It was the ultimate setting – like playing baseball at Yankee Stadium or singing opera at La Scala.
NVIDIA’s Inception Award competition for promising AI startups found its perfect home today in a packed auditorium in Tel Aviv, the epicenter of Israel’s startup culture.
In a nation with the highest number of startups per capita — some 5,000 fledgling companies in a country of 8.5 million people — six of the best competed head-on at GTC Israel, our inaugural local edition of the GPU Technology Conference, which is on a global swing through six cities this fall.
In quarter-hour blocks, their CEOs described their business strategy and fielded questions from five judges with deep startup experience, including execs from three Israeli VC firms specializing in AI. The format, longer than the typical several-minute martial-arts like square-offs that place in other startup competitions.
As more than 150 GTC attendees looked on, the judges named as winner Cognata, which uses simulations to train autonomous vehicles across wide-ranging, realistic environments without ever pulling out of the driveway. The prize: an NVIDIA DGX Station personal AI supercomputer, with four of the latest NVIDIA V100 GPU accelerators providing the processing capability of 400 CPUs, worth $69,000.
“We’ll be number crunching and running generative adversarial networks all day and night with this,” said Danny Atsmon, Cognata’s jubiliant CEO, as he received the award. He was joined on stage by the company’s full complement of 16 cheering staff members.
Atsmon, an expert in ADAS and deep learning, said that he’s now founded three companies and all were built on NVIDIA’s GPU technology. He said AI and GPUs are especially critical for Cognata, which estimates that an autonomous car will need 11 billion miles of road travel before it can be fully trained — something impossible without simulation.
“We’re grateful to NVIDIA for democratizing the AI market for the rest of the world,” he said.
Other companies that competed for the Inception Awards in Tel Aviv included:
- Deep Trading – Seeks to unravel the mystery of international financial markets by using deep learning and CUDA to predict the movements of financial markets.
- VaultML – Deploys AI to sort out flops from hits by predicting how audiences will react to content — such as scripts, movies and trailers.
- Cognitive ID – Anticipates computer screens without passwords, which it seeks to replace with authentication algorithms that analyze up to 700 cognitive, physiological and psychological parameters.
- VayaVision – Fuses data from lidar, radar and cameras to generate 3D models of surroundings.
- Presenso – Uses AI to provide real-time predictions of when industrial equipment, like heavy machinery, are likely to fail.