Nvidia has come across more than 1,300 AI startups in its Inception program, but the graphics and AI chip maker believes that there aren’t enough artificial intelligence startups out there.

Jeff Herbst, vice president of business development, said that Nvidia wants to partner with venture capitalists to fund even more AI startups that will build out the necessary parts of the ecosystem for AI, which is expected to become a huge part of the technology landscape going forward. He said the doors are open to partnering because there are so many more startups than Nvidia can help or finance on its own.


Learn how developers and brand marketers are using AI to grow their businesses, at MB 2017 in July 11-12 in SF. We cut through the hype to show how marketers are achieving real ROI. Get your 50% off Early Bird ticket by May 19!


Nvidia held a lunch on Thursday at its GPU Technology Conference to enlist dozens of venture capitalists in its crusade to change the world through artificial intelligence. Five startups made presentations to the VCs during the lunch, and Herbst spoke with Jim McHugh, general manager of Nvidia’s deep learning group, about the trends that will create opportunities for AI startups in the future.

Above: Jeff Herbst of Nvidia

Image Credit: Dean Takahashi

“Every meaningful startup is using a GPU in some form now,” Herbst said. “Every relevant AI company comes through us. We are at the center of the revolution. We can be a great source of deal flow” for venture capitalists.

Nvidia also launched its Tesla V100 AI processors today, which has 21 billion transistors and is about three times as powerful as the chip that Nvidia launched a year ago.

McHugh said that it’s amazing to see a chip platform move so fast and stay ahead of Moore’s Law, the prediction that chip performance will double every couple of years as the number of transistors double every couple of years. But he thinks that startups in particular will benefit from the Nvidia DGX Station, which essentially lets startups get started on AI processing right away.

Nvidia is a strategic investor, and it has invested in about 15 of the 1,300 startups in its Inception program. But there are more AI startups than Nvidia can invest in, and it wants to keep investing in companies that find a new way to use Nvidia’s GPUs.

Jay Eum, managing director at Translink Capital, said he thought the idea of partnering with strategic investors was a good one because they can open so many doors for startups.

During the session, VCs heard presentations from five AI startups: Datalogue (automating data preparation), Drive.ai (building the brain of self-driving cars), Fastdata.io (using GPUs to accelerate streaming data), MapD (accelerating SQL database processing with GPUs), and SoundHound (a leader in voice-enabled AI).

MapD recently closed a $25 million round led by NEA, and it has open-sourced its platform. Herbst said that Nvidia has three more investments in the pipeline. Nvidia plans to have more GPU Ventures events in the future.

“We are enabling the software ecosystem around our platform,” Herbst said. “It’s a platform focus. Don’t think of us as guys in Cupertino, Santa Clara, and San Diego. We are completely different. We are financial investors, but we are primarily strategic investors.”