LOTA latency and throughput performance specifications
Company: CoreWeave
The claim, verbatim
CoreWeave's AI Object Storage LOTA delivers reads at local NVMe speeds, reducing latency by 8 times compared to traditional storage clusters and providing up to 7 GB/s throughput per GPU.
Source (primary)
CoreWeave launches multi-rack NVIDIA Vera Rubin AI cloud clusters - ROI-NJ (-, news_article)
View cached copy (2026-09-19)Live source ↗
Quote: “The company said LOTA reduces latency by eight times compared with reading from a traditional storage cluster and provides up to 7 GB/s of throughput per GPU.”
How we checked this
Checked on September 24, 2026. The cited source supports every part of this claim.
The article reports all three LOTA performance figures as CoreWeave's own stated specifications.
Confirmed in the source:
- CoreWeave AI Object Storage LOTA (Local Object Transport Accelerator) delivers reads at local NVMe speeds, using managed caching on each CoreWeave Kubernetes Service node
- According to CoreWeave, LOTA reduces latency by eight times compared with reading from a traditional storage cluster
- According to CoreWeave, LOTA provides up to 7 GB/s of throughput per GPU
What we did: Read our cached copy of the publisher (https://www.roi-nj.com/2026/09/16/tech/coreweave-launches-multi-rack-nvidia-vera-rubin-ai-cloud-clusters/) in full (13,018 characters, retrieved September 19, 2026) and checked each assertion in the claim against it.
Additional evidence
confirms CoreWeave launches multi-rack NVIDIA Vera Rubin AI cloud clusters - ROI-NJ
Quote: “The company said LOTA reduces latency by eight times compared with reading from a traditional storage cluster and provides up to 7 GB/s of throughput per GPU.”