In 2011, Cycle Computing reported building a temporary 30,472-core cloud cluster called Nekomata for an unnamed pharmaceutical company. Its reported peak cost was $1,279 per hour, and the deployment ran for about seven hours. That is a historical price for one project—not a current quote or confirmation that the same cluster can still be rented.
What was Nekomata?
Nekomata was a high-performance computing cluster assembled from 3,809 Amazon EC2 instances. It was not one enormous computer in a single room: the instances were combined across Amazon data centers in California, Virginia and Ireland to create a temporary pool of computing capacity.
Contemporary reporting described the deployment as the world’s 30th-fastest computer. That ranking refers to the 2011 reports about Nekomata; it should not be read as a current ranking or as a like-for-like comparison with today’s systems.
What did the 30,472-core cluster include?
| Reported specification | What it means |
|---|---|
| 3,809 EC2 instances | The number of cloud compute instances used, as reported by Cycle Computing and Ars Technica in 2011. |
| 30,472 cores | The aggregate core count: 3,809 instances with eight cores each, as reported by Cycle Computing and Ars Technica in 2011. |
| 26.7 TB of RAM | Aggregate memory reported by Ars Technica in 2011. |
| 2 PB of disk | Aggregate disk capacity reported by Ars Technica in 2011. |
| About seven hours | The reported duration of the Nekomata run; this is not a stated minimum rental period. |
Those aggregate figures describe the combined deployment. They do not mean every instance had access to all the cluster’s memory or disk as if it were one local resource; software had to distribute work and data across the machines.
Free tools Windows power users keep installed
One-click scans. No signup required.
#1 Best Overall
Who rented it, and what did it calculate?
Cycle Computing built the cluster for an unnamed “Top 5 Pharma” customer. Public accounts describe high-performance computing and molecular modeling, but do not name the company or disclose a specific drug, compound, or scientific result. The available reporting therefore supports describing the type of work, not claiming a particular discovery.
How much did Nekomata cost?
The reported peak rate was US$1,279 per hour in 2011, including Amazon and Cycle Computing charges. The cluster ran for about seven hours. The hourly figure is the reported peak rate for that deployment, not a current AWS price, a guaranteed rate for every hour of the run, or a general price for renting a 30,472-core machine.
Rank #2
The point was short-term elasticity: the customer could bring together a large amount of computing capacity for a project and then shut it down, rather than buying and operating a permanent data center sized for peak demand. The reporting does not establish a total invoice, a present-day equivalent price, or how a current cloud provider would price comparable work.
Could you rent this exact cluster today?
The 2011 account does not establish that Nekomata or the reported $1,279-per-hour arrangement is available in 2026. Cloud high-performance computing remains a service category, but anyone planning a comparable job would need a new quote from AWS or another provider rather than relying on Nekomata’s historical rate.
Rank #3
A useful modern comparison would account for more than hourly compute charges. Capacity and price depend on the requested instances and their availability; the job’s total cost also depends on runtime and supporting resources. Core count alone does not establish whether a system will finish a particular workload quickly: memory per task, storage, networking between instances, software scaling, and data movement can all matter.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How was a cluster this large assembled?
Cycle Computing combined ordinary cloud instances instead of commissioning a dedicated physical supercomputer. The operator spread the deployment across multiple Amazon regions to obtain enough capacity and provide resilience. At the time, Cycle Computing manager Dave Powers noted that the number of instances obtainable from any one region was uncertain.
The reporting described HTTPS and SSH access and 256-bit AES encryption. Those are security measures cited for the historical deployment; they do not by themselves establish the controls, compliance posture, or configuration a present-day customer would receive. For a current project, security requirements need to be checked against the specific provider, service, region, and workload.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
