AWS AI Watch posted an update
AWS has expanded the regional availability of its EC2 G7 instances, bringing them to London, Mumbai, Sydney and São Paulo. The instances use NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, giving customers in those regions another option for running AI workloads closer to where their data and users are.
Why it mattersAWS says G7 can deliver up to 4.6 times the AI inference performance of G6 instances. Each instance can include up to eight GPUs, with 32 GB of memory per GPU. Those are AWS’s performance figures, not independent benchmark results. Customers can buy the instances on demand, through Spot Instances or with Savings Plans. For teams considering a local deployment, the practical change is regional access, not a new GPU generation.
Discuss: For AI inference, is having GPU capacity in more regions worth paying for local deployment, or should teams prioritise performance and cost over proximity?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.