Nervana Systems

Nervana Systems
Name	Nervana Systems
Type	Private
Founded	2014
Founders	Ali Farhadi, Amir Khosrowshahi, Awni Hannun
Fate	Acquired by Intel Corporation
Location	San Diego, California
Industry	Computer hardware, Artificial intelligence

Nervana Systems was an American startup founded in 2014 that developed deep learning software and custom hardware for neural networks. It attracted attention from investors, researchers, and technology companies for its work on accelerating machine learning workloads and for its role in the broader competition among hardware vendors such as NVIDIA Corporation, Google LLC, and AMD for AI compute. Nervana combined academic research ties, startup culture, and venture capital backing to pursue end-to-end solutions for large-scale deep learning.

History

Nervana Systems was founded by entrepreneurs with academic backgrounds linked to institutions such as University of Washington, MIT, and Stanford University and staffed by engineers previously affiliated with labs like Google Brain, Microsoft Research, and Facebook AI Research. Early funding rounds involved investors including Shasta Ventures, Data Collective DCVC, and later strategic interest from corporations such as Samsung Electronics and Intel Corporation. The company evolved amid competitive pressure from incumbents like NVIDIA Corporation and challengers like Google DeepMind and OpenAI, positioning itself within the ecosystem that included projects such as TensorFlow, PyTorch, and frameworks from Theano and Caffe.

Technology and Products

Nervana developed a full-stack approach encompassing software frameworks, compiler technology, and custom silicon intended to accelerate models similar to those in research from Yann LeCun, Geoffrey Hinton, and Yoshua Bengio. Product efforts included a deep learning framework and a matrix-multiplication optimized runtime designed to compete with offerings from NVIDIA Corporation's CUDA ecosystem and Intel's MKL. The company highlighted support for architectures used in work by teams at BAIR and labs such as Allen Institute for AI and targeted applications demonstrated in research from Stanford AI Lab and Berkeley AI Research.

Architecture and Performance

Nervana's technical claims emphasized throughput and scalability on workloads inspired by published results from researchers at University of Toronto, Carnegie Mellon University, and Google Research. The firm's designs addressed tensor operations central to models introduced at conferences such as NeurIPS, ICML, and CVPR, and benchmarks compared against accelerators from NVIDIA Corporation and custom designs like Google TPU. Nervana's architecture incorporated ideas related to systolic arrays and matrix multiplication optimizations seen in academic work at MIT CSAIL and industrial labs including Microsoft Research; performance discussions referenced challenges noted in papers from Cornell University and ETH Zurich teams on memory bandwidth and interconnects.

Partnerships and Customers

Nervana engaged with cloud providers and research institutions comparable to Amazon Web Services, Microsoft Azure, and Google Cloud Platform in seeking customers for model training and inference. Potential and announced collaborators included universities and labs such as Stanford University, University of California, Berkeley, and industrial R&D groups at Intel Corporation and Samsung Electronics. The company aimed to serve enterprises active in sectors represented by firms like Uber Technologies, Airbnb, Facebook, Apple Inc., and Microsoft Corporation, where large-scale model training had been demonstrated in published work.

Acquisition by Intel

In 2016 Nervana Systems was acquired by Intel Corporation, a transaction reflecting Intel's strategic response to investments by competitors including NVIDIA Corporation and initiatives like Google TPU and Facebook AI Research hardware efforts. Following the acquisition, leadership movements and organizational integrations linked Nervana engineers with Intel labs such as Intel Nervana groups and collaborations with teams associated with Intel Labs and Intel AI Products Group. The deal mirrored other industry consolidations like NVIDIA Corporation's acquisitions and strategic hires from OpenAI and DeepMind.

Legacy and Impact

Nervana's legacy is tied to the acceleration of interest in AI-specific hardware and end-to-end stacks exemplified by follow-on work at Intel Corporation, renewed investment by firms such as Samsung Electronics, and competitive responses from NVIDIA Corporation and Google LLC. Its influence can be traced through continued research cited in proceedings of NeurIPS, ICLR, and publications from institutions such as MIT, Stanford University, and Carnegie Mellon University, as well as engineering efforts at cloud providers like Amazon Web Services and Google Cloud Platform. The company's trajectory illustrates dynamics seen across the technology industry involving startups, acquisitions, and the interplay between academic research groups and corporate research labs like Microsoft Research and Facebook AI Research.

Category:Companies established in 2014 Category:Intel acquisitions