NVIDIA has launched Cosmos 3, an open physical AI foundation model built on a mixture-of-transformers architecture. The model family targets robotics, autonomous vehicles and vision agents, with Super, Nano and upcoming Edge variants. The signal is NVIDIA extending its influence from AI compute into the model and software stack for physical AI.
AI infrastructure and accelerated computing company launching the Cosmos 3 physical AI model family
NVIDIA controls core AI compute infrastructure and is expanding its role in physical AI software, model and deployment layers.
NVIDIA controls core AI compute infrastructure and is expanding its role in physical AI software, model and deployment layers.
AI infrastructure and accelerated computing company launching the Cosmos 3 physical AI model family
Cosmos 3 could influence robotics, autonomous vehicle and vision AI development stacks by combining synthetic data, world modelling and action-policy training.
NVIDIA has launched Cosmos 3, an open physical AI foundation model built on a mixture-of-transformers architecture. The model family targets robotics, autonomous vehicles and vision agents, with Super, Nano and upcoming Edge variants. The signal is NVIDIA extending its influence from AI compute into the model and software stack for physical AI.
Cosmos 3 could influence robotics, autonomous vehicle and vision AI development stacks by combining synthetic data, world modelling and action-policy training.
| 0.90–1.00 | A | High — direct sources |
| 0.75–0.89 | A/B | Strong |
| 0.55–0.74 | B/C | Medium |
| 0.35–0.54 | C/D | Weak–medium |
| 0.10–0.34 | D | Weak signal |
| 0.00–0.09 | D | Internal monitoring |
Direct public sources
• Les versions Super, Nano et Edge ciblent la précision, la vitesse et l'inférence en périphérie Voir aussi: Megaport obtient quatre contrats IA pour construire un cloud d'inférence.
• Cosmos 3 domine les benchmarks de modèles ouverts; les membres de la coalition couvrent la robotique, la génération vidéo et les véhicules autonomes Voir aussi: Les perspectives des puces IA de Broadcom déçoivent les investisseurs.
Le fait
NVIDIA a lancé Cosmos 3, un modèle de fondation d'IA physique ouvert construit sur une architecture mixture-of-transformers. Il prend en charge nativement la compréhension et la génération de texte, d'images, de vidéo, de son ambiant et d'actions. Entraîné sur des milliards d'échantillons multimodaux d'IA physique, Cosmos 3 peut réduire les cycles d'entraînement et d'évaluation de quelques mois à quelques jours. Cosmos 3 se classe premier parmi les modèles ouverts sur plusieurs benchmarks d'IA physique pour la génération de mondes, la politique d'action et la compréhension visuelle. La gamme comprend les versions Super, Nano et la prochaine version Edge, tandis que la Cosmos Coalition comprend Agile Robots, Runway et Skild AI.
L'analyse
Cosmos 3 étend NVIDIA au-delà du matériel dans la couche logicielle du développement de l'IA physique. En combinant la génération de données synthétiques, la modélisation du monde et l'entraînement aux politiques d'action dans une famille de modèles ouverts, NVIDIA réduit les coûts d'entrée pour la robotique, les véhicules autonomes et les agents de vision. Le signal stratégique n'est pas seulement la performance du modèle, mais la tentative de NVIDIA de faire de sa pile le chemin par défaut de la simulation au déploiement dans le monde réel — un scénario qui fait écho à l'emprise de CUDA sur le calcul GPU. Pour les lecteurs de BTW, les exigences d'inférence en temps réel de la variante Edge mettront à l'épreuve la capacité du silicium embarqué de NVIDIA (Jetson) à suivre le rythme des charges de travail robotiques sensibles à la latence. Voir aussi: Nvidia lance la puce RTX Spark pour les PC IA.
Ce qu'il faut surveiller
Surveillez la latence et la consommation électrique de Cosmos Edge dans les déploiements réels, les applications construites par la coalition, et si la famille de modèles stimule la demande de Jetson et DGX. Voir aussi: Les États-Unis comblent la faille des puces IA Nvidia à l’étranger.
Domain of operation
NVIDIA has launched Cosmos 3, an open physical AI foundation model built on a mixture-of-transformers architecture. The model family targets robotics, autonomous vehicles and vision agents, with Super, Nano and upcoming Edge variants. The signal is NVIDIA extending its influence from AI compute into the model and software stack for physical AI.
- Public role: NVIDIA is framed by ai infrastructure and accelerated computing company launching the cosmos 3 physical ai model family and public market context. Evidence basis: NVIDIA Newsroom Cosmos 3 announcement — NVIDIA launched Cosmos 3, described its architecture, model capabilities, benchmark claims, versions and Cosmos Coalition members.; NVIDIA Cosmos product page — NVIDIA presents Cosmos as a physical AI platform for world foundation models, datasets and developer workflows.
- Operating surface: technology and Global provide the public context for this institution profile. Evidence basis: NVIDIA Newsroom Cosmos 3 announcement — NVIDIA launched Cosmos 3, described its architecture, model capabilities, benchmark claims, versions and Cosmos Coalition members.; NVIDIA Cosmos product page — NVIDIA presents Cosmos as a physical AI platform for world foundation models, datasets and developer workflows.
Timeline
- NVIDIA public profile updated
Public coverage records NVIDIA as a subject for role, operating context, and evidence review.
At A Glance
- Name: NVIDIA
- Type: open physical AI foundation model launch
- Base: Global
What It Does
- Public records support monitoring of its role, services, and key relationships.
Why It Matters
- Cosmos 3 could influence robotics, autonomous vehicle and vision AI development stacks by combining synthetic data, world modelling and action-policy training.
- Operational criticality: Medium
- Time horizon: Next quarter
What To Watch
- Monitoring focuses on verified service continuity, governance changes, and relationship signals.
Track verified source updates, role changes, and current public evidence.
Cosmos 3 could influence robotics, autonomous vehicle and vision AI development stacks by combining synthetic data, world modelling and action-policy training.
Longer-term relevance depends on verified operating, policy, and relationship changes.
Member Briefing
Deeper Profile Context
Login is required to unlock the full profile briefing and source notes.
Only for Strategy Circle
Strategic Circle Access
Open to all readers. Unlock profile briefings after joining and logging in.
Join Strategic CircleOnly for Leadership Alliance
Leadership Alliance Access
For owners and management of IP-holding companies. Login required to unlock.
Join Leadership AlliancePublic Sources and Linked Organizations
6 linked-organization notes require member access.
Public View
The public read of NVIDIA is limited to visible role, operating context, and relationship evidence.
Watchpoints
- New public role, affiliation, product, policy, or market disclosures.
- Verified relationship changes involving named organizations or people.
Caveats
- Private or unverified claims are excluded from this public view.
FAQ
Why is NVIDIA included?
NVIDIA has public evidence that makes the institution relevant to BTW's coverage of digital infrastructure, governance, or markets.
What is public about this profile?
The public layer covers visible role, operating context, linked organizations, and evidence-backed watchpoints.
What should readers watch next?
Readers should watch for source-backed role changes, new partnerships, regulatory exposure, operating expansion, or evidence that changes the public assessment.
