Habana Labs is an AI Processor company building ground-up platforms optimized for training deep neural networks and for inference deployment in production environments. Habana is unlocking the true potential of AI with orders of magnitude gains in processing performance, scalability, cost and efficiency.
Habana Labs was acquired by Intel for ≈$2B to advance Intel's AI strategy with a high-performance training processor family and a standards-based programming environment to address evolving AIworkload
David Dahan, Eitan Medina, Ran Halutz, Shlomo Raikin
AI / Machine Learning
Artificial Intelligence & Machine Learning
Habana Labs Ltd.
100% AI. Processors purpose-built to unlock performance.
Why we backed Habana Labs?
Innovation Gaps & Market Needs
CPU/GPU architectures are not optimized for neural network workloads. Habana approached this problem by developing its processor platforms to be optimized for training deep neural networks and keeping in mind deployment environments.
Execution & Outcome Driven Leadership
For three years, Habana developed cutting-edge AI processors: Gaudi (for training with integrated remote direct memory access over Converged Ethernet) and Goya (for inference). In 2019, Goya demonstrated performance of 15,453 images-per-second throughput on ResNet-50. Intel acquired Habana to strengthen its AI portfolio and accelerate its efforts in the AI silicon market.