device that provides hardware acceleration for artificial intelligence applications
Intel's Crescent Island AI accelerator officially powers new GPU D-Series products — chip delivers 1.3 peak PFLOPS of FP8 compute and 1 ...
tomshardware.com/tech-industry/artificial-intelligence/intels-crescent-island-ai-accelerator-officially-powers-new-gpu-d-series-products-chip-delivers-1-3-peak-pflops-of-fp8-compute-and-1-5-tb-s-of-memory-bandwidth
Paid articlePaywall SK hynix details H-TPU, tiny cores that offload the Arm CPUs in its SSD controllers
tomshardware.com/pc-components/ssds/sk-hynix-to-place-custom-tiny-cpu-cores-in-its-ssd-controllers-to-ease-arm-load-as-drive-speeds-rocket-each-h-tpu-reportedly-draws-2-5-percent-of-a-cpu-cores-power
Paid articlePaywall
VideoCardz Oct 12
Intel GPU D270 is Xe3p “Crescent Island” card with 480GB LPDDR5X memory: Intel GPU D270 is a 350W AI accelerator with 1.5 TB/s of memory bandwidth ...
videocardz.com/newz/intel-gpu-d270-is-xe3p-crescent-island-card-with-480gb-lpddr5x-memory
Chiding in plain sight: Inside the Taxpayers’ Union’s in-your-face push for change. Part 1 of a 2 part series.
newsroom.co.nz/2023/10/31/chiding-in-plain-sight/
Google thinks SpaceX’s Starship has to launch 1,800 times before space data centers get off the ground
techcrunch.com/2026/10/01/google-thinks-spacexs-starship-has-to-launch-1600-times-before-space-data-centers-get-off-the-ground/
Decatur will become home to Alabama’s first statewide artificial intelligence accelerator, a first-of-its-kind initiative designed to both accelerate promising ... Show more
Decatur will be home to Alabama’s first AI accelerator
Decatur will be home to Alabama’s first AI accelerator
github.com Sep 30
GitHub - lokutor-ai/oido: Oído: open-vocabulary speech recognition that fits in a $5 ESP32-S3. 3.7% LibriSpeech WER, no cloud, no NPU.
github.com/lokutor-ai/oido
Windows 11 Version 26H2: New feature update available now: Is it worth switching to the new Windows 11 update v26H2, available now, for your computer? The update ...
winfuture.de/news,161626.html
'No country controls every layer': The AI complex systems reshaping global power
abc.net.au/news/2026-09-24/ai-race-chips-data-energy-cake-stack-us-china-asia-pacific/107071946
Nicky Hager’s new book, Dirty Work: Taxpayers’ Union’s Jordan Williams responds: Taxpayer Union's Jordan Williams says it's 'tempting' to want to sue him.
nzherald.co.nz/nz/politics/journalist-nicky-hager-releases-latest-book-dirty-work-on-taxpayers-union/FJTG2JKT2JBA3ADWGQ43AVIWC4/
Google announces Tensor G6, bringing 4K Portrait Video and more to the Pixel 11 series
androidauthority.com/google-tensor-g6-3695351/
trendforce Sep 02
[News] Google Says Memory Tops 75% of Server BOM; TPU-Software Dual Strategy Targets Memory Wall
trendforce.com/news/2026/09/02/news-google-says-memory-tops-75-of-server-bom-tpu-software-dual-strategy-targets-memory-wall/
![[News] Google Says Memory Tops 75% of Server BOM; TPU-Software Dual ...](https://img.trendforce.com/blog/wp-content/uploads/2025/09/04111503/Google-Ironwood-TPU-624x352.jpg)
AMD Threadripper Halo Station: 576 GB for local AI: During the opening keynote of IFA 2026 in Berlin, AMD showcased a liquid-cooled workstation for local AI ...
pcgameshardware.de/CPU-CPU-154106/News/AMD-Threadripper-Halo-Station-96-Kerne-MI350P-1552945/
3D DRAM: 100 TB/s beats HBM4 in AI accelerator: At Hot Chips 2026, D-Matrix showcased Raptor, an AI accelerator that stacks the computational logic directly onto ...
pcgameshardware.de/Kuenstliche-Intelligenz-Hardware-279517/News/3D-DRAM-Raptor-100-TB-s-HBM4-d-Matrix-1552420/
Nvidia NVHBM: Custom HBM brings 30 percent more bandwidth: Nvidia hat Ende August mit NVHBM eine eigene HBM-Variante vorgestellt, bei der der Speichercontroller ...
pcgameshardware.de/RAM-Hardware-154108/News/Nvidia-NVHBM-Custom-HBM-mehr-Bandbreite-Chipflaeche-1552421/
Sneaker News Aug 27
The AIR JORDAN 41 packs a punch.
- New Nanoprint upper
- Full length Zoom strobel
- ZoomX foam
- Extra Zoom bag at the heel
- Carbon fiber midfoot with a TPU ... Show more
IBM's first dual-ISA core natively executes ARM and z/Architecture in the same core; all cores run at 5.7 GHz base frequency — next-gen mainframe AI ...
tomshardware.com/pc-components/cpus/ibms-first-dual-isa-core-natively-executes-arm-and-z-architecture-in-the-same-core-all-cores-run-at-5-7-ghz-base-frequency-next-gen-mainframe-ai-processor-is-built-on-2nm-node-with-11-cores
Microsoft brings improved Windows 11 hardware monitoring with upgraded Task Manager
neowin.net/news/microsoft-brings-improved-windows-11-hardware-monitoring-with-upgraded-task-manager/
OpenAI says its Jalapeño chip offers spicy performance: The chip is designed for serving up AI models, rather than training new ones.
axios.com/2026/08/25/openai-says-its-jalapeno-chip-offers-spicy-performance
Marvell $MRVL is expanding its Google partnership to build custom AI chips for the TPU ecosystem!
KEY HIGHLIGHTS:
Issued Google a warrant for up to 58 ... Show more
Notebookcheck Aug 26
Garmin Fenix 9 (Pro) expected to use new ARM chip with 50% more RAM and AI accelerator
notebookcheck.net/Garmin-Fenix-9-Pro-expected-to-use-new-ARM-chip-with-50-more-RAM-and-AI-accelerator.1379157.0.html
Google reportedly taps AMD to design next-generation TPU — hybrid AI ASIC could integrate on-package CPU cores for reinforcement learning
tomshardware.com/tech-industry/artificial-intelligence/google-reportedly-taps-amd-to-design-next-generation-tpu-hybrid-ai-asic-could-integrate-on-package-cpu-cores-for-reinforcement-learning
Windows 11: Unified Memory for graphics and AI configurable: Microsoft is testing a setting in the current Insider build of Windows 11 that permanently allocates ...
pcgameshardware.de/Windows-Software-277633/News/Unified-Memory-Grafik-KI-Speicher-reservieren-1551856/
Key FactsWindows 11 tests fixed Unified Memory reservations
Aug 23, 2026 · Reading time: ~30 sec.
Summary created with AI assistance · Source: pcgameshardware.de
Summary created with AI assistance · Source: pcgameshardware.de
- The previously unannounced Intelligent Carveout feature was discovered in the Experimental channel of build 29648.1000 from August 17, with Windows Latest reporting the find and phantomofearth confirming it.
- The switch, apparently located under Settings > System > Advanced, currently appears to offer five fixed options: Custom, Recommended, High, Maximum, and Don't allow.
- Unlike dynamically used Shared GPU Memory, Intelligent Carveout is intended to reserve a fixed portion of shared memory for graphics and AI, leaving it unavailable to other programs afterward.
- The report identifies notebooks and mini-PCs with Nvidia RTX Spark or AMD Ryzen AI Max as possible initial systems, while the feature could still be changed or abandoned before release.
Microsoft brings improved Windows 11 hardware monitoring with upgraded Task Manager
neowin.net/news/microsoft-brings-improved-windows-11-hardware-monitoring-with-upgraded-task-manager/
Intel's cheapest CPUs do what Google's discontinued Coral accelerators used to do, and they cost less
xda-developers.com/intels-cheapest-cpus-do-what-googles-coral-accelerators/
trendforce Aug 11
[News] Microsoft Eyes Maia 300 Launch in Sep., Reportedly Seeks 300K TSMC Capacity as Google, AWS Race Ahead
trendforce.com/news/2026/08/11/news-microsoft-eyes-maia-300-launch-in-sep-reportedly-seeks-300k-tsmc-capacity-as-google-aws-race-ahead/
![[News] Microsoft Eyes Maia 300 Launch in Sep., Reportedly Seeks 300K TSMC ...](https://img.trendforce.com/blog/wp-content/uploads/2026/08/11101504/Microsoft-Maia-624x350.png)
trendforce Aug 03
[News] MediaTek Advances 400G SerDes on 2nm, Targeting Google TPU v10 and Meta AI ASIC Orders
trendforce.com/news/2026/08/03/news-mediatek-advances-400g-serdes-on-2nm-targeting-google-tpu-v10-and-meta-ai-asic-orders/
![[News] MediaTek Advances 400G SerDes on 2nm, Targeting Google TPU v10 and ...](https://img.trendforce.com/blog/wp-content/uploads/2025/05/20135115/MediaTek_MasterBrand_EN_BlackBG_Combo_0923-624x624.png)
Google Pixel 11 prices to go up? Here’s why upcoming Pixel phones may cost you more
financialexpress.com/life/technology-google-pixel-11-prices-to-go-up-heres-why-upcoming-pixel-phones-may-cost-you-more-4303664/
Intel's cheapest CPUs do what Google's discontinued Coral accelerators used to do, and they cost less
xda-developers.com/intels-cheapest-cpus-do-what-googles-coral-accelerators/
Ilya Sutskever's SSI chooses Nvidia Vera Rubin systems after earlier Google TPU partnership
neowin.net/news/ilya-sutskevers-ssi-chooses-nvidia-vera-rubin-systems-after-earlier-google-tpu-partnership/
AMD confirms Zen 7 Epyc "Florence" for 2028 and previews Zen 8 Epyc "Ravenna" for 2030
techspot.com/news/113233-amd-confirms-zen-7-epyc-florence-2028-previews.html
9to5google.com Jul 14
Google announces Gemma 4 optimized for the Pixel 10’s TPU: Google today announced Gemma 4 E2B for TPU that’s designed to run natively on the Pixel 10’s TPU ...
9to5google.com/2026/07/14/pixel-10-gemma-4/
dlvr.it Jul 07
Broadcom and Apple extend custom silicon pact to 2031: Probably an extension of long-running iThing innards arrangement, not a new custom AI accelerator deal
theregister.com/personal-tech/2026/07/07/broadcom-and-apple-extend-custom-silicon-pact-to-2031/5267491
Samsung is building a dedicated AI chip for PCs, and HP and Lenovo are already testing it
techspot.com/news/113074-samsung-building-dedicated-ai-chip-pcs-hp-lenovo.html
github.com Jul 16
GitHub - jegly/Box: The most advanced, fully offline client-side AI suite on Android today.
github.com/jegly/Box
Nvidia Corp. ($NVDA) CEO Jensen Huang dismissed reports of Vera Rubin AI hardware delays.
He said the company’s high-end AI accelerator systems remain on ... Show more
The Tensor G6 isn't a downgrade at all; it's Google admitting what really matters
androidpolice.com/tensor-g6-downgrade-is-smart/
Intel's Starfire SoC combines 18A CPU, Intel 3 GPU, and NPU for space-grade computing
techspot.com/news/113110-intel-starfire-soc-combines-18a-cpu-intel-3.html
Sachsens Justiz will einen Ferrari versteigern – die Spuren führen nach Hessen
scotsman.com/business/networks-are-the-glue-that-holds-scotlands-businesses-together-8786999
What tech companies actually mean when they say they're challenging Nvidia: Company after company has lined up to challenge Nvidia's grip on AI chips ...
qz.com/challenging-nvidia-chip-competition-meaning-inference-training-070626
Laptop specs are getting more confusing – here’s what actually matters in 2026: Look past the marketing hype to get the best Windows laptop
techradar.com/computing/windows-laptops/laptop-specs-are-getting-more-confusing-heres-what-actually-matters-in-2026
Cathie Wood’s ARK Invest says Advanced Micro Devices ($AMD) is becoming a more serious challenger to Nvidia ($NVDA) in the AI accelerator market ... Show more
I switched my local AI setup to AMD's Lemonade after Nvidia support landed, and solved my local AI portability problem
xda-developers.com/local-ai-amd-lemonade-nvidia-support-solved-local-portability-problem/
Nvidia now sells AI accelerators in exchange for revenue sharing: A new business model is intended to bring even more money into Nvidia's coffers ...
heise.de/news/Nvidia-verkauft-KI-Beschleuniger-jetzt-gegen-Umsatzbeteiligung-11355200.html
How to 3D-Print a Book: Using flexible TPU and an ingenious vertical printing technique, a new research project bypasses traditional publishing entirely ...
inc.com/fast-company-2/how-to-3d-print-a-book/91365229
Marie Claire UK Jun 27
Jelly Heels Are Summer's Most Unexpectedly Chic Shoe—and Grown Women Can't Stop Wearing Them
marieclaire.co.uk/fashion/jelly-heels
Taiwan's MediaTek secures key order for Google Tensor Processing Unit | Taiwan News | Jun. 23, 2026 09:55
taiwannews.com.tw/en/news/6387636
‘Ignore The Noise:’ Top Bank Says AVGO Stock Dip Is A Buying Opportunity, Points To Locked-In Broadcom–Go
benzinga.com/analyst-stock-ratings/analyst-color/26/06/60010282/ignore-the-noise-top-bank-says-avgo-stock-dip-is-a-buying-opportunity-points-to-locked-in-broadcom-google-ai-chip-deal-through-2031
AMD Zen 6 desktop CPUs may ditch integrated graphics for a built-in NPU: AMD is expected to launch its Zen 6 "Morpheus" processors later this year or in early ...
techspot.com/news/112788-amd-zen-6-desktop-cpus-may-ditch-integrated.html
WindowsCentral Jun 16
Microsoft just unveiled the new Surface Laptop 8 and Surface Pro 12, both powered by Snapdragon X2 chips and built to push Copilot+ PCs forward again ... Show more
AMD Zen 6: Ryzen X likely with AI engine instead of graphics unit: AMD's upcoming Ryzen X/10000 series desktop processors ("Olympic Ridge") based on Zen 6 ...
pcgameshardware.de/CPU-CPU-154106/News/Zen-6-Olympic-Ridge-NPU-statt-Grafikeinheit-1545434/
WSL 3 will finally let Linux apps use your GPU and NPU without the performance tax
xda-developers.com/wsl-3-will-finally-let-linux-apps-use-your-gpu-without-the-performance-tax/
Acer’s Swift Spin 14 AI was just named PCMag’s Best Laptop at Computex. 🏆
This 2-in-1 packs a Snapdragon X2 Elite, 16GB RAM, a 120Hz touch screen ... Show more
Acer’s new Swift Spin 14 AI was just named PCMag’s Best Laptop at Computex. 🏆
This 2-in-1 packs a Snapdragon X2 Elite, 16GB RAM, a 120Hz touch screen ... Show more
trendforce May 20
🤖 Can Google TPU Replace the GPU?
TPU v7 delivers over 8× the BF16 compute of TPU v4, closing the performance-per-dollar gap to NVIDIA ... Show more
dlvr.it May 27
Windows 11 update preview promises faster launches, puts Task Manager on NPU patrol
theregister.com/personal-tech/2026/05/27/windows-11-update-preview-promises-faster-launches-puts-task-manager-on-npu-patrol/5246940
dlvr.it May 28
Google, Canonical team up to certify Ubuntu images for TPU VMs: Chocolate Factory shifts Tensor Processing Unit Ubuntu support back upstream
theregister.com/off-prem/2026/05/28/google-canonical-team-up-to-certify-ubuntu-images-for-tpu-vms/5248028
Windows 11 is a complete traffic accident: Windows 11 hat den Ruf, ziemlich fehlerhaft zu sein. Wir wollten wissen: Was genau ist aktuell kaputt?
heise.de/news/Windows-11-ist-ein-kompletter-Verkehrsunfall-11304565.html
Blackstone to invest $5 billion in AI infrastructure venture with Google, powered by TPU chips
cnbc.com/2026/05/19/blackstone-google-ai-data-center-joint-venture-tpu.html
Key FactsGoogle Transforms Search with AI Agents and Wearables
May 20, 2026 · Reading time: ~45 sec.
Summary created with AI assistance · Source: axios.com
Summary created with AI assistance · Source: axios.com
- Google is expanding the search box to accommodate extended and conversational queries.
- AI agents enable users to automate recurring requests, such as alerts for nearby concert announcements.
- AI integration extends to new hardware, including audio-only smart glasses releasing later this year.
- The company aims to embed AI across all products to stay ahead of standalone chatbot competitors.
Agents in search is the next step.— Google DeepMind chief Demis Hassabis said in an interview.
github.com May 19
Releases · google-ai-edge/gallery: A gallery that showcases on-device ML/GenAI use cases and allows people to try and use models locally ...
github.com/google-ai-edge/gallery/releases
AMD users have reason to celebrate: FSR 4.1 is officially coming to RX 7000 and RX 6000 graphics cards as a free driver update. Unlike FSR 3, FSR Upscaling 4 ... Show more
The Pixel 11’s Tensor G6 sounds promising — until you compare it to rival flagship chips
androidauthority.com/pixel-11-performance-rumors-3664645/
Tech Advisor May 05
After a bad start, Google could still win 2026 with the Pixel 11: After the underwhelming Pixel 10a, leaked Pixel 11 specifications suggest that Google could be ...
techadvisor.com/article/3131621/google-pixel-11-spec-leak-points-to-progress.html
Tiny company steals AMD's thunder and challenges Nvidia with old-tech PCIe AI accelerator that runs 700B LLMs locally ...
techradar.com/pro/tiny-company-steals-amds-thunder-and-challenges-nvidia-with-old-tech-pcie-ai-accelerator-that-runs-700b-llms-locally-sipping-just-240w-thanks-to-decade-old-ddr4-and-28nm-chips
Wccftech May 07
This PCIe AI Accelerator Card Can Run 700B LLMs Locally With 384 GB Memory at Just 240W, Less Than Half The Power of RTX PRO 6000 Blackwell
wccftech.com/this-pcie-ai-accelerator-card-packs-384-gb-memory-run-700b-llms-240w/
dlvr.it Apr 30
AWS says acute server memory shortage is driving customers to the cloud: : When you can't get 'em with a 'transformation plan,' supply chain pain ...
theregister.com/2026/04/30/server_memory_shortage_pushing_you/
New Tensor G6 leak has good and bad news for Pixel 11 series: A new leak has revealed a ton of Tensor G6 specs, with this processor expected to power the Pixel ...
androidauthority.com/google-tensor-g6-full-specs-pixel-11-leak-3663212/
dlvr.it Apr 30
Google to sell its TPUs to some customers, who also fancy big-G GPUs: : AI is driving more searches and ads
theregister.com/2026/04/30/alphabet_google_q1_fy26/
Key FactsReview: Bambu Lab X2D Uses a Dual-Nozzle Design
Apr 14, 2026 · Reading time: ~30 sec.
Summary created with AI assistance · Source: tomshardware.com
Summary created with AI assistance · Source: tomshardware.com
- The listed release date is April 14, 2026, and the machine costs $649 standalone or $899 as a Combo.
- The hybrid toolhead combines a direct-drive main nozzle with an external Bowden auxiliary nozzle, helping with two-color printing and support removal.
- The auxiliary nozzle limits speed and, according to the test, is unsuitable for normal soft TPU; a calibration cube took 41 minutes 59 seconds compared with 20 minutes 35 seconds on the main nozzle.
- Features include a 256 × 256 × 256-millimeter build volume, active temperature control, three-stage filtration, automatic leveling, an AI monitoring camera, and an AMS 2 Pro with the Combo model.
The New Stack Apr 25
Google splits its TPU line in two for the agentic era: At Google Cloud Next on Wednesday, the company introduced TPU 8t and TPU 8i. TPU 8t is built for ...
thenewstack.io/google-splits-tpu-line/
Gizchina Apr 24
Google TPU Chips Reach Its 8th Generation With Separate Chips For Training And Inference
gizchina.com/google-2/google-tpu-chips-reach-its-8th-generation-with-separate-chips-for-training-and-inference
VideoCardz Apr 25
Windows 11 Task Manager will show GPU neural engines on Performance page: Microsoft Task Manager update will show Neural Engines in Windows 11 with more detailed ...
videocardz.com/newz/windows-11-task-manager-will-show-gpu-neural-engines-on-performance-page
Google challenges Nvidia with new dedicated training and inference AI chips: The TPU 8t targets model training while the TPU 8i handles inference workloads ...
qz.com/google-tpu-8t-8i-training-inference-ai-chips-042226
Google debuts eighth-gen TPU chips for training and inference with big performance claims - Neowin
neowin.net/news/google-debuts-eighth-gen-tpu-chips-for-training-and-inference-with-big-performance-claims/
on.wsj.com Apr 22
Google Introduces Specialized Chip for New Wave of AI Computing: New version of company’s TPU is first designed for inference as agent boom sends demand soaring
wsj.com/tech/ai/google-tpux-inference-chip-7930f2d0
Paid articlePaywallSomeone finally got an RTX 5090 running on a Mac – no hacks required: Tiny Corp, the same company that built the tinybox AI accelerator ...
techspot.com/news/112082-someone-finally-got-rtx-5090-running-mac-no.html
Taiwan’s TSMC posts NT$415.19 billion revenue for March | Taiwan News | Apr. 10, 2026 18:01
taiwannews.com.tw/en/news/6338234
Broadcom to supply Anthropic with 3.5 gigawatts of Google TPU capacity from 2027 — Claude pioneer says its annual revenue run rate has passed $30 billion
tomshardware.com/tech-industry/broadcom-expands-anthropic-deal-to-3-5gw-of-google-tpu-capacity-from-2027
