Nvidia chief executive Jen-Hsun Huang said that deep-learning artificial intelligence has become a new computing platform, and the company is dealing with hundreds of startups in the space that plan to take advantage of the platform.
Speaking at the GPUTech conference in San Jose, California, Huang noted that $5 billion was invested last year in A.I. startups, and there are probably a thousand companies working on the technology for applications ranging from face recognition to self-driving cars.
“Deep learning is not an industry,” he said. “Deep learning is going to be in every industry. Deep learning is going to be in every application.”
The computational complexity of Go is nearly infinite. But Google’s Alpha Go A.I. was able to beat the best human Go player. It runs on 200 graphics processing units (GPUs).
“It’s a pretty amazing achievement,” he said. “Every single year, we do more and more in this area. We think this will change computing. Deep learning is that significant. It’s a new platform. After working on this for five years, it is our fastest-growing business.”
Deep learning has one general algorithm with many different versions. The team at Baidu was able to deconstruct both English and Chinese with the same algorithm. Now the algorithm can be used with massive amounts of data and huge amounts of processing power to yield big benefits in things like computer vision.
“It started in research and moved to computer platform providers,” he said. “They are incorporating it into frameworks and engines. These are the tools of modern network design, the authoring tools for neural networks.”
Nvidia’s CUDA programming language allows these software frameworks and cloud platforms to tap the power of GPU chips.
“Industry after industry is taking advantage of deep learning,” Huang said. “It’s like Thor’s Hammer that fell from the sky. It’s relatively easy to apply.”