AI hardware and compute news
Stories 361 to 420 of 421, from 13 September 2026 to 23 September 2026, newest first. The latest are on the first page →
-
AMD adds GDDR7 support to its Linux GPU driver
AMD has begun adding support for GDDR7 memory and several new IP blocks to its open‑source Linux kernel driver. The patches, reported by Phoronix, include identification of GDDR7, IH 8.0 interrupt handler, NBIF 7.10…
1 sourceTom's Hardware -
China launches Supercomputing-1 AI satellite for orbital data processing
China lifted nine satellites into orbit on September 20 using CAS Space’s Kinetica 1 Y18 rocket, including the S‑AIDC‑1 "Supercomputing-1" AI satellite developed by S‑AIDC. According to Digitimes, the satellite will…
1 sourceTom's Hardware -
Viture announces Vonder glasses that map memories, preorder begins Sep 28
Viture, a San Francisco‑based wearable maker, unveiled its Vonder glasses, a pair of frames that look like ordinary eyeglasses but contain microphones, speakers and bone‑conduction sensors instead of cameras. The…
2 sourcesanalyticsinsight.netWired AI -
Global AI glasses shipments rise 263% YoY in H1 2026, led by Meta and EssilorLuxottica
Counterpoint Research reported that worldwide shipments of AI‑enabled glasses, covering both display‑less and AR/AI devices, jumped 263% year‑over‑year in the first half of 2026. The analysts behind the study are…
1 sourcecounterpointresearch.com -
Broadcom says Anthropic likely to become its biggest XPU customer in 2027
Broadcom indicated that Anthropic is likely to become its largest XPU customer starting in 2027 and continuing through 2028, according to remarks made on the company's third‑quarter earnings call. The firm plans to…
1 sourcefinance.yahoo.com -
SpeakON releases $129 MagSafe AI voice button for iPhone
SpeakON has launched a 25‑gram magnetic button that snaps to the back of an iPhone and includes its own microphone, battery and 128 MB of storage. The device captures speech up to about 60 cm away, buffers on‑device,…
2 sourcesMarkTechPost -
Google launches Googlebook laptops starting at $899 with Android integration and AI features
Google has launched pre-orders for Googlebook, a new line of laptops built by Acer, ASUS, Dell, HP, and Lenovo, starting at $899. The devices run on a hybrid OS combining Android and ChromeOS, feature Qualcomm…
6 sourcesnote.comindiatoday.inTechCrunch AIUnite.AI +1 more -
China builds hundreds of data centres in Inner Mongolia to power AI expansion
In the grasslands of Inner Mongolia’s Ulanqab region, rows of rectangular data‑centre blocks are rising alongside potato fields. Hundreds of facilities are being built for Chinese tech giants such as Huawei, ByteDance…
1 sourceBBC Technology -
CoolIT says near‑total liquid heat capture enables fanless AI servers beyond 250 kW
Hybrid liquid‑air cooling works up to about 250 kW per rack, but a 70/30 split leaves roughly 75 kW of air load. CoolIT argues that moving that heat to liquid makes the air component drop below one percent, allowing…
1 sourceIEEE Spectrum AI -
Kurnal Insights says DRAM per‑area price exceeds TSMC N2/N3 wafer price
Kurnal Insights observes that a 300‑mm wafer processed with TSMC's N3 technology is priced at $20,000, which works out to about $0.283 per mm², while an N2 wafer is assumed to cost $30,000 or $0.424 per mm². These…
1 sourceTom's Hardware -
OpenAI and Anthropic reportedly seek 20‑30 MW data centers as gigawatt projects lag
OpenAI and Anthropic are reportedly scrambling to secure existing data centers with about 20‑30 MW of capacity, according to a September 18 CNBC report. The move aims to provide immediate inference power while their…
1 sourceTom's Hardware -
Opinion: M5 Ultra Mac Studio enables fast local AI agents, reviewer says
The reviewer tested Apple’s M5 Ultra Mac Studio with 256 GB of unified RAM, comparing it to the previous M3 Ultra (512 GB) and a desktop PC built around an RTX 5090 GPU. Using local models such as Qwen3.8‑Flash‑Next,…
2 sources HN 41timdettmers.commacstories.net -
Jeff Dean says automation could cut chip design teams to ten and cycles to three months
He said a team of about ten engineers could create a new custom chip in three months, compared with the traditional approach of roughly 150 engineers over two years. Dean cited his earlier work, including a Nature…
1 sourcewccftech.com -
Mixture of experts reshapes inference compute and data movement pipelines
Mixture‑of‑Experts (MoE) models have altered the architecture of inference services beyond simply adding parameters. Inference now runs in a cluster managed by orchestration layers such as NVIDIA Dynamo or Mooncake,…
1 sourceSemiAnalysis -
Nvidia launches DSX Ready program to qualify power and cooling products
Nvidia announced DSX Ready, a qualification program that helps AI‑factory builders select power and cooling solutions that meet the company’s DSX reference design requirements. The initial rollout covers two…
1 source primary sourceNVIDIA Blog -
Nvidia introduces Halos safety system for autonomous vehicles and robots
Physical AI is moving from research to large‑scale deployment, with ABI Research forecasting 49 million level‑3‑5 autonomous vehicles by 2035 and Omdia estimating 60 million industrial robots will be deployed between…
1 source primary sourceNVIDIA Blog -
Apple and Tesla were better insulated from Hugging Face breach, analysts say
Apple reported fiscal Q3 2026 revenue of $109.42B, up 16.4% year‑over‑year, with earnings per share of $2.02 beating the $1.89 consensus. iPhone sales contributed $54.25B and Services $30.74B. Tim Cook introduced a new…
1 sourcefinance.yahoo.com -
Huawei limits new Ascend AI chips to China due to capacity constraints
Huawei announced that its upcoming Ascend 960‑series neural processing units will be sold only within China because the company cannot meet domestic demand with its current production capacity. Rotating chairman Eric…
1 sourceTom's Hardware -
Googlebooks are phone-first, not ai-first
Google unveiled a line of laptops called Googlebooks that run an Android‑derived OS called Aluminum. The devices are designed to sync tightly with Android phones, offering instant access to phone apps, file systems,…
1 sourcepcworld.com -
Acer CEO says memory makers hype shortage fears, predicts PC prices to fall after 2027
Acer chairperson and CEO Jason Chen told reporters that only high‑end DDR5 parts such as LPDDR5X‑9600 and niche CPUs like Nvidia’s N1 and N1X chips are in short supply. He said supplier pricing lags consumer pricing by…
1 source HN 33Tom's Hardware -
Nvidia's RTX 60 GPUs might not launch until 2028, leaker says
Hardware leaker Kopite7Kimi posted on X that Nvidia's upcoming RTX 60 series, built on the Rubin architecture, may not be released until at least 2028. The leaker previously thought a new generation could appear in…
1 sourceTom's Hardware -
Apple is reportedly developing enterprise AI servers powered by M8 Ultra chips
Apple is developing an enterprise AI server utilizing its own M8 Ultra chips, according to a report from The Information. The hardware would target businesses, governments, and developers seeking to run AI inference…
2 sourcesfinance.yahoo.com -
Nvidia says its GPUs contain 10 to 40 RISC‑V cores, totaling over a billion in 2024
Nvidia reports that every GPU includes a set of 10 to 40 small RISC‑V management processors, which it estimates adds up to more than a billion cores shipped in 2024. These cores are not part of the shader array; they…
1 source HN 42xda-developers.com -
Humanoid robot decommissioning faces technical, safety, and data risks
Humanoid robots contain between 10,000 and 15,000 individual parts, organized into 200 to 500 major sub‑components. Their architecture includes 20‑40 electric motors with integrated drive modules, a kinematic skeleton…
1 sourceThe Robot Report -
Noctua says its NF-A4x10 5V PWM fan cools CAIM1 anti-AI 4K camera
Noctua announced its NF‑A4x10 5V PWM fan is being used in the CAIM1 camera, a niche 4K device built to prove that captured media have not been altered or generated by AI. The camera runs high‑bitrate imaging and…
1 sourceTom's Hardware -
Apple's A20 Pro enables iPhone 18 Pro to run 27B model twice as fast as iPhone 17 Pro
Apple introduced the A20 Pro chip, featuring a dual‑16‑core Neural Engine designed for on‑device AI. In a recent demo, an iPhone 18 Pro executed a 27B‑parameter Bonsai model at double the speed of an iPhone 17 Pro,…
1 sourcewccftech.com -
Apple and OpenAI pursue opposite hardware strategies for the AI era
The hardware landscape is splitting into two distinct directions as artificial intelligence advances. Apple has introduced its first foldable phone, the "iPhone Duo," which expands from 5.4 inches to 7.6 inches to…
1 sourcenote.com -
AMD Radeon RX 10800 XT could outperform RTX 5090, insider leaks suggest
Insider leaks attribute massive specifications to AMD's upcoming flagship graphics card, the Radeon RX 10800 XT, which is based on the RDNA 5 microarchitecture. Leaked details suggest the chip could feature up to 200…
1 sourceen.gamegpu.com -
Petlibro launches Granary 2 AI feeder series for multi‑cat homes
Petlibro’s Granary 2 series adds four new automatic feeders for dry food, priced from $129.99 to $249.99. The entry‑level Granary 2 provides app‑controlled feeding and precise portioning, while the $189.99 Granary 2…
1 sourceTechCrunch AI -
Valve releases Steam Frame VR headset that works as standalone and PC device
Valve announced the Steam Frame, a new VR headset that can function both as a standalone unit and when connected to a PC. Review writer Brandon Hill tested the hardware and interviewed Valve engineers about the design…
1 sourceTom's Hardware -
Meta develops camera-free 'Luna' smart glasses to address privacy concerns
Meta is reportedly developing a new smart glasses model named Luna, which notably lacks the integrated cameras that have drawn significant privacy criticism from consumers. The device is designed to function as an…
4 sourcesThe Rundown AIengadget.comTechCrunch AIyourstory.com -
Apple launches iPhone 18 Pro and Pro Max worldwide as CEO John Ternus greets fans
Apple began selling the iPhone 18 Pro and iPhone 18 Pro Max on Friday, with the Pro starting at $1,199 and the Pro Max at $1,299. Both phones are also offered through Apple’s Upgrade lease program. The devices feature…
2 sourcesibtimes.com.aunews.com.au -
Apple intelligence in iOS 27 needs up to 14 GB on iPhone 17 Pro models
Apple announced that its Apple Intelligence features in iOS 27 require considerably more on‑device storage, with the requirement climbing from 8 GB to as much as 14 GB for certain newer iPhones. According to Apple’s…
1 sourcemacobserver.com -
AMD says 96-core Venice CPU is about 20% faster than Nvidia Vera in SPEC CPU 2026
AMD claims its 96-core, high‑frequency Venice processor is around 20% faster than Nvidia's 88-core Vera in the SPEC CPU 2026 Integer Rate benchmark. AMD also reports a 78% speedup over its prior 192-core EPYC 9965 and…
1 sourceTom's Hardware -
Apple plans 2029 AI server launch using M8 Ultra and Nvidia NVLink tech
Apple is reportedly preparing to re-enter the server hardware market, a segment it abandoned in 2011 with the discontinuation of the Xserve line. Driven by surging demand for efficient compute resources in the AI…
9 sourcesTom's Hardwarefinance.yahoo.comArs Technica AIThe Decoder +4 more -
Engram offloading boosts DeepSeek-V4.1-Flash performance on B300 by up to 1.6x
The SemiAnalysis report describes Engram, a token‑embedding extension that stores recurring patterns in a lookup table. By moving the Engram table—about 24 rows per layer, roughly 12.4 KiB per token position (3.1 KiB…
1 sourceSemiAnalysis -
Apple reportedly developing AI-powered smart glasses without built-in display
Apple is reportedly working on a new pair of smart glasses that rely on AI and Siri rather than a traditional heads‑up display. The devices would use cameras, microphones and open‑ear audio speakers to interpret the…
1 sourcegeeky-gadgets.com -
AI developer vibe codes DLSS 5 onto Intel CPU's integrated graphics —Intel Arc 140T runs neural rendering in 360p at 10 frames per second
A GitHub repository named "dlss-nr-on-intel" claims to port NVIDIA's DLSS 5 neural rendering to Intel's Xe architecture. The author, who uses the handle Uzbekunknown, focused on the Intel Arc 140V GPU in a Lunar Lake…
1 sourceTom's Hardware -
Modder reportedly runs Nvidia DLSS 5 in a web browser using WebGPU
Modder MAAN reportedly got Nvidia DLSS 5 running in a browser window with an interactive demo, according to a report by VideoCardz. The demo is hosted on Cloudflare Workers, starts with a scene called "Cowboy Gramps",…
1 sourceTom's Hardware -
Huawei Announces New AI Chips to Compete with Nvidia
Huawei Technologies is set to unveil new artificial intelligence (AI) chips, the Ascend 960, at its annual summit in Shanghai this week. These chips are designed to replace those of Nvidia and compete globally. The…
5 sourcesTom's HardwareTechCrunch AIsify.comirishtimes.com +1 more -
Meta Plans In-House AI Chip Deployment in 2027
- Meta is planning to deploy its internally designed AI chip, Arke (MTIA 450), in data centers starting in the first half of 2027. The chip aims to improve cost and energy efficiency for running AI models. - Meta has…
3 sourcesfinance.yahoo.compulse2.com -
AMD may raise prices about 10% across GPUs, chipsets, possibly CPUs, report says
Industry sources say AMD is preparing to lift prices by roughly 10% on most of its product lines, including graphics cards, motherboard chipsets and potentially its CPUs. The move is attributed to higher wafer costs…
1 sourcetechpowerup.com -
Delta launches 800 VDC power and liquid‑cooling infrastructure for NVIDIA DSX AI factories
Delta Electronics announced a new integrated power‑and‑cooling solution designed to accelerate the deployment of AI factories built on NVIDIA’s DSX platform. The offering combines 800 VDC power delivery with conversion…
1 sourceUnite.AI -
OpenAI CFO says spending will follow ROI, may not slow AI buildout, Cramer asks
Jim Cramer sat down with OpenAI chief financial officer Sarah Friar, who said the company bases its investment choices on a strong return on investment and did not commit to slowing its AI buildout, despite broader…
4 sources247wallst.cominvezz.comfool.comibtimes.co.uk -
Nvidia’s Inference Opportunity: Beyond Training
Nvidia (NVDA) has dominated the AI training market with its GPUs, powering most server-based AI models. However, a key opportunity investors may be overlooking is the inference phase – where trained AI models are…
3 sourcescryptobriefing.comfool.com -
Micron, SanDisk and SK Hynix Selloff: AI Safety Concerns and Memory Scarcity
Micron, SanDisk, and SK Hynix experienced significant declines in US premarket trading on Monday. These companies are at the center of an AI selloff due to concerns over potential slowdowns in artificial intelligence…
1 sourceinvezz.com -
Fujitsu Launches 2nm Monaka CPU and Sovereign AI Server for Secure Inference
Fujitsu has announced the global launch of its FUJITSU-MONAKA CPU and the associated MONAKA Server, positioning them as critical infrastructure for sovereign AI. Starting in November 2026, the 2nm processor will be…
2 sourcesUnite.AIphoronix.com -
Meta donates AI glasses to blind veterans in Cincinnati, says it will help 130,000 veterans
Meta is donating a pair of Ray‑Ban Meta AI glasses to legally blind veterans in Cincinnati as part of a pledge to provide the technology and training to more than 130,000 veterans nationwide, an emailed statement said.…
2 sourceswlwt.comyahoo.com -
Alphabet shares rise as Crux AI secures $22B debt financing for TPU expansion
Alphabet’s cloud venture Crux AI, a joint effort with Blackstone, has arranged a $22 billion debt facility from a ten‑bank syndicate, using its Tensor Processing Units and existing customer contracts as collateral.…
1 sourceblockonomi.com -
Huawei says OceanStor M900 delivers 64 PB of KV cache for AI inference
Huawei unveiled the OceanStor M900 Context Memory Storage at HUAWEI CONNECT 2026, claiming a single cluster can hold 64 PB of KV cache for large‑scale AI inference. The system, built on the company’s UnifiedBus…
1 sourceUnite.AI -
Chip Equipment Stocks Fall Amid AI Pacing Concerns
Chip equipment stocks like Applied Materials, Lam Research, and ASML have seen significant drops. After Anthropic CEO Dario Amodei called for slowing the pace of AI model improvements over the weekend, these companies'…
1 source247wallst.com -
CoreWeave shares rise after deploying first Nvidia Vera Rubin NVL72 cluster
CoreWeave Inc. shares increased by 4.5% following the announcement that the company has successfully deployed its first multi-rack Nvidia Vera Rubin NVL72 cluster. This infrastructure marks a significant milestone for…
3 sourcesblockonomi.comfinance.yahoo.cominvezz.com -
Arm CEO Boosts Confidence in $2B AGI CPU Revenue Target
Arm Holdings’ chief executive Rene Haas told CNBC’s Jim Cramer that the company’s new data‑center chip, the AGI CPU, is on track to hit a $2 billion revenue goal. The chip marks Arm’s first in‑house CPU, a shift from…
1 sourceCNBC Technology -
NVIDIA NVRx on AWS EKS cuts GPU training downtime via async checkpointing
AWS has published a technical guide demonstrating how to integrate NVIDIA Resiliency Extension (NVRx) into PyTorch Fully Sharded Data Parallel (FSDP) training on Amazon Elastic Kubernetes Service (EKS). The solution…
1 source primary sourceAWS Machine Learning Blog -
NVIDIA Vera Rubin NVL72 claims 30x throughput gain for agentic AI workloads
NVIDIA announced significant efficiency advancements for its Vera Rubin NVL72 platform at the AI Infra Summit, emphasizing a shift from peak performance metrics to validated tokens per megawatt. The company highlighted…
6 sources primary sourceUnite.AINVIDIA Blogfinance.yahoo.comSemiAnalysis -
PieceMakers debuts on Taiwan board, betting on hybrid-bonded DRAM for edge AI
PieceMakers, a DRAM design firm backed by Nanya Technology, began trading on Taiwan’s Emerging Stock Board on September 16, 2026. The company values at approximately $1.4 billion and is pivoting from standard memory…
1 sourceTom's Hardware -
Single-phase direct liquid cooling validated for next decade of ultra-dense AI compute
As AI accelerators and high‑performance CPUs push past 1,000 W per chip and rack power exceeds 100 kW, traditional air‑based cooling is hitting its limits. A new white paper from IEEE Spectrum and Wiley, sponsored by…
1 sourceIEEE Spectrum AI -
Syensqo leverages AI to accelerate advanced materials for AI infrastructure
Syensqo, a global specialty materials company, is addressing the physical limits of AI infrastructure by developing high-performance materials for semiconductors and data centers. Mike Finelli, the company’s chief…
1 sourceMIT Technology Review AI -
AWS orders clients to migrate data from Middle East sites after drone strikes
Amazon Web Services has instructed customers to migrate their workloads away from data centers in Abu Dhabi and Bahrain, six months after the facilities were damaged by Iranian drone and missile attacks. The company…
2 sources HN 11Tom's HardwareCNBC Technology -
Nvidia Denies RTX 5090 Warranty Over Faded Serial Number
A GeForce RTX 5090 Founders Edition owner in the UK reported that Nvidia rejected a warranty claim after the card’s metal bracket serial number faded and became unreadable. The user, Willing‑Avocado‑4830, had…
1 sourceTom's Hardware
