AI Lab

Pentagon unveils top performers in latest ‘Gauntlet’ for Drone Dominance Program

Pentagon plans to purchase approximately 60,000 drones from UAS vendors.

Cisco Blogs: “Partner Study Shows a Maturing AI Market, Expanding Opportunity for Partners”

Cisco Blogs reports the 2026 Global AI Partner Study surveyed 1,702 partners navigating the rise of agentic AI.

Qwen-Image-2.1: Compact, Efficient, and Unified Image Creation

Qwen Research opensources Qwen-Image-2.1, a compact, efficient, and unified image model balancing quality and cost.

Industry’s first MLPerf™ Inference benchmarking across multi-vendor accelerators

Cisco submitted the industry's first MLPerf Inference benchmark across multi-vendor GPU accelerators, optimizing data center unit economics without workload penalties.

IAC completes autonomous racing event at Laguna Seca

Unimore Racing won the Indy Autonomous Challenge at Laguna Seca, marking the IAC's first autonomous car overtakes on a road course.

What do you do with a humanoid robot when it breaks down?

Decommissioning a humanoid robot requires a highly technical, high-stakes surgical approach rather than traditional scrapping.

Introducing Amazon SageMaker HyperPod Inference Gateway

Amazon SageMaker HyperPod Inference Gateway is a Kubernetes-native, GPU-aware routing add-on for Amazon EKS.

A shared agentic platform for Wood Mackenzie, on Amazon Bedrock AgentCore

Wood Mackenzie built APEX, a shared agentic AI platform on Amazon Bedrock AgentCore to enable teams to ship production agents without rebuilding runtime, identity, observability, and guardrails from scratch.

Amazon SageMaker Inference: 2026 year-to-date launches in review

Amazon SageMaker AI shipped 13 inference launches year-to-date across fully managed endpoints and HyperPod Inference.

F-Prime Capital to give insights on the state of robotics investments at RoboBusiness

F-Prime Capital's Sanjay Aggarwal and Betsy Mulé will unveil the State of Robotics Investments report at RoboBusiness.

Benchmarking LLM Inference at Scale with AIPerf

NVIDIA Developer Blog reports that model deployment begins with prompt responses on the target system.

Qwen3.8-Omni-Flash: Omni Senses. Agentic Delivery.

Qwen Research launches Qwen3.8-Omni-Flash, a next-generation native omnimodal model.

How to Use AI Agents to Prepare 3D Scenes for Simulation

Agentic AI workflows prepare and validate digital twins for physical AI systems.

Hirebotics adds line tracking and linear rail capabilities to its cobots

Hirebotics adds line tracking and linear rail capabilities to its cobots, extending working range and giving manufacturers greater flexibility.

The picks-and-shovels strategy behind the robotics boom

The real robotics opportunity isn’t a humanoid in every home—it’s packaging robots for the jobs businesses need done today.

Introducing Kimi K3 on Amazon Bedrock

Kimi K3 from Moonshot AI is now available on Amazon Bedrock as an open-weight model for coding and knowledge work.

Migrating multi-model AI agents to Amazon Bedrock AgentCore runtime

AWS migrates multi-model healthcare AI agents from self-managed ECS to Bedrock AgentCore, preserving triple-model orchestration and vector-enhanced knowledge retrieval while reducing infrastructure management.

Robotic lab sets up and runs optics experiments on demand

A robotic lab autonomously sets up and runs optics experiments on demand could speed up testing of high-tech materials.

Skild AI Taps NVIDIA Physical AI to Teach Robots New Tasks From a Single Video

Skild AI uses NVIDIA Physical AI to enable robots to learn new tasks from a single video, adapting to changing manufacturing floors and production lines without reprogramming.

Reimagining robotics for sustainability

Robots are often promoted for their increasing intelligence and autonomy, according to EPFL Research News.

Cloudflare Blog: “When scanners miss the attack: how Cloudflare Client-Side Security protects storefronts”

See how Cloudflare's machine learning models surface evasive client-side attacks for analyst investigation from malicious JavaScript that silently siphons revenue, hijacks clicks, or rewrites analytics.

Build a serverless PII redaction pipeline with Amazon Bedrock Data Automation

AWS Bedrock Data Automation automates end-to-end PII detection and redaction from scanned documents at scale using a custom blueprint, Step Functions, and Lambda.

Glyph: A Multi-Strategy Agentic System for Column Description and Sensitivity-Ontology Tagging of Enterprise Data Catalogs

Enterprise data lakes accumulate tables faster than human stewards can document or classify them, leaving columns with missing descriptions and unassigned governance labels.

$3.5M to advance autonomous trauma robotics

DARPA awards $3.5 million to advance autonomous trauma robotics for mass casualty events.

New AI technique could make minimally invasive surgeries safer and more precise

MIT News reports xvr is a patient-specific AI method enabling X-ray use for surgical navigation in orthopedics and neurosurgery.

Translating CUDA Tile Operations from Python to Rust Using Agentic AI

cuTile Rust (cutile-rs) enables safe, idiomatic GPU kernel authoring in Rust, translating CUDA tile operations from Python.

Abnormal AI: Amazon Bedrock AgentCore for agentic email security at scale

Abnormal AI deployed Amazon Bedrock AgentCore Code Interpreter as an ephemeral compute scratch pad for billion-message email threat detection, detailing sandbox design decisions and practical lessons for builders.

Shared Selective Persistent Memory for Agentic LLM Systems

Apple researchers identify that agentic LLM systems lose productive context across sessions, requiring persistent memory to retain configuration and tool-use patterns.

NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut

NVIDIA Blog reports system performance, efficient infrastructure scaling and continuous software optimization are key levers determining AI inference economics.

How NVIDIA Groq 3 LPX Deterministic Execution Drives Power-Efficient High-Interactivity Inference on NVIDIA Vera Rubin

NVIDIA Developer Blog states power is a defining constraint for AI factories.

Build an AI-powered product tagging system with Amazon SageMaker serverless model customization

Amazon SageMaker serverless model customization enables cost-efficient product tagging by fine-tuning Qwen3-8B with supervised and reinforcement learning on catalog data.

Introducing ChatGPT for Financial Services

OpenAI introduces ChatGPT for Financial Services, combining built-in financial data and GPT-6 Astra for research, modeling, and client-ready materials.

MIT Schwarzman College of Computing launches pilot to help educators teach AI across disciplines

MIT hosted a weeklong summer workshop for higher education faculty to explore adapting AI and machine learning materials for classroom use.

AI takes root in university curriculum

More than 12,000 students have returned to classes at EPFL. How has the advance of artificial intelligence transformed higher education?

New method enables AI for safety-critical situations

MIT News reports the The “HardFlow” algorithm could help generative AI models produce high-quality outputs that obey strict requirements for safety-critical situations.

AMD Newsroom: “F-Secure and AMD Silo AI Help Secure Agentic AI Journeys”

F-Secure and AMD Silo AI explore adaptive AI routing across local, private and cloud infrastructure to help keep user data secure as AI agents work.

Powering the AI era: How wave energy can complement a 24/7 energy mix

CorPower Ocean, a Cisco Investments portfolio company, has explored how wave energy technology could complement other sources for 24/7 energy needs.

High-Throughput Structure Prediction with BioNeMo Inference Runtime

BioNeMo Inference Runtime enables proteome-scale biomolecular structure prediction for efficient worklist processing.

IBM and NASA Release Open-Source AI Model to Support Lunar Exploration

IBM and NASA release open-source Lunar Foundation Model to surface patterns across decades of lunar observations.

NASA, IBM Launch AI Foundation Model for Lunar Science

NASA and IBM launch AI foundation model to transform lunar surface analysis.

Satellite imagery gives experts a better understanding of conflict

Satellite data supports humanitarian conflict analysis to reduce civilian war toll.

Govern models with MLflow and Amazon SageMaker AI Model Registry sync: Part 2

AWS blog states governing models across accounts follows automatic model registration.

Govern models with MLflow and Amazon SageMaker AI Model Registry sync: Part 1

Managed MLflow on Amazon SageMaker AI now syncs richer model metadata, including training metrics, evaluation results, inference specs, and lineage, into the SageMaker AI Model Registry with lifecycle stage promotion.

Build a Physical AI model factory with NVIDIA Cosmos 3 on SageMaker HyperPod

Building a Physical AI system requires a continuous pipeline, not a single training job.

Reduce inference cold starts on Amazon SageMaker HyperPod with model caching

Amazon SageMaker HyperPod now supports model caching for inference, pre-loading model weights and container images onto cluster nodes for local NVMe storage access.

Evolving With Agentic Risk: Updating Our Integrated AI Security & Safety Framework

Cisco introduced nine months ago a unified taxonomy to identify and mitigate AI security and safety risks.

Simplify and support your TorchServe workloads using Ray Serve Deep Learning Containers

TorchServe is no longer maintained, requiring teams to own the entire GPU inference stack.

Physical AI Takes the Wheel: How the World’s Robotaxi Leaders Are Building With NVIDIA Technologies

Robotaxi market projected to reach $400 billion by 2035 with over 6 million commercial vehicles in operation.

Monitoring production agent lifecycle with AWS DevOps Agent and AgentCore Evaluations

AWS presents a dual-layer monitoring approach for production agents using Bedrock AgentCore Evaluations and DevOps Agent, demonstrated on a four-agent airline reservation system.

Beyond the price per token: Choosing the right OpenAI model on Amazon Bedrock for your workload

Choosing the right OpenAI model on Amazon Bedrock prioritizes outcomes over price per million tokens.

Reduce LLM latency with prefix-aware routing on Amazon SageMaker Inference

Amazon SageMaker Inference introduces prefix-aware routing to keep KV cache warm for shared prompt prefixes.

When to Use Encode-Prefill-Decode Disaggregation to Accelerate Multimodal Model Serving

NVIDIA Developer Blog covers the topic: "When to Use Encode-Prefill-Decode Disaggregation to Accelerate Multimodal Model Serving".

ICYMI: What landed for AI builders in August 2026

AWS launched million-token context for OpenAI models, cross-Region inference, 14-day agent compute, and expanded GovCloud availability in August 2026.

d-Matrix Adopts NVIDIA NVLink Fusion for Rack-Scale XPU Deployment

d-Matrix announced it will use NVIDIA NVLink Fusion to connect its next-generation Raptor XPUs to NVIDIA's AI infrastructure platform.

MVICAD2: Multi-View Independent Component Analysis with Delays and Dilations

Machine learning techniques in multi-view settings face challenges integrating heterogeneous data, aligning feature spaces, and managing view-specific biases, particularly in neuroscience applications.

The yield imperative: Turning AI infrastructure into useful intelligence

Microsoft asks what defines progress in the next era, naming yield as the semiconductor industry's defining measure and questioning how elegance, time and infrastructure shape useful intelligence.

AWS Weekly Roundup: Welcome DuckLabs to the team, Agentic Resource Discovery (ARD), and more (August 31, 2026)

AWS has signed a definitive agreement to acquire DuckLabs, the Amsterdam-based company behind DuckDB, the open source analytical database that executes SQL directly against files like Parquet, CSV and JSON while remaining under its independent foundation.

AWS announces GPT-6 Astra availability on Amazon Bedrock

AWS presents the OpenAI model as generally available through its Bedrock inference platform.

Cisco shares a first-time analyst’s account of the Black Hat SOC

The account connects agentic workflows and technical tools with human mentorship during security investigations.

MIT presents a way to explain autonomous-driving AI decisions

CW-Net translates an autonomous vehicle’s AI reasoning into concepts people can understand.

NVIDIA presents local AI tools and RTX Spark PCs at IFA

NVIDIA, Microsoft and partners describe tools for running agents locally, with compact RTX Spark Windows PCs announced for October.

Cloudflare links robots.txt to AI bot preferences

Bot Preference Sync aligns the file with policies for search, agent and training bots.

Cisco outlines its Security Cloud role at Black Hat

Cisco presents its Security Cloud contribution to the Black Hat conferences and its work on agentic security operations.

AWS connects Outlook workflows with Amazon Quick

The account covers email, calendar and workflow automation using Amazon Quick's agents, Flows and Automate.

EPFL and NYU study robotic swimmers in different sizes

The research describes fish-inspired robots intended for aquatic environments ranging from shallow creeks to open water.

Qwen-Drive-1.0 connects vision and language for autonomous driving

The project combines 3D perception and visual question answering during pretraining, with an extension to motion planning.

AWS connects HyperPod operations to an agent-driven control plane

AWS describes HyperPod InstantStart as an open-source control plane combining Amazon EKS orchestration with SageMaker HyperPod capabilities.

Cisco describes ThousandEyes' shift towards continuous promotion

Cisco reports that ThousandEyes moved from fragmented manual release cycles towards a centralised, GitOps-driven deployment model.

AMD reports progress towards its AI energy-efficiency target

AMD reports an estimated fourfold improvement in AI energy efficiency between 2024 and 2026.

Google outlines AI tools for advertising and analytics workflows

Google describes new AI and agentic experiences across Google Ads and Google Analytics.

Frontier Reasoning Reaches the Edge: How to Deploy and Optimize Models on NVIDIA Jetson

Running reasoning and agentic AI at the edge has been harder than it needs to be. Until recently, models capable of multi-step reasoning were too large to run.

Introducing WeatherNext 3, our most advanced and accurate global weather AI model

DeepMind announces its most advanced and accurate global weather AI model to date.

Introducing Adaptive Intelligence: undermining the economics of every bot attack

Bot operators have historically had the economic advantage, bypassing static, deterministic detection rules with cheap proxies and retooling. Cloudflare's new Adaptive Intelligence engine flips this dynamic by autonomously learning from the meta-signals of live traffic and deploying disposable rules, making automated

Migrate agentic workloads to Amazon Bedrock AgentCore

An agent that works in a notebook is not an agent in production. This post walks through migrating a LangGraph customer support agent to Amazon Bedrock AgentCore in two stages: onto Runtime, Gateway, and Memory, then to model-driven planning on Strands Agents, retiring operational burdens along the way.

NVIDIA PAIR Virtual Inference Router Expands Available Compute on Your Local Network

A lead agent can break a complex task into smaller jobs and assign those jobs to specialized subagents, learning to do more by working together.

Accessing OpenAI models on Amazon Bedrock from Australia with global cross-Region inference

Australian teams can now access OpenAI GPT-5.6 Sol, Terra, and Luna models on Amazon Bedrock with global cross-Region inference from the Asia Pacific (Sydney) and Asia Pacific (Melbourne) Regions. This post shows how to invoke the models, use prompt caching, set up Codex with OpenID Connect authentication, and monitor

Co-Designing AI Models Using Speculative Decoding for Faster LLM Inference

This post is the third in a series on AI model co-design. It explores how to accelerate LLM inference while maintaining accuracy using speculative decoding and...

NVIDIA Nemotron powers adaptive agentic cybersecurity system

Agentic systems coordinate work and pursue complex objectives over long horizons; security teams are beginning to deploy adaptive defences.

t54 builds x402-secure trust layer for Amazon Bedrock AgentCore payments

Session budgets, credential isolation and deterministic trust gate govern over 20 million agent-initiated transactions with no human in the loop.

From theory to delivery: How Atos upskilled 400 engineers in agentic AI

When Atos set out to upskill 400 engineers in agentic AI, hands-on learning was the missing ingredient. Over three days, engineers built multi-agent systems on AWS through an AI League event. This post explains why Atos chose the format, what engineers built and learned, and what other enterprises should consider.

MIT engineers create system for building shape-changing smart devices

Bifur-circuits interactive building blocks could be used to develop reconfigurable robotic grippers or customized assistive devices.

Build observable enterprise agentic retrieval using Managed Amazon Bedrock Knowledge Base with AWS CloudFormation

This post builds an enterprise agentic retrieval solution on the Amazon Bedrock Managed Knowledge Base and Amazon Bedrock AgentCore. An agent reasons, routes across multiple knowledge bases, and returns cited answers, with seven layers of observability and both on-demand and continuous evaluation, all deployed with a s

Agentic AI is changing how research is done

AI scientists can read papers, propose hypotheses, call models, and determine which experiments to prioritize next, as described in the NVIDIA Developer Blog overview of BioNeMo NIM Microservices.

Building the systems behind the next generation of intelligent robots

As robots expand capabilities and intelligence, the next challenge is integrating AI inference, sensing, control, safety, and orchestration into systems that can operate reliably at scale in the real world.

IDEA Prune: An Integrated Enlarge-and-Prune Pipeline in Generative Language Model Pretraining

Recent advancements in large language models have intensified the need for efficient and deployable models within limited inference budgets. Structured pruning pipelines have shown promise in token efficiency compared to training target-size models from scratch.

Build multi-tenant agentic chat applications on enterprise data with Amazon Bedrock Managed Knowledge Base

Learn how to build a multi-tenant agentic document chat application on Amazon Bedrock Managed Knowledge Base, where users upload documents and immediately ask grounded questions. This post covers the ingestion and retrieval flows, the asynchronous indexing lifecycle, per-user data isolation, and best practices for oper

Deploy an Open Model from Checkpoint to Inference in Two Commands with NVIDIA TensorRT Model Connect

Open AI models are evolving faster than ever, but bringing them into native applications can still require model-specific conversion, preprocessing,...

How Decathlon runs demand forecasting at scale with Chronos-2

Decathlon, one of the world's largest sporting goods retailers, forecasts weekly demand for tens of thousands of products across multiple continents. Learn how they deployed Chronos-2 on AWS to improve forecast accuracy by 11-15 points while cutting operational complexity and running weekly inference for about $0.03.

With Groq 3 LPX in Full Production, NVIDIA Extends Vera Rubin Inference for Agents

The next era of AI inference won’t be defined by a single breakthrough chip, network or system. It’ll be defined by how every layer of the AI factory works together. That’s why NVIDIA is extending Vera Rubin NVL72 with fast token generation for agentic systems.

Intel Outlines Architectures for Agentic AI at Hot Chips 2026

Intel is showcasing three architectures for Agentic AI and Enterprise-scale workloads, combining the Diamond Rapids processor for high-performance orchestration, the Crescent Island GPU for efficient inference and the Wildcat Lake SoC for intelligent client and edge computing.

Spreading the load: How Salesforce met Multi-AZ HA with SageMaker Inference Components

Learn how Salesforce used Amazon SageMaker AI Inference Component placement (the SchedulingConfig parameter) to distribute model copies across multiple Availability Zones, meeting their Multi-AZ high availability compliance requirements without sacrificing the cost efficiency of multi-model co-hosting.

From Isolated Agents to Collective Intelligence: Why A2A Is the Protocol the Agentic SOC Has Been Waiting For

The Agentic SOC needs two protocols, not one. Learn how MCP and A2A work together as the vertical and horizontal layers of multi-agent security operations

OpenAI discontinues model provision to Cursor following SpaceX acquisition

OpenAI announces it is winding down its contract providing OpenAI models to Cursor following its acquisition by SpaceX.

The Hugging Face incident and the road ahead

OpenAI shares findings from the Hugging Face security incident and the steps being taken to strengthen AI model security, monitoring, and alignment.

Beyond Visual CoT: Internalized Visual Thinking for Proactive Video Reasoning

Multimodal large language models increasingly use visual chain-of-thought (Visual CoT) to reason about spatial, temporal, and embodied environments, with internalized visual thinking emerging as a proactive approach for video reasoning.

NVIDIA Vera Rubin NVL72 Sets New Efficiency Standard for AI Agents

OpenRouter data shows agentic AI workloads consume 15x more tokens than simple chat requests, driven by multi-step research and sub-agent coordination.

Alibaba Qwen3.8-Flash-Next Preview and Agentic Coding on NVIDIA GB300 NVL72

Developers can experiment with Qwen3.8-Flash-Next 176B model on NVIDIA GB300 NVL72, previewing the upcoming Qwen4 architecture for agentic coding tasks.

Build agentic creative workflows with Amazon Quick and fal

Creative teams produce more assets than ever, but fragmented tools and manual context transfer slow production. This post shows how to build a reusable agent harness with Amazon Quick and fal, connected through the Model Context Protocol (MCP), using two hands-on workflows: an eight-panel storyboard and a music-video c

Amazon Bedrock adds India cross-Region inference for OpenAI GPT-5.6 models

New capability allows scale processing of Terra and Luna models while keeping data within India for local compliance requirements.

Amazon Bedrock Expands OpenAI GPT-5.6 Model Access in India

New geographic cross-Region inference capability allows local data processing requirements to be met while keeping inference requests and data within India.

Alibaba Qwen3.8-Flash-Next Preview Released on NVIDIA GB300 NVL72

Model weights released as preview of upcoming Qwen4 architecture for developers to experiment with and evaluate agentic coding capabilities.

Agentic observability with Amazon OpenSearch Service MCP Apps

Amazon OpenSearch Service now supports MCP Apps, which return interactive visualizations alongside your AI agent's text responses. Learn how a single, locally run MCP server lets your agent move from alert to trace to logs to root cause in one conversation, and how you can verify every step inline without leaving your

Why Agentic AI is a Systems Challenge

As AI agents move from generating answers to autonomously achieving increasingly complex goals, performance depends on how effectively compute, memory, software, networking, security, and power management work together as a system.

Hugging Face Inference Endpoints, Jobs, and Buckets Power Search on Papers with Code

Hugging Face Inference Endpoints, Jobs, and Buckets provide the infrastructure for searching papers with code, enabling researchers to find and experiment with models efficiently.

AWS SageMaker HyperPod adds managed Ray support on Amazon EKS

AWS Machine Learning Blog announces new Ray capabilities on SageMaker HyperPod, enabling managed Ray clusters on Amazon EKS with out-of-the-box observability and resilient distributed training and accelerated inference.

Up to 3.2x Faster Inference with LFM2.5-DSpark

OpenAI unveils Jalapeño inference chip for AI workloads

OpenAI News reports Jalapeño is a custom inference chip delivering faster, more power-efficient AI inference with higher throughput and lower latency for modern models.

Restore LLM Inference Capacity in Seconds with Shadow Engine Recovery in NVIDIA Dynamo

When an LLM engine process fails, the standard recovery path involves a cold restart. This requires loading weights into HBM from storage, compiling kernels,...

Six in 10 Leaders Bet Big on Robots. Only Four in 10 Are Ready.

Intel Newsroom reports that 60% of senior business and IT leaders, robotics specialists, government and healthcare officials expect their organizations to operate robot fleets within five years, with leaders predicting full-scale robotics deployment could double operational output.

Cross-lingual Knowledge Transfer under Data Constraints via Lexical Interventions

Apple Machine Learning Research explores methods for transferring knowledge across languages when target language data is scarce, focusing on scientific reasoning, commonsense inference, and world knowledge.

NVIDIA Vera Rubin and Blackwell Set a New Standard for Agentic AI Performance per Watt

NVIDIA Developer Blog outlines advancements in AI agent hardware and performance efficiency for multi-step workflows.

AWS Agent Registry Introduces Open Agentic Resource Discovery Standard

Centralized catalog for agents, tools, and skills enables cross-environment discovery and governance at scale.