I’m Alice, a computational geneticist and software engineer. I recently launched Carolina Cloud, a cloud service built for researchers, freelancers, and small teams who need high CPU throughput — especially for bioinformatics pipelines and ML workloads, where many algorithms are still single-threaded and don’t lend themselves well to GPU compute, despite the hype!
What makes it different:
- 100% dedicated VMs and containers - no CPU steal
- Up to 128 vCPUs and 512GB RAM resources
- Pricing based on single core performance, no arbitrary "vCPU"
- Unlimited 8Gbps bandwidth, zero egress fees
- GPU nodes available for ML/AI experiments and hybrid pipelines
- 7ms latency to AWS Virginia for hybrid workflows
The goal isn’t to replace AWS/GCP, but to give you a flexible, affordable supplement for CPU-heavy or single-threaded or hybrid workloads. Our happiest customers have saved a lot of money on bioinformatics pipelines, credit modeling, weather simulations, and ML experiments.
If you have questions, want to try it out, or have suggestions, comment here or reach me at alice@carolinacloud.com. I’d love feedback.
Check it out: https://carolinacloud.io/ Console : https://console.carolinacloud.io/
For example, how much would you charge for this machine with 128vCPUs and 256G RAM (1)
Also, what is an "arbitrary vCPU"?
We feel that you should be charged less for older/slower hardware, and more for newer/faster hardware, so we scaled our pricing accordingly.
And regarding egress, because we're fully vertically integrated (we own everything aside from our ISP) nobody else charges us egress and therefore we don't charge our customers egress.
What is the roadmap? Do you see this service continuing to specialize in single region, inexpensive compute with minimal overhead or expand the service with a user facing API, etc?
Technically, we're looking at adding S3-compatible storage onsite, hosting Solana Validator and RPC nodes and even offering colocation to other providers.