Loading organizations...
§ Private Profile · Lewes, DE, USA
AI-Native Infrastructure for Spatial AI based in Lewes, Delaware, United States of America.
AI-Native Infrastructure for Spatial AI
Zibra sits at the intersection of spatial computing and AI infrastructure. Our compression-native infrastructure serves as an enabling layer for training directly on large-scale 3D data - exactly the kind of infrastructure required for next-generation AI systems across multiple verticals. Our GPU-native, PyTorch-integrated compression technology is purpose-built for massive 3D datasets. It achieves up to 100X compression while supporting real-time decompression at 600+ GB/s, significantly improving GPU utilization, training throughput, and model convergence in spatial-AI and simulation-heavy domains such as robotics, autonomous systems, aerospace, digital twins, and climate modeling. By removing low-signal and redundant data, our compression accelerates training: we see >12% faster model convergence in vertical scaling setups on a single GPU and >30% improvements on multi-GPU and distributed clusters. In addition, our instant decompression drives GPU utilization up to ~95%.
Zibra AI has raised $2.0M across 1 funding round.
Zibra AI has raised $2.0M in total across 1 funding round.
Zibra AI has raised $2.0M across 1 funding round. Most recently, it raised $2.0M Seed in August 2023.
| Date | Round | Lead Investors | Other Investors | Status |
|---|---|---|---|---|
| Aug 1, 2023 | $2M Seed | Troy Kirwin | Hartmann Capital, ParaFi Capital, Mona EL ISA, Stani Kulechov, Phil Libin, Sebastien Borget | Announced |
Zibra AI has raised $2.0M in total across 1 funding round.
Zibra AI's investors include Troy Kirwin, Hartmann Capital, ParaFi Capital, Mona El Isa, Stani Kulechov, Phil Libin, Sebastien Borget.
AI-Native Infrastructure for Spatial AI
Zibra sits at the intersection of spatial computing and AI infrastructure. Our compression-native infrastructure serves as an enabling layer for training directly on large-scale 3D data - exactly the kind of infrastructure required for next-generation AI systems across multiple verticals. Our GPU-native, PyTorch-integrated compression technology is purpose-built for massive 3D datasets. It achieves up to 100X compression while supporting real-time decompression at 600+ GB/s, significantly improving GPU utilization, training throughput, and model convergence in spatial-AI and simulation-heavy domains such as robotics, autonomous systems, aerospace, digital twins, and climate modeling. By removing low-signal and redundant data, our compression accelerates training: we see >12% faster model convergence in vertical scaling setups on a single GPU and >30% improvements on multi-GPU and distributed clusters. In addition, our instant decompression drives GPU utilization up to ~95%.