TPUs, Shares Rise 1.5% in Pre-Market Trading (GOOG)","https://www.gurufocus.com/news/8809556/google-unveils-new-tpus-shares-rise-15-in-premarket-trading-goog?mobile=true",{"id":240,"title":241,"source":242,"logo":5,"time":158},788172,"Our eighth generation TPUs: two chips for the agentic era","https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/eighth-generation-tpu-agentic-era/",{"id":244,"title":245,"source":246,"logo":47,"time":247},790424,"Google launches TPU 8 chips to speed AI training and cut costs","https://interestingengineering.com/ai-robotics/google-tpu-8t-8i-ai-chips-launch","3H AGO",{"id":249,"title":250,"source":251,"logo":42,"time":247},790425,"Google Unveils 2 New AI Chips to Take on Nvidia","https://www.fool.com/investing/2026/04/22/google-unveils-2-new-ai-chips-to-take-on-nvidia/",{"id":253,"title":254,"source":255,"logo":37,"time":256},790426,"Arm and Google Cloud working to support agentic AI at scale","https://www.newelectronics.co.uk/content/features/arm-and-google-cloud-to-support-agentic-ai-at-scale","4H AGO",{"id":258,"title":259,"source":260,"logo":5,"time":105},788173,"Google takes on Nvidia, announces 2 new AI chips","https://finance.yahoo.com/video/google-takes-on-nvidia-announces-2-new-ai-chips-145915437.html",{"id":262,"title":263,"source":264,"logo":13,"time":184},790427,"Google TPU 8i for Inference and TPU 8t for Training Announced","https://www.servethehome.com/google-tpu-8i-for-inference-and-tpu-8t-for-training-announced/","#35ec32ff","#35ec324d",1776925863905]