Skip to main content

Cloud Workstations

The Cloud-Native Lab: An Alternative to Physical Hardware​

If you do not have access to RTX-enabled workstations, a cloud-native lab is a viable alternative. This approach relies on cloud-based instances (like AWS RoboMaker or NVIDIA's cloud delivery for Omniverse) and is ideal for rapid deployment or if you have a less powerful laptop.

info

Building a "Physical AI" lab is a significant investment. You will have to choose between building a physical On-Premise Lab (High Capital Expenditure) versus running a Cloud-Native Lab (High Operating Expenditure).

The "Ether" Lab (Cloud-Native)​

This option is best for:

  • Rapid deployment
  • Students with weaker laptops

1. Cloud Workstations (AWS/Azure)​

Instead of buying powerful PCs, you can rent cloud instances.

  • Instance Type: AWS g5.2xlarge (A10G GPU, 24GB VRAM) or g6e.xlarge.
  • Software: NVIDIA Isaac Sim on Omniverse Cloud (requires a specific Amazon Machine Image - AMI).

Cost Calculation​

  • Instance Cost: ~$1.50/hour (using a mix of spot and on-demand instances).
  • Usage: 10 hours/week × 12 weeks = 120 hours.
  • Storage: ~$25/quarter (for EBS volumes to save your environments).
  • Total Cloud Bill: ~$205 per quarter.