{"slug": "modal-says-nvidia-s-vera-cpu-doubles-sandbox-throughput-under-load", "title": "Modal says NVIDIA's Vera CPU doubles sandbox throughput under load", "summary": "Modal is testing NVIDIA's Vera Arm CPU for its burst-oriented sandbox fleet and reports the chip doubles sandbox throughput under load while keeping task latency similar, according to a Modal Newsroom post. Modal, led by founder and CEO Erik Bernhardsson, is weighing whether Vera can reduce the CPU contention that slows agents and leaves reinforcement-learning GPUs idle, and says the result could improve utilization if it holds in production.", "body_md": "# Modal says NVIDIA's Vera CPU doubles sandbox throughput under load\n\n**Modal is testing an Arm server for its burst-oriented sandbox fleet, where CPU contention can slow agents and leave training GPUs idle.**\n\n        By [Ryan Merket](https://runtimewire.com/author/ryan-merket)\n        · Published \n\nPrimary source: [Modal Newsroom](https://modal.com/blog/nvidia-vera-sandboxes)\n\n## Why it matters\n\nModal is testing whether NVIDIA's Arm CPU can let it run more agent sandboxes per host while keeping task latency similar. If the result holds in production, it could improve utilization and reduce the CPU bottlenecks that slow agents and keep reinforcement-learning GPUs waiting.\n\n[Modal](https://modal.com/?ref=runtimewire), whose founder and CEO is [Erik Bernhardsson (@bernhardsson)](https://x.com/bernhardsson?ref=runtimewire), is weighing whether [NVIDIA's Vera CPU](https://nvidianews.nvidia.com/news/nvidia-unveils-vera-the-cpu-for-agents?ref=runtimewire) can ease a problem AI workloads make unusually visible: the computer running an agent can hold up the rest of the system.…\n\nWhen you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.", "url": "https://wpnews.pro/news/modal-says-nvidia-s-vera-cpu-doubles-sandbox-throughput-under-load", "canonical_source": "https://runtimewire.com/article/modal-vera-cpu-sandbox-throughput", "published_at": "2026-10-19 15:47:00+00:00", "updated_at": "2026-10-19 18:47:36.130436+00:00", "lang": "en", "topics": ["ai-infrastructure", "ai-chips", "ai-agents"], "entities": ["Modal", "NVIDIA", "Vera", "Erik Bernhardsson"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/modal-says-nvidia-s-vera-cpu-doubles-sandbox-throughput-under-load", "markdown": "https://wpnews.pro/news/modal-says-nvidia-s-vera-cpu-doubles-sandbox-throughput-under-load.md", "text": "https://wpnews.pro/news/modal-says-nvidia-s-vera-cpu-doubles-sandbox-throughput-under-load.txt", "jsonld": "https://wpnews.pro/news/modal-says-nvidia-s-vera-cpu-doubles-sandbox-throughput-under-load.jsonld"}}