Liberating A.I. From Hardware Lock-In: Software Is The Key To Our Investment In Lemurian Labs
AI development is increasingly constrained not by hardware, but by software. Today, if you build on one chip, you’re effectively stuck there - rewriting workloads to switch hardware is costly, slow, and energy-inefficient. NVIDIA’s software lock-in with CUDA - an archaic software paradigm - is a tax on AI data centers, foundational model companies, hyperscalers, neoclouds, other chip designers, and any company using AI infrastructure.
Lemurian Labs eliminates this bottleneck. Their software-first, hardware-agnostic compute platform lets developers write AI code once - in a language they already know - and optimize it anywhere. By unlocking heterogenous hardware use cases, Lemurian delivers faster deployment, lower infrastructure costs, and improved energy efficiency for anyone deploying AI workloads at scale.
The timing couldn’t be better. As new accelerators and GPU alternatives emerge, AI software portability has become a mission-critical need. Lemurian’s approach provides the flexibility and sustainability the market has been missing.
The team includes veterans from NVIDIA, Qualcomm, Sun, IBM, and Intel, bringing deep expertise in the architectures on which today’s AI depends on. The future of AI requires open, software-defined compute, not vertically locked stacks.
Origin Ventures VI participated in Lemurian’s oversubscribed $28M Series A co-led by Pebblebed Ventures and Hexagon, which included participation from Oval Park Capital, Uncorrelated Ventures, Untapped Ventures, and Silicon Catalyst, as covered in EE Times and siliconANGLE.
Lemurian Labs sits right at the heart of Origin Ventures' focus on the Artificial Intelligence economy, including opportunities to improve the unit economics of AI training and inference. Our investment in Torus provides energy and storage solutions to AI data centers, and our investment in Opticore provides better computing hardware for AI workloads. We’re excited to back co-founder and CEO Jay Dawani and the Lemurian team as they build the software portability layer that will make AI faster, more flexible, and more sustainable.
