THE FUTURE IS HERE

The First AI Chip Startup with Naveen Rao, Nervana Systems

Over the past five years, over 50 companies dedicated to making artificial intelligence run faster have sprung up. Whether in data centers or in cars, like Tesla’s automobiles, all of these innova-tions arguably started with NVIDIA and the move to deep learning on graphics processing units (GPUs). This created an entire industry. The first AI chip company was Nervana Systems, and its Founder, Naveen Rao, joins us to shares his insights. In this episode, we hear about Naveen’s academic and professional backgrounds and how he used his engineering and neuroscience skills to start Nervana after working at Qualcomm. We discuss what fundraising for Nervana was like and how the venture capital (VC) landscape changed its attitude to hardware concurrent with Nervana’s growth and innovation. The company was ultimately acquired by Intel, and we learn about how it went from a business development proposition to acquisition. Naveen shares the personnel and resource constraints Nervana faced despite being part of a much larger organi-zation such as Intel, and how his role ultimately changed over time. While Intel has kept some of Nervana’s developments, they have also gone in different directions, and Naveen shares his take on this. The discussion also touches on Nervana and NVIDA’s love/hate relationship, what is piquing Naveen’s interest now that he is no longer at Intel, and where he thinks the outcomes of the work in the space are likely to lead. To hear more, tune in today!

Key Points From This Episode:

· Naveen’s background and moving from engineering to computational neuroscience.
· Hear about Naveen’s time as Qualcomm and the neuromorphic computing research he was doing.
· How a failed internal pitch for an artificial neural network at Qualcomm led to Nervana’s crea-tion.
· Nervana’s founding story, building the team, and fundraising for the company.
· The difficulty of pitching chip investments to VCs, who has lost money in the early 2000s.
· Changes in VC attitudes towards hardware by the time Nervana had raised 20 million dollars.
· Hear more about the nature of Nervana’s relationship with NVIDIA.
· How the deal with Intel went from biz dev to acquisition.
· Some of the obstacles that came with being bought by a big company.
· Getting an offer to head up the AI team at Intel and the challenges this brought for Naveen.
· How the release of Volta, NVIDIA’s chip, changed the competition in the space.
· The game plan changes after Volta’s release and splitting into inference and training paths.
· The importance of patience in the space; NVDIA has always bet on their team.
· Naveen’s position on Intel acquiring Habana and how his job changed after this acquisition.
· Naveen’s impression of the AI companies that have sprouted up since.
· Unpacking the critical software for this type of development.
· Google’s GPU team and the advantage they have on the inference side.
· NVIDA’s acquisitions and Naveen’s take on what they are trying to do.
· What Naveen is up to now that he is no longer at Intel.

Tweetables:

“Once we got acquired, we were actually put under a hiring freeze. It became very difficult, or ac-tually, impossible for us to hit any timelines, because we knew we had to grow by 50 people.” — @NaveenGRao [0:20:04]

“I think, if Intel doesn’t react well enough to actually maintain the markets they have, NVIDIA is going to eat at them and that’s what they’re going to do with Arm. That’s really going to be painful.” — @NaveenGRao [0:47:18]

More FYI Podcasts: https://ark-invest.com/research/podcast

To learn more about ARK: https://ark-invest.com/

For more updates, follow us on:
– Twitter: https://arkinv.st/3corDiY
– LinkedIn: http://arkinv.st/1f7AiVX
– Facebook: http://arkinv.st/3r4EDOU
– Instagram: http://arkinv.st/39td8bO

Disclosure: http://arkinv.st/39rzF94