Amazon Web Services has made EC2 G7 instances generally available, pairing NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs with AWS networking and storage options for AI inference, graphics, analytics, and real-time data workloads.
Amazon Web Services announced the general availability of Amazon EC2 G7 instances accelerated by NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, positioning the new instance family for AI inference, graphics, and data analytics workloads.
In an AWS News Blog post, Amazon Web Services said EC2 G7 instances are designed for organizations deploying AI models, running graphics-intensive applications, and processing data analytics workloads. AWS says the instances use NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs and can deliver up to 4.6 times the AI inference performance of comparable G6 instances.
AWS’s EC2 G7 product page lists the family as supporting configurations with up to eight NVIDIA RTX PRO 4500 Blackwell GPUs. The same AWS product page says the instances support up to 700 Gbps of Elastic Fabric Adapter networking, which is intended for low-latency, high-throughput communication between instances.
According to AWS, G7 instances are aimed at use cases including AI inference, graphics rendering, visualization, data analytics, and real-time workloads. AWS’s regional availability notice says the instances are now available in the US East (N. Virginia) Region and can be used for AI model deployment, recommender systems, retrieval-augmented generation inference, and real-time data pipelines.
NVIDIA also described the G7 launch as part of its collaboration with AWS to bring AI workloads into production environments at larger scale. In a NVIDIA Blog post, the company said EC2 G7 brings RTX PRO 4500 Blackwell GPUs to AWS infrastructure for production AI use cases.
NVIDIA also pointed to integration with Amazon OpenSearch Serverless, saying GPU-accelerated vector indexing can be paired with the new compute instances. That matters for retrieval-augmented generation and semantic search workloads, where vector search is often used to retrieve relevant information before a model generates a response.
The announcement reflects a broader shift in cloud AI infrastructure from model training alone toward deployment and inference. AWS’s own materials emphasize workloads such as recommender systems, RAG inference, model deployment, and real-time analytics rather than only large-scale training.
For enterprises, the significance is practical: inference workloads often need predictable latency, scalable deployment options, and integration with existing cloud services. AWS is presenting G7 as a GPU-backed EC2 option for those production needs, while NVIDIA is highlighting the Blackwell-based GPUs as part of the infrastructure stack for deployed AI applications.
The launch also expands the number of NVIDIA Blackwell-based options available through a major cloud provider. AWS has not positioned G7 as replacing its other GPU families; instead, its product materials frame G7 as a specialized EC2 family for AI inference, graphics, and analytics.
AWS says Amazon EC2 G7 instances are generally available, and its regional announcement states that the instances are available in the AWS US East (N. Virginia) Region. Customers can consult the Amazon EC2 G7 instance page for supported sizes, hardware specifications, and workload guidance.
As with any cloud infrastructure release, the practical impact will depend on application requirements, regional availability, pricing, and how customers configure surrounding services such as storage, networking, observability, and model-serving software.
What AWS announced In an AWS News Blog post, Amazon Web Services said EC2 G7 instances are designed for organizations deploying AI models, running graphics intensive applications, and processing data analytics workloads.
AWS says the instances use NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs and can deliver up to 4.6 times the AI inference performance of comparable G6 instances.
AWS’s EC2 G7 product page lists the family as supporting configurations with up to eight NVIDIA RTX PRO 4500 Blackwell GPUs.
Continue reading
