AWS launches llm-d disaggregated inference on SageMaker HyperPod and EKS, delivering up to 70% higher LLM throughput by splitting compute-heavy prefill and memory-bound decode phases across distributed GPUs using an open-source Kubernetes-native framework built on vLLM.
Mistral unveils 'Forge' at Nvidia GTC, a groundbreaking platform letting enterprises and governments train custom AI models from scratch on proprietary data, as the company races toward $1 billion in annual recurring revenue with partners including Ericsson and the European Space Agency.
Google Labs launches Stitch SDK, an open-source TypeScript library that lets developers instantly generate UI screens from simple text prompts, complete with HTML extraction, screenshot support, and seamless AI framework integration.
NVIDIA launches OpenShell, a sandboxed runtime securing autonomous AI agents through four-layer defense-in-depth protection, blocking unauthorized file access, data exfiltration, and uncontrolled network activity via declarative YAML policies with GPU passthrough support and integration with top AI agents like Claude and Codex.
Amazon and NVIDIA are joining forces to deliver a powerful multimodal AI assistant for vehicles by 2027, combining Alexa Custom Assistant with NVIDIA's DRIVE AGX platform to enable natural conversation, ambient awareness, and seamless cloud-connected features like smart home control and music streaming directly from the car.
OpenAI is nearing a $10 billion joint venture with major private equity firms TPG, Bain, Brookfield, and Advent, offering equity stakes, board seats, and early enterprise tool access in exchange for $4 billion as it aggressively scales its enterprise AI deployment strategy.
Meta commits up to $27 billion to cloud provider Nebius Group over five years, securing massive AI infrastructure capacity as it races to compete in the frontier AI model race.
NVIDIA launches NemoClaw, a one-command AI agent software stack that combines Nemotron models with the new OpenShell runtime, delivering built-in security, privacy guardrails, and continuous skill development across a wide range of NVIDIA hardware.
IndexCache slashes DeepSeek Sparse Attention compute by 75% by reusing token indices across transformer layers, delivering up to 1.82× faster AI inference on H100 GPUs with negligible quality loss and zero extra memory overhead, now available as a patch for SGLang and vLLM.
A new terminal dashboard called claudetop is revolutionizing how developers monitor Claude Code sessions, delivering real-time cost tracking, token usage, and smart budget alerts directly in the status line to prevent surprise API bills.