The latest news on OpenClaw, AI agents, and automation
Tag
110 articles
A hands-on test of Dell's $6,167 Pro Max with GB10 running OpenClaw and Qwen3.6 35B locally processed internal documents and accepted Telegram commands, with the device paying for itself in roughly one month for teams spending $187/day on cloud inference.
Leopold Aschenbrenner's AI-focused hedge fund delivered 439% net returns in H1 2026 by going long AI infrastructure and short disrupted software. Then July's selloff hit his leveraged positions. Within 24 hours, Ken Griffin's Citadel assembled a deal to buy the bulk of a $16 billion stock portfolio. The fund survives with a smaller book and its Anthropic stake intact, but the episode exposes a structural problem: even the most conviction-heavy AI thesis cannot survive 4x leverage when semiconductor stocks correct 15-20% in a month.
AI Infrastructure Capital AG, a newly founded Swiss company, closed €16 million in funding to buy NVIDIA GPU servers and operate them in Iceland on renewable power. The company rents compute capacity on long-term contracts, positioning itself as dedicated infrastructure for AI workloads that need persistent, cost-efficient GPU access.
As AI agents shift from generating text to writing and executing code autonomously, the hardware and infrastructure requirements change fundamentally. The Motley Fool reports that sandbox density, isolation speed, and state management are becoming competitive metrics alongside inference performance, with Amazon, Microsoft, and Google each building dedicated agent execution environments.
Nvidia committed approximately $5 billion in equity and priority access to its Vera Rubin GPU platform to Safe Superintelligence Inc., the stealth AI safety lab founded by former OpenAI chief scientist Ilya Sutskever. Nvidia said it made the deal after obtaining rare access to SSI's closely guarded research, which has produced no public papers or products in two years of operation.
Nvidia is negotiating a credit guarantee that would let OpenAI raise up to $250 billion in debt to build a 10-gigawatt data center campus in Pike County, Ohio, on a former uranium-enrichment site. The deal would make Nvidia the financial underwriter of the compute infrastructure required to train and run OpenAI's next generation of autonomous agent systems.
Three Harvard dropouts just closed the highest-valuation Sequoia-led Series C in history. Etched AI's $300M round values the inference chip startup at $10.3 billion, doubling from $5 billion in December, as demand for agent-scale inference infrastructure outpaces GPU supply.
NVIDIA brought together more than 50 technology companies to form the Open Secure AI Alliance, a coalition building open-source security tools for AI agents. Founding members include OpenClaw, Microsoft, Salesforce, CrowdStrike, Databricks, Hugging Face, and Palantir. The alliance will develop identity frameworks, isolation standards, and audit tools for agent deployments.
AMD and Anthropic announced a strategic partnership combining a $5 billion equity investment with a multi-gigawatt GPU deployment deal. Anthropic will deploy up to 2 gigawatts of AMD Instinct MI450-series GPUs in Helios rack-scale systems, with the first gigawatt beginning in the first half of 2027. The deal includes a multi-year engineering collaboration where both companies will use Claude to optimize AMD ROCm software development.
A coalition of 25 technology companies released a joint letter urging policymakers to avoid 'premature restrictions' on open-weight AI models, arguing that limiting access risks creating monopolistic control over AI development. OpenAI and Anthropic, both preparing for potentially massive IPOs, did not sign. The letter arrives as Chinese open-weight models gain adoption among U.S. companies and Congress considers new AI regulation after the OpenAI rogue system incident.
AMD will begin shipping Helios, its first rack-scale AI system, later this year with Microsoft deploying it in Azure data centers. Meta committed 1 GW on Helios this year out of a 6 GW AMD deal. Eight of the top 10 AI companies now run workloads on AMD Instinct GPUs.
Infinity announced a $15M Series A at a $100M valuation to scale Ignition, an AI research agent that writes, tests, debugs, and optimizes low-level inference code for non-Nvidia chips. The agent automates kernel development that would take human engineers months, reducing the process to hours or days. Customers include AI chip maker D-Matrix.
Etched, a San Jose startup building purpose-built inference chips, is negotiating two concurrent VC rounds that would value the company at $10 billion and $20 billion respectively. The company says it already has $1 billion in customer demand for chips that haven't shipped yet, backed by 400+ engineers and $800M in prior financing.
NVIDIA and Japanese startup Noetra Corp. are building a 140-megawatt AI factory with 27,500 Rubin GPUs and 13,750 Vera CPUs, backed by Japan's Ministry of Economy, Trade and Industry. The facility will train open multimodal foundation models for robotics, digital twins, and agentic AI as part of Japan's FRONTia Project.
Next Net and Sundial Media Group have launched the Standardized Agentic Intelligence Ledger (SAIL), a framework for how AI agents access, attribute, and compensate premium publishers. Built on NVIDIA's NeMo, RAPIDS, and NIM infrastructure, SAIL is the first scalable attempt to create a governance layer for agent-to-publisher content retrieval.
The world's largest technology companies are collectively raising over $200 billion through debt offerings, bond sales, and equity issuances to fund AI data center construction. SpaceX borrowed $25 billion and started buying AI companies. The shift from retained earnings to external capital signals that AI infrastructure buildout has exceeded the cash generation of even the most profitable firms.
NVIDIA's NemoClaw adds enterprise security governance, audit trails, and permission controls on top of OpenClaw's open-source agent framework. It installs with a single command and targets the gap between developer adoption and corporate deployment requirements.
NVIDIA Research released HORIZON, a hands-free AI agent that automates Register-Transfer Level chip design by treating each problem as repository-level code evolution. Using GPT-5.3 as its backbone and Git as an experience buffer, the system achieved 100% completion across all evaluated RTL benchmarks, with 91% of tokens coming from cache.
Anthropic's Claude Science is a multi-agent research workbench bundling 60+ scientific skills and database connectors alongside NVIDIA's BioNeMo toolkit. Beta users at Manifold Bio and the Allen Institute report compressing years of research work into days. Available now for Pro, Max, Team, and Enterprise plans.
OpenAI engineers found a way to halve the GPU compute required to serve ChatGPT users through software optimization alone. Asian chip stocks fell sharply on the news, with investors recalculating whether the $5.5 trillion AI infrastructure buildout thesis still holds when software can substitute for silicon.
AI inference cloud company Groq has closed $647 million in new growth funding led by Disruptive and Infinitum, using the capital to expand its 13-data-center network and serve over 5 million developers. The round follows Groq's $20 billion non-exclusive licensing deal with Nvidia in December 2025 and arrives as competitor Baseten secured $1.5 billion in its own inference-focused raise.
JPMorgan Global Research raised its AI capex forecast to $5.5 trillion through 2030 in its midyear outlook, up from $5.1 trillion. The bank now projects hyperscaler spending will hit $650 billion in 2026 and exceed $1.1 trillion in 2027, with AI-related debt financing reaching $4.1 trillion.
The risk premium on high-grade US tech bonds rose to 0.79 percentage points in June, up from 0.74 at the end of May. Bond investors are repricing risk on AI infrastructure spending as J.P. Morgan raises its five-year AI capex forecast to $5.5 trillion and projected bond financing to $2.1 trillion.
AI chip startup SambaNova Systems is seeking $800 million to $1 billion in new funding at an implied valuation of around $10 billion, according to a report from The Information. Intel CEO Lip-Bu Tan, who also serves as SambaNova's executive chairman, disclosed the fundraising plans during a public event.
OpenAI's first custom silicon is an inference-focused ASIC called Jalapeño, co-designed with Broadcom and built end to end in nine months using OpenAI's own models. A physical sample ships Wednesday, with prototype deployment in late 2026 and full-scale production expected by first half 2028.
Nvidia's Nader Khalil told The New Stack that the company has developers contributing to OpenClaw full-time, is building NemoClaw agent blueprints for enterprise harness adoption, and sees every industry building specialized agents. Nvidia is already working with CrowdStrike, Cadence, and Palantir on agent deployments.
Google, Meta, Amazon, and Microsoft are on track to spend $750 billion on AI infrastructure in 2026. But depreciation on existing equipment has nearly doubled to $116 billion annually, and Amazon has already shortened its server useful life from six to five years. The math behind the AI infrastructure race is getting harder to sustain.
NVIDIA's new open-source tool scans AI agent skills for 64 vulnerability patterns across 16 categories before installation. SkillSpector flags prompt injection, data exfiltration, and supply chain risks in skill packages used by Claude Code, Codex CLI, Gemini CLI, and other agent frameworks, producing standardized SARIF reports for CI/CD integration.
NVIDIA is marketing its next-generation Vera CPU to Chinese data center customers ahead of the processor's expected launch later this year. Built on a custom ARM architecture as the successor to Grace, Vera is designed to coordinate AI accelerators in data centers running generative and agentic AI workloads. The move signals NVIDIA's determination to hold its China position despite U.S. export controls on advanced GPU chips.
Chinese AI lab MiniMax released M3, a 428-billion-parameter mixture-of-experts model with native multimodal support and a one-million-token context window, on NVIDIA's accelerated infrastructure. NVIDIA is offering a free GPU-accelerated endpoint for the model, positioned for long-context reasoning and agentic workflows on Blackwell hardware.
Oracle beat Q4 earnings and revenue expectations but saw shares drop 7% after announcing plans to raise another $40 billion in financing for AI infrastructure. Net capital expenditures for fiscal 2027 will hit roughly $70 billion, with over half of the company's $638 billion backlog coming from OpenAI.
OpenAI is in advanced talks to lease a 10-gigawatt data center campus on federal land in Ohio, with Nvidia backing the project and a potential full buildout cost exceeding $500 billion. The lease structure means OpenAI would control the equipment without owning the campus, with the first phase expected to come online in 2028.
Super Micro Computer announced $7 billion in equity financing to fund component purchases for $39 billion in AI server orders from more than 20 customers. Shares fell 9% after hours on dilution concerns, exposing a hard constraint on AI deployment: even companies with massive order books can't fulfill them without raising billions in working capital.
Clive Chan left OpenAI's custom chip program to join Anthropic, bringing Tesla Autopilot ASIC and OpenAI-Broadcom silicon expertise to a company that had no dedicated hardware team two months ago. Combined with a $35 billion Apollo-Blackstone financing deal for Google TPUs and a Reuters report on custom chip exploration, Anthropic is assembling the infrastructure story that public market investors demand before a trillion-dollar debut.
Qualcomm CEO Cristiano Amon used a TIME essay to position agentic AI as the next paradigm shift comparable to the internet and smartphone. The timing is not accidental: Qualcomm just launched NPU-equipped chips for everything from $300 laptops to robotics at Computex, and every claim about agents needing better hardware maps directly to Qualcomm's product roadmap.
Four months after PCWorld warned readers not to install OpenClaw, the same publication now calls it the blueprint for mainstream AI agents. Nvidia's RTX Spark chip, Microsoft's Scout assistant, and a sandboxed Windows desktop version of OpenClaw from creator Peter Steinberger all shipped within the same week, putting agent-powered PCs months away from consumer shelves.
NVIDIA released Nemotron 3 Ultra on June 4, a 550-billion-parameter mixture-of-experts open model built for long-running AI agents. It delivers 5x faster inference and up to 30% lower cost than comparable open models. The release came alongside an expanded Agent Toolkit including NemoClaw blueprints, OpenShell secure runtime, and CUDA-X agent skills, with Cadence, Siemens, Synopsys, CrowdStrike, and Palantir among the first enterprise adopters.
Broadcom disclosed during its Q2 earnings call that it is financing custom AI chips for Anthropic, OpenAI, and other large language model developers through a vehicle backed by Apollo and Blackstone. The arrangement underscores how capital-intensive agent and model scaling has become: companies building frontier AI systems now need their chip supplier to help pay for the silicon.
Australian networking company Megaport is pivoting from connectivity plumbing to AI compute, raising A$827.3M to fund a globally distributed inference cloud spanning 1,100 data centers in 31 countries. The bet: inference workloads reward proximity, not centralization.
Nvidia CEO Jensen Huang confirmed at Computex 2026 that Anthropic, OpenAI, and SpaceX are the first major customers for Vera, Nvidia's first standalone data center CPU. The chip enters full production in Q3 2026 and competes directly with Intel Xeon and AMD Epyc. Huang claimed Vera is 1.8x faster than x86 chips on AI-related workloads.
DriveNets closed a $410M Series D at an $8.5 billion valuation, bringing total capital raised to $1 billion. AMD joined as a strategic investor alongside Bessemer Venture Partners, Atreides Management, Pitango, and D1 Capital Partners. The company says it is cash-flow positive with over $1 billion in orders and project backlog.
OpenClaw's ClawHub marketplace now runs every published skill through three independent scanners: static analysis, VirusTotal, and NVIDIA's new SkillSpector. The scanners agree on less than 10.4% of their combined findings, and only 0.69% of skills trigger all three. The full scan dataset, covering 67,453 skill versions, is now public on Hugging Face.
AI chip startup Groq is raising $650 million from existing investors to scale its inference cloud business, pivoting away from chip manufacturing after a $20 billion licensing deal with Nvidia stripped the company of its founder, president, and hardware IP. Backers Disruptive and Infinitium have agreed to fill any gaps in the round.
NVIDIA-backed Nebius Group raised its 2026 contracted power outlook from 3 GW to 4 GW after reporting 684% year-over-year revenue growth. The company doubled capex guidance to $20-25 billion, signaling that AI compute demand, driven by agentic workloads, is outstripping infrastructure supply. Analysts are split on whether the capital intensity will pay off.
NVIDIA unveiled a comprehensive Agent Toolkit at GTC Taipei, positioning itself as the infrastructure layer for enterprise AI agents. NemoClaw provides orchestration blueprints, Nemotron 3 Ultra delivers 5x faster inference for agentic workloads, and OpenShell introduces security primitives for autonomous agent execution. Cadence, Siemens, Synopsys, CrowdStrike, and Palantir are among early adopters.
The first Windows PCs built around Nvidia processors are expected to launch at Computex and Microsoft Build next week. Alongside the hardware, Microsoft is preparing a new local AI agent platform called Autopilot with an OpenClaw-like agent called Scout, consolidating its scattered Copilot tools into a single agentic interface.
Elon Musk posted on X that SpaceX's lease of Colossus compute to Anthropic is a 180-day agreement with 90-day mutual cancellation, and that the short term 'was our request, not Anthropic's.' But SpaceX's own S-1 filing describes Anthropic paying $1.25 billion per month through May 2029. The discrepancy raises questions about the stability of Anthropic's compute access ahead of both companies' IPOs.
NVIDIA released Verified Agent Skills on May 22, a capability governance framework that scans agent skills for prompt injection and software vulnerabilities, signs them cryptographically, and documents them with machine-readable skill cards before they enter enterprise workflows. The system builds on the open SKILL.md format used by Claude Code, Codex, and Cursor.
AMD's Ryzen AI Halo packs a Ryzen AI Max+ 395 processor and 128GB of unified LPDDR5x memory into a Mac mini-sized form factor for $3,999. Pre-orders open in June, with AMD pitching a six-month break-even against cloud AI costs and positioning the box as a complete local-first development platform for agent builders.
Nvidia is in advanced negotiations to lead a $20 million Series A for Simplismart, an AI inference optimization startup, at a valuation of approximately $100 million. The deal would represent a fourfold valuation jump from 2024. Simplismart helps enterprises deploy and optimize AI models in production, improving GPU utilization and cutting inference costs for customers including Tata 1mg and InVideo.
The multi-turn, tool-calling workloads generated by agent harnesses like OpenClaw and Claude Code are driving a structural shift in data center hardware. Intel reports server CPU prices up 10-20% since March as the ratio of CPUs to GPUs in agentic scenarios tightens toward 1:1. Nvidia's $20 billion Groq acquihire, AWS's Cerebras deployments, and Google shipping a 4GB local LLM inside Chrome all point to the same conclusion: the infrastructure built for training is being rebuilt for agents.
Anthropic's deal to take over SpaceX's entire Colossus 1 supercomputer cluster reveals the real bottleneck in the agent economy: not model quality, but inference compute. The company doubled Claude Code rate limits the same day, signaling that agent workloads are consuming GPU capacity faster than Anthropic can build it.
IREN Limited closed a $3 billion convertible senior notes offering, one of the largest capital raises by a publicly traded company pivoting from Bitcoin mining to AI infrastructure. The raise follows a $3.4 billion, five-year AI cloud contract with NVIDIA and a separate strategic partnership targeting 5 gigawatts of AI infrastructure deployment. Shares fell 8% on dilution concerns.
A crypto miner turned $14.6 billion AI infrastructure company in two years. Nscale's $2 billion Series C, the largest in European tech history, anchors a broader neocloud spending surge where CoreWeave reports $99.4 billion in revenue backlog, SpaceXAI leases 220,000 GPUs to Anthropic, and the entire sector is building gigawatt-scale AI factories with five-year contracts and 80%+ EBITDA margins.
Datavault AI sold 109 million shares at $0.55 each to fund deployment of distributed GPU infrastructure across roughly 100 U.S. metropolitan markets. Combined with a separate $120M cash commitment from Scilex Holding Company, the company is marshaling $180M for edge compute buildout. Investors responded by dumping the stock 25.5%.
Beijing-based Global Mofy has deployed OpenClaw's agent framework to automate storyboard generation, script parsing, and multimodal content assembly in its VFX production pipeline. Its U.S. subsidiary Eaglepoint AI was separately selected for the NVIDIA Inception Program, signaling a vertical-specific pattern: production studios are moving AI agents from experiment to infrastructure.
Project Arc puts autonomous AI agents directly on employee desktops with local file system access, terminal control, and application integration. Every action runs inside Nvidia's open-source OpenShell sandbox with policy-based governance from ServiceNow's AI Control Tower. It supports any agent family including OpenClaw and Claude Codex.
NVIDIA's Nemotron Labs blog series details how enterprises should deploy OpenClaw's persistent autonomous agents, positions NemoClaw as the reference stack for secure deployment, and estimates autonomous agents drive 1,000x the inference demand of reasoning AI. The company is also contributing security code directly to the OpenClaw project.
NVIDIA released Nemotron 3 Nano Omni, an open-weight multimodal model that unifies vision, audio, and language in a single architecture. With 30 billion parameters but only 3 billion active per inference pass, it runs on a single GPU while claiming 9x throughput over comparable open models. Companies including Foxconn, Palantir, and Docusign are already adopting or evaluating it.
Adobe's new CX Enterprise Coworker, built on NVIDIA's OpenShell secure runtime and paired with WPP's global marketing expertise, lets autonomous agents generate, adapt, and activate on-brand content across millions of product and audience combinations while operating inside containerized policy guardrails.
Adobe has rebranded Experience Cloud as CX Enterprise, an AI-first platform built around persistent AI agents called Coworkers that orchestrate creative production and marketing campaigns. More than 10 purpose-built agents are now in production across 1,770+ customers. NVIDIA provides the secure runtime and Nemotron models, while WPP brings global media and marketing scale.
Google unveiled TPU 8t and TPU 8i at Cloud Next 2026, splitting its custom silicon into specialized training and inference architectures for the first time. TPU 8t scales to 9,600 chips per superpod with 2 petabytes of shared memory. TPU 8i triples on-chip SRAM to eliminate latency bottlenecks for multi-agent inference. Both connect through Virgo Network for near-linear scaling to one million chips.
NVIDIA's OpenShell secure runtime will enforce policy-based governance for Adobe's CX Enterprise Coworker and WPP's media agents, creating containerized sandboxes that answer 'what can the agent do?' rather than just 'what policy is in place?' The collaboration, demonstrated at Adobe Summit's day-two keynote, connects creative production, customer experience orchestration, and media buying under a single governed architecture.
A four-month-old AI lab with no product and 20 employees just raised $500 million at a $4 billion valuation. Recursive Superintelligence, founded by former Salesforce chief scientist Richard Socher and ex-DeepMind principal scientist Tim Rocktäschel, is building systems that automate the entire AI development pipeline without human intervention.
Morgan Stanley published research projecting that autonomous AI agents will shift computing bottlenecks from GPUs to CPUs, adding $32.5–60 billion to a data-centre CPU market already exceeding $100 billion by 2030. The note names Nvidia, AMD, Intel, Arm, Micron, Samsung, SK hynix, TSMC, and ASML as potential beneficiaries.
Google Chief Scientist Jeff Dean told Bloomberg the company is building next-gen TPU chips specialized for inference, targeting the latency requirements of AI agents. Meta, Anthropic, and Citadel Securities are already using or expanding TPU access. The announcement is expected at Google Cloud Next in Las Vegas this week.
Cadence announced two partnerships at CadenceLIVE: a Google Cloud integration bringing Gemini models to its ChipStack AI Super Agent for automated chip design, and an expanded NVIDIA collaboration around physics-based simulation, digital twins, and robotics. Early deployments report up to 10x productivity gains in design and verification tasks.
NVIDIA dropped the full arXiv paper behind Nemotron 3 Super, its open-weight 120B model with 12B active parameters, 1M context window, and a hybrid Mamba-Transformer MoE backbone built specifically for multi-agent systems. The paper reveals how latent MoE, multi-token prediction, and native NVFP4 pretraining deliver over 5x throughput gains versus the previous Nemotron Super.
Cerebras Systems filed for a NASDAQ IPO on Friday under ticker CBRS, targeting a May 2026 listing at a $22-28 billion valuation. The filing dropped one day after OpenAI committed $20 billion over three years for Cerebras-powered servers, the largest non-Nvidia AI infrastructure contract on record.
TSMC posted a 58% year-over-year profit increase in Q1 2026, beating analyst estimates for the fourth consecutive record quarter. CEO C.C. Wei told analysts the demand driver is specific: the industry shift from 'generative AI and the query mode to agentic AI and command and action mode.' Full-year revenue guidance was raised to above 30% growth.
OpenAI has agreed to pay chip startup Cerebras more than $20 billion over three years for servers powered by its wafer-scale chips, doubling a previous $10 billion commitment. The deal includes warrants for up to a 10% equity stake in Cerebras and $1 billion in data center funding, as OpenAI moves to reduce its reliance on Nvidia.
Humanoid's HMND 01 robot completed autonomous tote-handling tasks inside a Siemens electronics factory in Erlangen, Germany, hitting 60 moves per hour, 8+ hours of uptime, and above 90% pick-and-place success. Announced at Hannover Messe 2026, the deployment uses NVIDIA's full physical AI stack including Jetson Thor, Isaac Sim, and Isaac Lab. It is the most concrete proof of a humanoid robot performing real logistics work in an operational production facility.
SiFive closed an oversubscribed $400M Series G at a $3.65B valuation, with NVIDIA among the investors. The Berkeley-founded company designs open-architecture RISC-V CPUs that plug directly into NVIDIA's NVLink Fusion rack servers and CUDA software stack, positioning open-standard processors as an alternative to Intel x86 and ARM in the AI data centers that run enterprise agent workloads at scale.
Tanium security analyst Melissa Bischoping warns that NVIDIA's NemoClaw, the open-source security stack for OpenClaw agents, is being deployed across Southeast Asia faster than governance frameworks can absorb it. The core risk: autonomous agents operating at machine speed with minimal human checkpoints dramatically increase the blast radius of any single credential misconfiguration.
CoreWeave expanded its existing Meta Platforms agreement to $21 billion in total cloud capacity through December 2032 on April 9, with the deal covering initial deployments of Nvidia's Vera Rubin GPU architecture. Combined with a separate Anthropic deal announced the next day, CoreWeave now serves 9 of the 10 leading AI model providers globally. The company holds $21 billion in debt, just raised $8.5 billion in infrastructure financing, and is raising another $3 billion in convertible notes to fund the Meta expansion.
Nutanix unveiled a multi-tenant management portal and AI governance gateway at its .NEXT conference, enabling neocloud providers to move beyond raw GPU rental into managed AI services including Kubernetes-as-a-Service and agent orchestration. General availability is targeted for H2 2026.
Google has committed to using multiple generations of Intel's Xeon 6 CPUs in its AI data centers for training and inference workloads. The companies also reiterated their ongoing collaboration on custom Infrastructure Processing Units, a partnership dating to 2022. Intel CEO Lip-Bu Tan said scaling AI 'requires more than accelerators.'
Engineers at UC San Diego developed a hybrid chip that combines piezoelectric resonators with capacitors to convert 48V data center power to GPU-level voltages at 96.2% peak efficiency. Published in Nature Communications, the prototype delivers roughly four times more output current than previous piezoelectric designs, addressing a bottleneck in GPU power conversion that wastes significant energy at scale.
Elon Musk confirmed on X that xAI is pausing its $80 million water recycling plant in Memphis to focus on completing the Colossus 2 data center. Without the plant, xAI and TVA will continue drawing billions of gallons of fresh drinking water from the Memphis Aquifer. Separately, CFO Anthony Armstrong departed after just six months, part of wider C-suite turnover reported by The Information.
CoreWeave announced a multi-year agreement to supply Anthropic with Nvidia GPU capacity across US data centers for production-scale Claude workloads. The deal arrives one day after CoreWeave expanded its Meta partnership to $21 billion, and makes Anthropic the latest addition to a customer roster that now includes nine of the ten leading AI model providers.
Chatham Financial estimates AI infrastructure capital expenditure will exceed $830 billion in 2026, roughly half the volume of the entire US investment-grade bond market. Data center capex grew 57% last year to $726 billion. The question nobody wants to answer: if AI is absorbing this much capital, what happens to everyone else trying to raise money?
XDA's Lead Technical Editor spent several days running Nvidia's NemoClaw security sandbox for OpenClaw on a Lenovo ThinkStation PGX. The verdict: NemoClaw's deny-by-default networking and kernel-level isolation are real improvements over bare OpenClaw, but the sandbox cannot stop prompt injection attacks arriving through approved service connections like email, Telegram, and cloud storage. A permissions bug locks the agent out of its own config, the Telegram bridge requires an Nvidia API key despite local inference, and the dashboard became unreachable overnight. The fundamental problem remains: every integration that makes OpenClaw useful is also a vector for attack, and no external sandbox can inspect the semantic content of what the agent processes.
Iran's Islamic Revolutionary Guard Corps has designated 18 American technology companies as military targets across the Middle East, issuing an April 1 deadline for attacks on their regional facilities. The list includes Nvidia, Palantir, Microsoft, Apple, Google, Meta, and others that the IRGC accuses of providing the AI infrastructure enabling US-Israeli precision strikes that killed Iran's supreme leader and dozens of senior military officials. The threat follows a March 1 drone strike on AWS data centers in the UAE and Bahrain that caused widespread regional outages, establishing a precedent for attacks on commercial cloud infrastructure hosting AI workloads.
CNET has published a category-defining explainer framing 'claw' as a distinct computing paradigm, not just a product. The piece documents a growing vendor ecosystem beyond OpenClaw — including NanoClaw and others — and quotes NVIDIA CEO Jensen Huang calling claws 'the new computer' at GTC 2026. The shift from single product to multi-vendor category marks a turning point in how the mainstream tech press covers agentic AI.
French AI startup Mistral has closed an $830 million debt deal backed by seven global banks to fund a 13,800-GPU data center near Paris. The facility, powered by Nvidia GB300 chips, is expected to go live in Q2 2026 and represents Mistral's first-ever debt raise as Europe's largest foundation model company bets on owning its own compute infrastructure.
Nvidia released ProRL Agent, an open-source infrastructure that decouples reinforcement learning training from agent rollout by exposing rollout as a standalone API service. Integrated into NeMo Gym, the system achieved near-linear throughput scaling and boosted a Qwen3-8B model from 9.6% to 18.0% on SWE-Bench Verified — gains from infrastructure alone, not bigger models.
After 35 years of exclusively licensing chip designs, Arm Holdings has broken from its founding business model and launched a physical processor: the AGI CPU, a 136-core data center chip built on TSMC's 3nm process and optimized for agentic AI workloads. Meta co-developed the chip and is the anchor customer, with OpenAI, Cerebras, Cloudflare, and SAP among the additional partners committed to deployment.
At a GTC 2026 panel discussion, Nvidia CEO Jensen Huang proposed a security framework for enterprise AI agents: grant any agent access to sensitive information, code execution, or external communication — but never all three at once. Mistral AI CEO Arthur Mensch warned that OpenClaw's governance and scalability primitives are not enterprise-ready, while LangChain CEO Harrison Chase argued that 'harness engineering' — building guardrails and tool constraints around the core model — is the practical path to safe deployment. The panel also debated whether open or closed models are better suited for enterprise agents, with Allen Institute's Hanna Hajishirzi and healthcare AI CEO Daniel Nadler making the case for open models in privacy-sensitive and specialized domains.
Indian sovereign AI startup Sarvam AI is negotiating a $200–250 million funding round that would value the company at $1.5 billion, with NVIDIA, Accel, and HCLTech as reported investors. The deal would mark the largest private AI funding round in India's history and make Sarvam the country's first AI unicorn of 2026.
Despite record quarterly results, NVIDIA and AMD shed a combined $711 billion in market value within 48 hours of their earnings reports. Developers are more bullish on AI agents than ever. Wall Street just flinched.
Senators Elizabeth Warren and Richard Blumenthal sent a letter to Nvidia questioning whether its $20 billion licensing deal with AI chip startup Groq could stifle competition in the AI chip market. Nvidia pushed back, clarifying it licensed Groq's LPU technology rather than acquiring the company outright. NVDA dropped 1.65% on the day.
U.S. prosecutors charged Supermicro co-founder Yih-Shyan 'Wally' Liaw and two associates with smuggling Nvidia-powered servers to China through a Southeast Asian front company. The alleged scheme generated $2.5 billion in sales since 2024. Supermicro shares fell 25% on the news.
Digitimes reports that OpenClaw's shift to always-on, locally-running AI agents is forcing Chinese chip firms to retool silicon roadmaps for edge inference. With cloud GPU markets locked up by Nvidia and US export controls limiting access to cutting-edge chips, domestic on-device AI silicon represents the opening China's semiconductor industry has been waiting for.
NVIDIA's three-day GTC 2026 conference ended with Disney revealing Kamino, its GPU-accelerated simulator that trained a robotic Olaf to walk on a moving boat using deep reinforcement learning. T-Mobile announced an edge AI partnership for cell tower infrastructure, and Jensen Huang reiterated his $1 trillion revenue forecast through 2027.
Gamers revolted against DLSS 5's AI-rendered faces in Starfield, calling them uncanny and meme-worthy. Jensen Huang fired back that they're 'completely wrong.' Bethesda responded by promising to let artists override it. GTC 2026's only real controversy reveals where AI capability outpaces consumer consent.
Nvidia confirmed at GTC 2026 that its Nemotron model family is being deployed for clinical AI agents running entirely on hospital infrastructure. The on-premises requirement addresses HIPAA and data sovereignty concerns that have blocked healthcare from adopting cloud-based agent systems. Two new model variants — Nemotron 3 Omni for multimodal document extraction and Nemotron 3 VoiceChat for listen-and-respond agents — expand the clinical toolkit.
Nvidia's Agent Toolkit — featuring the OpenShell runtime, Nemotron models, and AI-Q blueprint — now has confirmed integrations with six major enterprise software vendors. Adobe, Salesforce, SAP, ServiceNow, Atlassian, and Box have all signed on, covering creative, CRM, ERP, ITSM, collaboration, and content management categories.
Forbes analyst Janakiram MSV mapped out Nvidia's GTC 2026 announcements as five interlocking platform layers — from the Vera Rubin compute substrate to NemoClaw's agent security framework. The picture that emerges: Nvidia is building the full operating environment for AI agents, not just the chips that run them.
Nvidia launched two new Nemotron 3 open models at GTC 2026: a 4-billion-parameter Nano model for consumer GPUs and a 120-billion-parameter Super model for professional workstations. Combined with optimizations for Qwen 3.5 and Mistral Small 4, Nvidia is building the open-model layer for local AI agents.
Nvidia's NemoClaw stack now runs on RTX PRO 6000 Blackwell workstations with 4,000 TOPS and 96GB GPU memory, enabling enterprises to deploy OpenClaw agents entirely on-premises without cloud dependencies.
Nvidia's CEO escalated his OpenClaw endorsement at GTC 2026, calling the framework 'definitely the next ChatGPT' and demonstrating a live agent that autonomously studied images, learned design tools, and iterated on kitchen layouts without human prompts.
NVIDIA announces Groq 3 LPU for inference and Vera CPU for orchestration, explicitly positioning them as the hardware backbone for AI agents that combine GPU inference with real-world tasks.
NVIDIA assembles a closed enterprise stack. Mistral partners with NVIDIA to provide the open foundation. Together, they're defining what agentic AI infrastructure looks like — and fragmenting it into two incompatible worlds.
Business Insider reports Jensen Huang has shifted his OpenClaw analogy from Linux and GPT to Windows — the platform that someone monetized and controlled. If OpenClaw is Windows, NVIDIA is positioning NemoClaw as the Microsoft of agentic AI, and the revenue implications are fundamentally different.
NVIDIA announced Uber will deploy Drive AV-powered robotaxi fleets across 28 cities on four continents by 2028, with BYD, Hyundai, Nissan, and Geely joining as new OEM partners. Jensen Huang called it 'the ChatGPT moment of self-driving cars' during GTC 2026.
Siemens unveiled the Fuse EDA AI Agent at NVIDIA GTC 2026 — a domain-scoped autonomous AI agent that plans and orchestrates multi-tool workflows across semiconductor design, 3D IC integration, and PCB systems. One of the three dominant EDA vendors just put an AI agent in its most technically demanding product line.
The enterprise infrastructure publication NextPlatform published a thesis piece arguing OpenClaw occupies the same defining role for agentic AI that GPT-3 played for conversational AI. After Jensen Huang's GTC keynote canonized OpenClaw as foundational infrastructure, the comparison raises a specific question: if OpenClaw is the new GPT, who are the winners and who are the dead startups walking?
Abacus AI's DeepAgent shipped SOC 2-certified sandboxed OpenClaw execution with persistent memory and scheduled runs in mid-February. NVIDIA announced NemoClaw at GTC on March 17. The enterprise OpenClaw race started earlier than the headlines suggest.
Nvidia unveiled NemoClaw at GTC 2026, an enterprise-grade security and orchestration stack built on top of OpenClaw. With OpenShell sandboxing, partnerships with CrowdStrike and Cisco, and a new open-source model coalition, Nvidia is positioning itself as the infrastructure layer between autonomous AI agents and the corporate firewall.
NVIDIA's 2026 enterprise AI report confirms what many suspected: the industry has crossed from proof-of-concept into profit-and-loss, with nearly two-thirds of organizations running live deployments.