Reports
AI-generated structured vendor updates
NVIDIA Open-Sources Cosmos 3 Edge 4B World Model for Real-Time Robot Control at 15Hz on Jetson Thor
NVIDIA open-sources Cosmos 3 Edge, a 4B parameter world action model for edge robotics. It achieves 15Hz real-time inference with 32 actions per inference on the Jetson Thor module. This extends NVIDIA's physical AI stack from training to real-time deployment, enabling end-to-end robot control at the edge.
Microsoft's Project Perception Automates Vulnerability Remediation with Multi-Model AI Orchestration
In response to competitors like Anthropic and Palo Alto, Microsoft's Project Perception leverages multi-model AI orchestration to automate vulnerability discovery and remediation. This product signifies a transition from manual security operations to autonomous self-healing systems, with control shifting from security analysts to an AI-driven platform.
NVIDIA Expands Agent Toolkit with Omniverse Libraries for Physical AI Simulation
At SIGGRAPH 2026, NVIDIA announced an expansion to its Agent Toolkit, adding Omniverse libraries that enable AI agents to build and simulate 3D worlds. The company also open-sourced Cosmos 3 Edge, a 4B-parameter world action model, completing its physical AI ecosystem from training to edge deployment.
Apple Engages PrismML for 1-bit Quantization, Enabling 15x Memory Reduction for On-Device AI
Apple is evaluating PrismML's native 1-bit model compression technology, reducing model size to 1/14, memory usage by 90%, and boosting inference speed by 8x. The Bonsai 27B model can run on iPhone 15, marking a breakthrough in on-device AI that could reshape the mobile AI landscape.
NVIDIA Vera Rubin at BMS: Mission Control and BioNeMo Shift the AI Factory Control Plane
BMS deploys NVIDIA DGX SuperPOD with Vera CPU and Rubin GPU, managed by Mission Control and leveraging BioNeMo Agent Toolkit for drug discovery. This signals NVIDIA's shift from hardware vendor to AI factory control plane provider, potentially locking enterprises into its ecosystem.
Google's Frozen v2 Chip Hardwires Gemini Architecture for 6-10x TPU Efficiency, Set for 2028
Google is developing Frozen v2, a dedicated AI chip that hardwires the Gemini model architecture into silicon for 6-10x energy efficiency per token over current TPUs. It is a new product line, planned for 2028, with weight update flexibility but a frozen architecture. This validates the industry shift from general-purpose GPUs to dedicated ASICs for AI inference.
Microsoft Azure Deploys AMD Helios Rack with MI455X GPUs, Breaking NVIDIA's Cloud AI Monopoly
Microsoft Azure officially adopts AMD Helios rack-scale AI infrastructure, featuring 72 MI455X GPUs (432GB HBM4, 19.6TB/s), Venice EPYC CPUs, and Pensando DPUs. Three new instances (ND MI455X v7, HDv2, HXv2) are launched, marking Azure's shift from exclusive NVIDIA dependency to a multi-vendor AI strategy.
NVIDIA Invests $2B in CoreWeave, Debuts 'Compute Central Bank' Model
NVIDIA invests $2 billion in CoreWeave and launches the AI Compute Partner Program, featuring credit enhancement, revenue sharing, and GPU buyback. This transforms NVIDIA from a hardware vendor into a 'compute central bank', tightening control over the AI cloud leasing ecosystem and squeezing intermediaries.
Beyond the Model: Harnessing Frontier AI for Stronger Defense
...
Intel Foundry 18A Yields Jump to 85%+; EMIB Packaging Hits 98%, Challenging TSMC N2
Intel Foundry 18A yields surged from 65% to 85%+ in a single quarter, approaching TSMC N2's 90%. EMIB advanced packaging yields reached 90-98%, turning a former bottleneck into a selling point. NVIDIA, AMD, Apple signed on but mostly as secondary suppliers.
Meta Launches Muse Spark 1.1 API at 25% Competitor Price, Ends Open-Source Era
Meta releases Muse Spark 1.1, a multimodal reasoning model with 1M token context window, and launches its first paid API at 25% of competitors' price. This ends the Llama open-source era, signaling a strategic shift to proprietary API monetization and aggressive market share capture.
Anthropic Extends Claude Cowork Unified Interface to Web and Mobile, Compliance Gap Looms
Anthropic launches Claude Cowork unified interface on Web and Mobile for Max users, merging chat and task execution with local file access and cross-device continuity. However, Cowork activities are not captured in audit logs or Compliance API, creating a significant governance gap vs. Microsoft Copilot.
Microsoft July Patch Tuesday Hits Record 622 CVEs, AI Infrastructure Vulnerabilities Emerge as New Attack Surface
Microsoft's July 2026 Patch Tuesday addresses a record 622 CVEs, including three critical AI vulnerabilities: Copilot RCE (CVSS 9.6), Azure OpenAI EoP (CVSS 9.9), and M365 Copilot EoP (CVSS 9.3). The attack surface expands from OS to AI service infrastructure, signaling an era of AI-driven vulnerability inflation.
Google DeepMind AlphaEvolve GA: AI Self-Evolution for Data Center and Algorithm Optimization
On July 19, 2026, Google DeepMind announced the GA of AlphaEvolve, a Gemini-based multi-agent evolution system for algorithmic discovery, mathematical discovery, and data center efficiency optimization, already used in Borg and Orca, aiming to reduce Capex in massive AI compute investments.
Moonshot AI Launches Kimi K3: 2.8T Parameter Open-Source MoE Model at $3/$15
Moonshot AI unveils Kimi K3, a 2.8T parameter open-source MoE model with 896 experts (16 active), native vision, and 1M context. API pricing at $3/$15 input/output per million tokens undercuts rivals. Open weights release July 27. GPU capacity exhausted within 2 days. Arena score 1679 tops Fable 5.
HPE Expands Private Cloud AI with NVIDIA Vera Rubin, Enabling Agent-Native AI Factory
HPE expands its Private Cloud AI line with NVIDIA Vera Rubin NVL72 and HGX Rubin NVL8, introducing Compute XD700, Cray GX240 blade with Vera CPU, and Quantum-X800 InfiniBand. New software includes Agent Toolkit and NemoClaw for agent-native AI, with Alletra Storage MP X10000 in Q4 2026.
NVIDIA Agent Toolkit Shifts AI Agent Control from Cloud to Local DGX Station
NVIDIA launches Agent Toolkit for DGX Station, comprising NemoClaw, Nemotron 3 Ultra, Omniverse Libraries, and OpenShell. It enables local AI agent deployment in 30 minutes, locking developers into NVIDIA's hardware-software stack and shifting control from cloud services to on-premises hardware.
测试来源：36氪
...
Huawei Ascend 950 Supernode: Self-Developed HCCS Interconnect for Sovereign AI Compute Ecosystem
Huawei unveiled the Ascend 950 Supernode at WAIC, integrating 32 self-developed Ascend 950 AI processors with HCCS high-speed interconnect, achieving 2.5x compute density and supporting trillion-parameter model training, offering a sovereign AI compute alternative free from overseas supply chains.
Cisco Deploys AgenticOps Autonomous Network, AI Agents Take Over Operations Control
Cisco deploys AgenticOps autonomous network architecture globally, using an AI-native engine for self-healing. Over 90,000 employees use AI agents, reducing MTTR from hours to minutes, targeting 40% opex reduction. A third-party agent development framework is also launched.