Hardware & Chips
Analyses of news about the hardware behind AI: GPUs and accelerators, chipmakers, data centers, supply chains and export controls. Page 4 of 7.
Inference Compute Brokerage: Parasail Raises $32M Amid Shift to Open Models
Parasail, a cloud computing service, raised $32 million to capitalize on the growing demand for cheap, flexible inference compute as the AI industry shifts toward open-source model architectures.
OpenAI Mandates macOS Updates After Third-Party Supply Chain Security Breach
Following the compromise of a third-party developer tool, OpenAI is forcing all macOS users to update their desktop applications to new versions signed with a rotated security certificate.
Safetensors Joins PyTorch Foundation, Signaling Move to Vendor-Neutral Ecosystem Governance.
The widely adopted, secure model format, Safetensors, has formally moved under the PyTorch Foundation, securing its future governance within the broader ML community.
Intel Lands as Core Builder for Musk's Ambitious Terafab Chip Initiative
Intel has agreed to participate in Elon Musk's massive project (Terafab) with SpaceX and Tesla to build a new semiconductor facility in Texas, aiming to boost AI compute capacity.
Anthropic Secures Massive Compute Power from Google and Broadcom to Fuel Claude's Growth
Anthropic announced a major compute expansion agreement with Google and Broadcom, committing billions in capacity to power its Claude models through 2027.
Spanish Startup Xoople Raises $130M to Build Next-Gen Satellite Data Constellation for Enterprise AI
Xoople, a Spanish space data startup, secured $130M in funding and a major contract with L3Harris to build a highly precise satellite constellation aimed at becoming a 'System of Record' for enterprise-grade AI training data.
Holo3: Agentic AI Redefines Enterprise Computer Use
Hcompany's Holo3 achieves state-of-the-art performance on benchmark tests thanks to its agentic learning flywheel, which trains it to autonomously execute real-world workflows within synthetic enterprise environments. It achieves this with a fraction of the parameters of leading models.
AI Takes Control of Your Stream Deck
Elgato’s Stream Deck 7.4 update introduces Model Context Protocol (MCP) support, allowing AI assistants like Claude and ChatGPT to trigger Stream Deck actions automatically.
Anthropic’s Code Leak: A Recurring Packaging Issue
Anthropic experienced a recurring data leak incident, exposing source code for its Claude Code developer tool, highlighting ongoing challenges with secure software release processes.
IBM Granite 4.0 3B Vision: Focused Chart & Table Extraction Updates
IBM has released Granite 4.0 3B Vision, a compact vision-language model optimized for extracting data from complex documents, particularly charts and tables, through a modular design and enhanced data understanding capabilities.
Bezos Eyes $100B Fund to Automate Factories, AI Shifts Focus
Jeff Bezos is launching a $100 billion fund to revitalize aging manufacturing facilities using automation and AI, prompting a shift in focus for companies like Sift Stack, which specializes in data infrastructure for complex machines.
Lucid Bots Raises $20M, Focusing on Practical Robotics
Lucid Bots, a robotics startup specializing in drone window cleaning solutions, recently secured a $20 million Series B funding round led by Cubit Capital and Idea Fund Partners. The company is focused on deploying its Sherpa and Lavo robots to cleaning companies, demonstrating steady sales growth and expanding its application capabilities.
Arm Goes Silicon: A Historic Shift
Arm Holdings, a major chip designer, is entering the chip manufacturing business with the Arm AGI CPU, marking a significant departure from its long-standing licensing model.
Nvidia's Olaf Robot Demo: Hype vs. Reality
A recap of Nvidia CEO Jensen Huang's GTC keynote, including a memorable (and slightly chaotic) demo featuring a robot Olaf from Disney's ‘Frozen’.
Musk's Austin Chip Plant: Ambition Without a Timeline
Elon Musk plans a massive chip fabrication plant in Austin, Texas, aimed at scaling production for Tesla, SpaceX, and AI applications. However, no timeline has been provided.
Nvidia's Nemotron 3 Nano 4B: Edge-Optimized Model Released
Nvidia has unveiled Nemotron 3 Nano 4B, a compact hybrid language model designed for efficient inference on edge devices. Leveraging a novel Elastic compression technique and quantization strategies, it’s targeted at local AI applications like conversational agents.
Startup Targets Data Center Power Waste
Niv-AI, a Tel Aviv-based startup, is emerging with $12M in seed funding to tackle the growing problem of power waste in AI data centers. By deploying millisecond-level power usage sensors and developing predictive AI models, they aim to optimize GPU utilization and reduce strain on the electrical grid.
Community-Driven Dataset Accelerates Physical AI in Surgical Robotics
A collaborative effort by NVIDIA and a global network of institutions has released Open-H-Embodiment, a comprehensive dataset and accompanying models designed to accelerate the development of physically intelligent surgical robots.
Nvidia Projects $1 Trillion in Blackwell/Rubin Chip Demand
Nvidia CEO Jensen Huang announced a staggering $1 trillion in sales projections for the Blackwell and Vera Rubin AI chips, fueled by strong demand and performance improvements.
Memories.ai Partners with Nvidia to Build Visual Memory for AI Wearables
Memories.ai, a startup focused on enabling AI wearables and robotics with visual memory capabilities, has announced a collaboration with Nvidia to leverage Nvidia’s Cosmos-Reason 2 and Metropolis platform. This partnership aims to build the infrastructure needed for AI systems to ‘remember’ and recall visual data in real-world scenarios.
Frore Systems Hits Unicorn Status
Deep tech chip startup Frore Systems raises $143 million, achieving a $1.64 billion valuation.
Nvidia’s GTC Kicks Off: AI Chip Focus and Strategic Partnerships
Nvidia’s annual GTC developer conference begins in San Jose, highlighting announcements related to AI advancements, strategic partnerships, and new hardware developments aimed at accelerating AI inference.
U.S. Army Contracts With Anduril - $20B Deal Signals Military Tech Shift
The U.S. Army has signed a 10-year, $20 billion contract with defense startup Anduril, consolidating existing procurements and signaling a shift towards software-defined military capabilities.
Amazon S3 Celebrates 20 Years – A Storage Platform Evolving for AI
To mark its 20th anniversary, Amazon Web Services highlights the evolution of S3, emphasizing its increasing role as a foundational element for AI systems and data infrastructure.
Kalanick Diversifies Again: Robotics Venture 'Atoms' Gains Uber Backing
Travis Kalanick, co-founder of Uber, is launching a new robotics venture, 'Atoms,' backed by Uber. The company aims to automate tasks across restaurants, logistics, and mining, leveraging existing assets like City Storage Systems and Pronto AI’s autonomous driving technology.
AWS Integrates Cerebras’ WSE-3 Chip for Accelerated AI Inference
Amazon Web Services is partnering with Cerebras Systems to make the company’s WSE-3 wafer-size AI chip available on its cloud platform. This integration is designed to significantly speed up the generation of AI model outputs by a factor of five through a disaggregated architecture.
NanoClaw's Rapid Rise: Docker Integration Signals Community Momentum, Not Revolutionary Shift
The NanoClaw AI agent project, born from a weekend coding binge and amplified by viral attention on Hacker News and a notable endorsement from Andrej Karpathy, has just secured a key partnership with Docker. This integration, coupled with a large, engaged community, suggests growing momentum but doesn’t represent a fundamentally disruptive change in the AI agent landscape.
NVIDIA's KGMON Agent Toolkit Achieves SOTA on DABStep – Modular Reasoning Gains Traction
NVIDIA's KGMON Agent Toolkit achieves state-of-the-art performance on the challenging Data Agent Benchmark for Multi-step Reasoning (DABStep) through a novel modular reasoning approach, emphasizing reusable tool generation and efficient execution.
Robotic Greenhouse Startup Hits Automation Milestone - A Slow Burn
Canopii, a robotic greenhouse startup focused on autonomous crop production, has achieved a significant milestone: fully automated seed-to-harvest operations. The company's approach, emphasizing slow, capital-light development and avoiding traditional VC funding, contrasts with the rapid scaling attempts of other vertical farming ventures.
Sandbar Raises $23M to Bring Smart Note-Taking Ring to Market
Sandbar, a startup developing a smart note-taking ring, recently secured a $23 million Series A funding round led by Adjacent and Kindred Ventures. The company is focused on refining its app and reducing latency, with plans to ship the ring this summer.
Hugging Face Introduces Mutable Storage Buckets for ML Artifacts
Hugging Face has launched Storage Buckets, a new S3-like object storage solution built on its Xet backend, designed for efficiently managing the constantly changing intermediate files generated during machine learning training.
Neura Robotics and Qualcomm Partner on Robot 'Brain' Development
Neura Robotics and semiconductor giant Qualcomm have announced a partnership to develop the next generation of robots, leveraging Qualcomm’s IQ10 processors and Neura’s robotic simulation platform. This collaboration aims to accelerate the deployment of humanoid and general-purpose robots across various industries.
Nscale Secures $14.6B Valuation, Fueling European AI Infrastructure Ambitions
British AI infrastructure company Nscale has raised a $14.6 billion valuation, marking it as one of Europe's largest AI infrastructure firms. The substantial funding, driven by a Series C round and debt financing, will accelerate expansion across Europe, North America, and Asia, leveraging renewable energy and local skills.
LeRobot v0.5.0: Expanding Embodied AI Capabilities
LeRobot v0.5.0 delivers significant hardware and policy updates, including full humanoid support, enhanced dataset pipelines, and new, more responsive policies.
Netflix Acquires Ben Affleck’s AI Film Production Startup
Netflix has acquired InterPositive, Ben Affleck’s AI startup focused on streamlining film and television production workflows using AI-powered tools for post-production.
NXP Shares Best Practices for Deploying VLA Models on i.MX95
NXP provides a detailed guide on deploying Vision-Language-Action (VLA) models on their i.MX95 processor, focusing on dataset recording techniques and optimization strategies for embedded robotics applications.
Anthropic Rolls Out Voice Mode for Claude Code
Anthropic is introducing voice control to its AI coding assistant, Claude Code, allowing developers to interact with the tool through spoken commands. This feature is being rolled out gradually to a small percentage of users.
Xiaomi Prioritizes Hardware Over AI in New Leica Phone
Xiaomi’s new Leica Leitzphone emphasizes camera hardware advancements, including continuous zoom and a LOFIC sensor, signaling a shift away from a heavy reliance on AI processing, according to company executives.
Zuckerberg Spotted at Prada Show, Signaling Potential AI Glasses Collaboration
Mark Zuckerberg's attendance at Prada’s Fall/Winter 2026 Fashion Week show, alongside Prada’s Chief Merchandising Officer Lorenzo Bertelli, fuels speculation of a forthcoming collaboration involving Meta’s AI glasses.
Passive Cooling Breakthrough Could Transform Space Data Centers
Sophia Space is pioneering a novel approach to cooling space-based processors using a thin, sail-like design derived from solar plant research, potentially revolutionizing the economics of space computing.
Nvidia’s Explosive Growth Fuels AI Demand – But Key Risks Remain
Nvidia’s record quarter, driven by soaring AI compute demand, highlights the industry's transformation but reveals ongoing challenges regarding China market access and investment risk.
Google Intensifies Physical AI Push with Strategic Intrinsic Acquisition
Google is bolstering its foray into physical AI by acquiring Intrinsic, a robotics software company, leveraging Google DeepMind and Gemini AI models.
Wearable Startup CUDIS Launches AI-Powered Health Ring with Points System
CUDIS, a wearable health tech startup, is releasing its newest health ring featuring an AI ‘agent coach’ designed to motivate users toward healthier habits through a points system and reward redemption.
MatX Raises $500M, Positioning as Nvidia's Challenger
AI chip startup MatX secured $500 million in Series B funding, led by Jane Street and Situational Awareness, aiming to deliver significantly better performance than Nvidia's GPUs for training large language models.
Open Source VLM Deployment on Jetson Devices: A Practical Tutorial
This article details a hands-on tutorial demonstrating the deployment of the NVIDIA Cosmos Reasoning 2B VLM model on NVIDIA Jetson devices, utilizing vLLM and the Live VLM WebUI.
G42 and Cerebras Deploy 8 Exaflops Supercomputer in India
G42, in partnership with Cerebras Systems, is launching an 8 exaflops supercomputer in India to bolster local AI development, focusing on sovereignty and data residency.
Mirai: Optimizing AI Inference at the Edge
Mirai, a London-based startup, is developing a framework to improve AI model performance on consumer hardware, addressing the growing demand for edge AI. They’ve built an Apple Silicon inference engine and plan to expand to Android, targeting developers with a streamlined integration experience.
Freeform Raises $67M for High-Throughput Metal 3D Printing Platform
A startup, Freeform, developing a novel 3D printing system for metal components, has secured a $67 million Series B funding round led by investors including Nvidia and Two Sigma Ventures, aiming to scale its manufacturing platform.
Google Cloud VP Warns Startups on ‘Check Engine Light’ Cloud Costs
Google Cloud’s VP for startups, Darren Mowry, cautions founders about the potential for runaway cloud costs, urging them to monitor their infrastructure usage and recognize ‘check engine light’ signs early to avoid unexpected bills.
Amazon Halts Blue Jay Warehouse Robot Project, Signaling Tech Experimentation Challenges
Amazon has paused its Blue Jay warehouse robotics project, a multi-armed robot designed for package sorting, highlighting potential difficulties in scaling its ambitious robotics initiatives despite significant investment and rapid development timelines.
Sarvam Bets on Edge AI: Deploying Models on Feature Phones and Beyond
Indian AI startup Sarvam is launching its AI models on a surprising range of devices, including Nokia feature phones, cars, and smart glasses, leveraging edge AI technology for offline operation and local language support.
Apple's AI Wearables Race Heats Up
Apple is accelerating the development of three new AI-powered wearables – an AirTag-sized pendant, smart glasses (N50), and revamped AirPods – signaling a significant push into the burgeoning wearable AI market.
Apple Poised to Launch Smart Glasses, Pendant, and AI AirPods
Apple is reportedly developing a suite of smart devices, including AI-powered glasses, a pendant, and camera-equipped AirPods, designed to integrate seamlessly with the iPhone and leverage Siri’s capabilities based on real-time environmental context.
Mesh Optical Technologies Secures $50M to Disrupt AI Data Center Interconnect
Mesh Optical Technologies, founded by former SpaceX engineers, has raised a $50 million Series A led by Thrive Capital to mass-produce optical transceivers for AI data centers. The company aims to provide a secure and efficient alternative to existing solutions dominated by Chinese suppliers.
India Launches Aggressive $200 Billion AI Infrastructure Push
India is embarking on a bold strategy to attract over $200 billion in AI infrastructure investment, leveraging existing commitments from tech giants and offering tax incentives, venture capital, and policy support to become a global AI computing hub.
Ricursive Intelligence: AI Chip Design Startup Raises $300M, Leveraging Google Brain Roots
Ricursive Intelligence, co-founded by former Google Brain employees Anna Goldie and Azalia Mirhoseini, has secured a $300 million Series A funding round led by Lightspeed, signaling a significant bet on their AI-powered chip design platform. The startup is leveraging their expertise and prior work on Google’s Alpha Chip to fundamentally change how computer chips are designed.
India's C2i Secures $19M to Tackle Data Center Power Efficiency
C2i Semiconductors, an Indian startup focused on plug-and-play power solutions for AI data centers, has raised a $19 million Series A round led by Peak XV Partners to address escalating energy demands.
Nvidia's Next Move: The Groq Gamble Signals a Shift in AI's Trajectory
Nvidia's strategic investment in Groq, a company specializing in ultra-fast inference, signals a critical pivot away from raw compute power and towards addressing the 'thinking time' latency crisis inherent in rapidly evolving AI models.
Meta’s Facial Recognition Smart Glasses Back on Track
Meta is planning to integrate facial recognition technology into its smart glasses, dubbed ‘Name Tag,’ potentially releasing the feature this year despite previous ethical concerns and technical challenges.
OpenAI’s Spark Model Launches, Prioritizing Speed Over Nvidia Dependence
OpenAI has released its first production AI model designed to run on non-Nvidia hardware via Cerebras’ Wafer Scale Engine 3, focusing on significantly faster code generation at 1,000 tokens per second. This move represents a key diversification strategy away from Nvidia’s dominance in AI compute.

