Today's Hardware Pulse
Nvidia driver adds DLSS 5 profile options, hinting at next‑gen AI rendering
Nvidia's GeForce Game Ready driver 610.47 introduces three new profile settings—Enable DLSS NR override, Enable DLSS NR SL override, and Override DLSS NR presets—suggesting continued development of its upcoming DLSS 5 neural‑rendering technology. The earlier DLSS 5 reveal sparked debate over AI‑driven image alteration, after which Nvidia fell silent on the feature.
Also developing:
By the numbers: Cyient Semiconductors raises $30M Series A
Will Digital ‘Dynamic Pricing’ Tags Help or Harm Customers?
Electronic shelf labels (ESLs) have moved from pilot projects to core operations, with Woolworths deploying roughly 17 million tags across more than 770 ANZ stores and Bunnings following suit. The digital tags allow prices to be updated centrally in minutes, opening the door to real‑time or near‑real‑time pricing adjustments. While the technology promises reduced errors, paper savings and faster promotions, regulators stress that any price changes must be clearly communicated, and political pressure warns against aggressive dynamic pricing. Consequently, Australian grocers are likely to restrict price swings to overnight cycles rather than hourly fluctuations.
Jensen Holds Court
Jensen Huang delivered a two‑hour, unscripted keynote at Nvidia’s GTC in San Jose, addressing roughly 30,000 attendees. He combined rapid technical detail with a Jobs‑style humor, even ending with a musical campfire number featuring robot singers. The presentation emphasized Nvidia’s...
ZimaCube 2 Personal Cloud NAS Opens for Pre-Order with Multiple Configurations
IceWhale has launched pre‑orders for the ZimaCube 2, a compact personal‑cloud NAS built on 12th‑Gen Intel i3 and i5 processors. The system offers DDR5 memory up to 64 GB, dual 2.5 GbE Ethernet with optional 10 GbE, and two PCIe slots for GPUs or...
How to Build High-Performance GPU-Accelerated Simulations and Differentiable Physics Workflows Using NVIDIA Warp Kernels
The MarkTechPost tutorial demonstrates how NVIDIA Warp lets Python developers write GPU‑accelerated kernels for scientific simulations and differentiable physics. It walks through environment setup, kernel creation for vector math, signed‑distance fields, particle dynamics, and a gradient‑based projectile optimizer. Performance tests show...
ASRock AI BOX-A395 Runs Ryzen AI Max+ 395 with Up to 128GB LPDDR5x
ASRock Industrial unveiled the AI BOX‑A395, a compact bare‑bone system built around AMD’s Ryzen AI Max+ 395 platform. The device combines a 16‑core Zen 5 CPU, integrated Radeon 8060S GPU, and an XDNA 2 NPU delivering up to 50 TOPS, all fed by up to 128 GB of 8000 MT/s...
Market View: Nvidia’s $1 Trillion AI Bet, Buffett’s Oil Windfall & Markets on the Move
In this episode, Michelle Martin and Ryan Huang dissect Nvidia’s bold $1 trillion AI revenue forecast, highlighting the shift to agentic AI, the demand for next‑gen Blackwell and Vera Rubin chips, and even Nvidia’s space‑based data‑center ambitions. They then pivot to Warren...
Every Pool Robot Needs Cleaning, Except This One
Beatbot unveiled the AquaSense X, a premium robotic pool cleaner paired with AstroRinse, the first self‑cleaning docking station that rinses the filter, empties debris and wirelessly charges the unit. Priced at $4,250, it ships on April 30 2026 and boasts 29 sensors, AI...
NVIDIA GTC Signals OpenAI’s Shift Toward Enterprise
Visiting the NVIDIA GTC expo hall shows you why OpenAI is changing strategy. All the big booths are enterprise. Does this mean @sama is gonna give up his dreams of stealing Tim Cook’s lunch? The biggest news here is how NVIDIA is bringing...
WEKA Releases NeuralMesh AI Data Platform Based on NVIDIA AI Data Platform Design
WEKA announced the general availability of its NeuralMesh AI Data Platform (AIDP), an enterprise‑ready, composable storage solution built on NVIDIA’s AI Data Platform reference design. The platform promises to shrink AI project timelines from months to minutes by delivering high‑performance,...
Penguin Solutions Introduces Industry’s First Production-Ready CXL-Based KV Cache Server
Penguin Solutions unveiled the MemoryAI KV cache server, the industry’s first production‑ready key‑value cache built on Compute Express Link (CXL) memory. The appliance combines 3 TB of DDR5 with up to eight 1 TB CXL add‑in cards, delivering up to 11 TB of...
Meta Renewing Investment Into The Jemalloc Memory Allocator
Meta has announced a renewed commitment to the jemalloc memory allocator, a component it has used for nearly two decades across its infrastructure. The company plans to modernize the codebase, reduce technical debt, and enhance features such as the hugepage...
NVDA's $1T Demand Signal Highlights Heterogeneous Compute Moat
$NVDA just dropped the biggest forward demand signal in semiconductor history. Jensen Huang disclosed at GTC 2026 that he sees at least $1T in Blackwell and Vera Rubin demand through 2027, up from $500B in forward visibility just 12 months...
China Approves First Invasive Brain‑computer Interface, Launches Research Hubs
“Days after Chinese regulators approved…the world’s first invasive brain-computer interface device cleared for commercial use in patients, Shanghai announced three new research centers aimed at accelerating the technology’s development.” @SixthTone https://t.co/q1mrOAY82F
Coda Octopus Group Inc (CODA) Q1 2026 Earnings Call Transcript
Coda Octopus Group reported fiscal‑year 2025 revenue of $26.6 million, up 30.7% year‑over‑year, driven by a 30.5% jump in hardware sales and contributions from its newly acquired acoustic sensors unit. The company launched the ultra‑compact Echoscope NanoGen sonar and delivered the...
Robotic Tea Harvester Cuts Manual Labor Drastically
#WhosNext? Tea harvesters? This motorized tea plucking machine eliminates a lot of manual labor and as a result laborers. (GiGadgets) #AgTech #Automation https://t.co/saGtJTH3Dk
Check and Update Your Kobo eReader Firmware Easily
Have a @Rakuten @Kobo eReader? Smart choice, it's quite as capable as an Amazon Kindle, with strengths all its own. But is it up-to-date with system software? Here's how to check (and update it, as needed): https://t.co/L67emcu2fa #rakuten #kobo #software...
TORMAX iMotion Operators Automate V&A East Storehouse Entrance
TORMAX UK installed its iMotion 1301 door drives to automate a pair of heavy, custom‑fabricated swing doors at the new V&A East Storehouse museum entrance. The gearless, high‑torque motors deliver smooth, silent operation while handling the doors' substantial weight and...
NVIDIA Brands Itself Before Customers Need Its Chips
NVIDIA GTC 2026: The Company That Made Sure You Knew Its Name Before You Needed Its Chips https://t.co/BQ4crizikL
Jensen Huang Predicts $1 Trillion GPU Orders by 2027
Jensen Huang forecasts $1 trillion in orders for Blackwell and Vera Rubin through ’27. The technical term for this is "a lot." https://t.co/0mqhlqtnzA
IDC Significantly Lowers PC Sales Forecast Amid Chip Shortage
IDC has sharply lowered its 2026 PC market outlook, now forecasting an 11.3% decline in shipments, far steeper than the previous 2.4% drop. Tablet shipments are also projected to fall 7.6% amid persistent memory shortages, rising component prices, and ongoing...
NVIDIA Offers Modest 5× Boost, Not 35× Leap
NVIDIA does it again. NVIDIA gives us more exponential multipliers, but not the one that could be 35x by itself. I suppose we can live with 5x out of existing infrastructure. 🙂 Here's a great Synopsis of the GTC Keynote:
Pasqal Integrates NVIDIA CUDA-Q with QRMI for Slurm-Native Hybrid HPC-Quantum Workflows
Pasqal announced the integration of NVIDIA’s open‑source CUDA‑Q platform with its Quantum Resource Management Interface (QRMI), allowing quantum workloads to be scheduled through standard Slurm job queues. The coupling makes Pasqal’s neutral‑atom QPUs appear as native accelerators alongside CPUs and...
NVIDIA and Bolt Team up for European Robotaxis
At GTC 2026, NVIDIA and European rideshare firm Bolt announced a partnership to develop robotaxis across Europe. Bolt will leverage NVIDIA’s Cosmos, Omniverse, the Alpamayo autonomous‑vehicle model, and Drive Hyperion platform to turn its fleet data into a scalable learning...
NVIDIA Launches Vera CPU, Purpose-Built for Agentic AI
NVIDIA unveiled the Vera CPU, the first processor engineered specifically for agentic AI and reinforcement‑learning workloads. The chip claims 50% higher speed and twice the energy efficiency of conventional rack‑scale CPUs, leveraging 88 custom Olympus cores and LPDDR5X memory. A...
Kodiak Teams with NVIDIA to Power Truck‑Scale Driverless AI
The era of Physical AI requires us to solve the hardest problems in unpredictable conditions in real time. @KodiakRobotics is doubling down on its ability to do exactly that. We announced a collaboration with @nvidia today in which our next-generation driverless...
HPE Unveils Next-Gen AI Factory and Supercomputing Advancements with NVIDIA
HPE announced a suite of next‑generation AI factory and supercomputing solutions co‑engineered with NVIDIA, featuring the first NVIDIA Vera CPU compute blades, Quantum‑X800 InfiniBand switches, and the Vera Rubin NVL72 rack‑scale system for trillion‑parameter models. The portfolio also adds the...
Schneider Electric Teams with NVIDIA to Develop Validated Blueprints to Design, Simulate, Build, Operate and Maintain Gigawatt-Scale AI Factories
Schneider Electric, NVIDIA and AVEVA unveiled a suite of validated blueprints to design, simulate, build, operate and maintain gigawatt‑scale AI factories. The new NVIDIA Vera Rubin reference design confirms 480 VAC power distribution and 45 °C loop temperatures, while supporting MaxP/MaxQ operating...
Groq Integration and Nemoclaw Promise AI Security, HPC Boost
I had a chance to speak with Josh Lipton at @YahooFinance about the biggest takeaways from today's @nvidia GTC event. At least two things stood out for me. One in Nemoclaw and it's potential to be used for...
GTC 2026 - Everpure Tackles Data Readiness and Flexible Consumption for Enterprise AI
Everpure unveiled three major announcements at GTC 2026, beginning with record‑setting AI storage benchmarks for its FlashBlade//EXA and FlashBlade//S500 platforms. The company introduced DataStream, an AI‑ready data pipeline that automates ingestion, anonymization, vectorization and indexing, complemented by the 1touch.io acquisition...
Physical AI to Power Nvidia’s $70B Growth
Physical AI gets its time in the keynote (as expected). I believe physical AI is going to be a meaningful growth driver for Nvidia starting in CY28. Today, only about 1% of Nvidia’s sales are used to deliver physical AI, which...
AI-Driven Automation in Semiconductor Design: The Fuse EDA AI Agent
Siemens has launched the Fuse EDA AI Agent, an agentic AI platform that automates and optimizes electronic design automation across the full chip development lifecycle. The system targets the soaring complexity of modern SoCs, where design costs can exceed $300 million...
Dell’s AI Story Electrified by Lightning
Dell unveiled an AI Data Platform built with Nvidia that adds a four‑layer architecture and a new parallel file system called Lightning. Lightning delivers up to 150 GB/s per 1‑RU enclosure and can be stacked to achieve roughly 6 TB/s per rack,...
Synopsys Builds Hardware-Accelerated Agentic AI Stack with Nvidia
Synopsys announced a collaboration with Nvidia to create an open, secure, hardware‑accelerated agentic AI stack spanning from silicon design to system‑level applications. The stack leverages Nvidia GPUs and AI accelerators to deliver high‑performance inference and training for agentic models. In...
Memories AI Is Building the Visual Memory Layer for Wearables and Robotics
Memories.ai announced a partnership with Nvidia to build a visual‑memory layer for AI wearables and robotics, using Nvidia's Cosmos‑Reason 2 and Metropolis tools. The startup launched its Large Visual Memory Model (LVMM) in July 2025, positioning it against Google’s Gemini Embedding 2. It...
Nvidia DGX Spark and DGX Station with NemoClaw Deliver Autonomous Agents Platform
Nvidia announced that its DGX Spark and DGX Station systems, paired with the NemoClaw software stack, now deliver a full‑stack platform for building and running autonomous AI agents. The solution is designed to support always‑on agents that can reason, plan,...
Agilent Launches S540MD Slide Scanner System
Agilent Technologies introduced the S540MD slide scanner, a whole‑slide imaging system now on sale in select European markets. The scanner holds up to 540 slides, supports continuous loading, standard rack compatibility, automated scanning modes, and AI‑assisted tissue detection. It is...
GNU C Library Lands X86_64 FMA'ed Cosh For A ~35% Improvement
The GNU C Library (glibc) now includes an FMA‑optimized implementation of the hyperbolic cosine function, cosh(), delivering roughly a 35% speed boost on x86_64‑v3 Intel and AMD CPUs. The change was merged by Linaro engineer Adhemerval Zanella alongside other core‑math...
What’s New in the AirPods Max 2? Breaking Down Apple’s Headphone Update
Apple unveiled the AirPods Max 2, a refreshed version of its premium over‑ear headphones, priced at $549 and shipping in April. The update centers on the H2 chip, delivering 1.5× stronger active noise cancellation, a new amplifier, and software features like...
Rubin+Groq Shows Custom Accelerators Becoming Unnecessary
The Rubin + Groq narrative is the best case $NVDA has made to date about less real need for custom accelerators. I don't think it changes the calculus for certain hyperscalers, but as it comes to silvering for training and...
NVIDIA’s Visual Guide to Scaling Up, Out, Across
In case it helps here is a visual on how NVIDIA describes scale up, out, and across. https://t.co/BVhQbE17Qk
Dynamo + VAST Makes KV Cache Shareable, Saves GPU Memory
Inference doesn’t break on compute these days, it's GPU memory that's the choke. But Dynamo + VAST turns KV cache into a first-class, shareable system resource so GPUs stop redoing context. Cool stuff https://t.co/4v0Au4rcM5
Nvidia Expands AI Inference Beyond GPUs with Processor Family
Nvidia GTC 2026: @Nvidia’s hardware strategy goes beyond GPU in AI inference pivot https://t.co/1ZWled5xXB >> It takes a family of processors to build the #AIFactory. @LDignan is first out. #NvidiaGTC
Run OpenClaw on $59 Arduino UNO Q
Hey, did you know you can run OpenClaw on the $59 @arduino UNO Q? Amazing. https://t.co/Nrl1lpBYdY
NVDA Hits $1 Trillion Market Cap, Sparks Competition Talk
Enjoyed chatting with @CNBCOvertime on the $NVDA keynote while it was still going on inside. Paired up with @dylan522p. The $1T number got the highlights. Talked about the competition, as well. https://t.co/QwmyZ9SZRx
Nvidia's 750GB Unified-Memory Dell Pro Max Arrives
.@nvidia hand delivered a pre-production unit of the @Dell Pro Max with GB300 to my house. 100lbs beast with 750GB+ of unified memory to power the best open-source models in the world. What should I test first? https://t.co/fmHXIUr1NR
Iran Conflict Jeopardizes Chip Supply, Hikes TSMC Costs
Iran Conflict Threatens Global Chip Supply Chain, Risks Higher Costs for TSMC - Bloomberg https://t.co/nu9qEGPpB0
Jensen Huang Hails AI Natives Powered by Nvidia
And Jensen Huang break a lance for the AI natives that run on @Nvidia. #NVIDIAGTC https://t.co/8tNGm7JDOo
Nvidia Nails Heterogeneous Computing Showcase at GTC
$NVDA was going to make its best case yet for heterogenous computing at GTC. Mission accomplished today. What I told @cvpayne today as the keynote kicked off. 👏🏻💪🏻🚀 https://t.co/Jxor9SWxXD
CudaJet Unveils High‑Performance Underwater Jetpack
CudaJet: A High-Performance Underwater Jetpack Redefining Aquatic Mobility via @WevolverApp #Innovation #EmergingTech #Technology #Tech https://t.co/cF5kNZliKd
NVIDIA's BlueField-4 STX Redefines Storage for Agentic AI
NVIDIA Launches BlueField-4 STX Storage Architecture With Broad Industry Adoption With agentic AI, storage must evolve to meet the demands of real-time responsiveness and longer context windows. #NVIDIAGTC @nvidianewsroom @NVIDIAGTC https://t.co/RWwx1pijbv