Sales of GPUs into the gaming market rose 40%, year over year, to $1.348 billion. Sales of the company's products for data center were up 67% from the prior-year period, and up 23% from the prior quarter, at $296 million. More data on the quarter is available in a document of commentary from CFO Colette Kraus on Nvidia's IR Web site.

CEO Jen-Hsun Huang was kind enough to talk with me by phone following the report. "I think that the results show we are the fastest growing technology company in the world," he said. "We had a record quarter, a record year, and all of our businesses are growing."

Said Huang, "this is the beginning of an exciting new period in computing," based on his belief that artificial intelligence, and ways of achieving it, such as deep learning, are taking over all of software.

"A.I. is just the future of software," says Huang, in reply to my question as to whether deep learning showed up in other parts of the business besides just the "data center" product line.

Well, that's an interesting question. The most important thing that's happening is A.I. computing. Data center tripled. This is just beginning. More and more of the game developers are using A.I. to control bots, and animation, and the list goes on. In self-driving car business, if not for A.I., it would be complete nonstarter. In the enterprise business, you see healthcare with imaging, and the large pharma companies that are looking into A.I. Across the board, A.I. is going to revolutionize every industry. A.I. is just the future of software.

I asked Huang about a contention made by NXP Semiconductors (NXPI) CEO Rick Clemmer, in an interview I did with him last month at the Consumer Electronics Show in Las Vegas. Clemmer told me he doesn't think any auto maker would support putting Nvidia chips cost $3,000 into automobiles to support self-driving computation.

Huang replied, "He's missing it by a factor of 10," by which he meant Nvidia self-driving approaches require chips costing hundreds, not thousands. "We have a Level 3 [autonomy] solution in one chip, and that's a couple to three hundred dollars. Next year, we'll have Level 4 in one chip. So, you're talking, $300, $400, maybe $500.

Huang added that for Level 5 autonomy, which means full autonomy, with no person doing anything, it would require "several chips," by which he meant four of the Xavier" modules he showed off at CES. "But that's Level 5, without a driver at all." When I asked how much that might cost, he conceded it might be several thousands of dollars.

"Okay, so, he wasn't entirely right," Huang said of Clemmer.

I also asked about Nvidia competitor Advanced Micro Devices (AMD) making a new push into data center applications. AMD has claimed that Nvidia's software technology for programming machine learning, CUDA, is not open enough. I phrased it as "open source."

Huang shot back, "CUDA is open source, everyone can access it -- and every application for deep learning has been ported to CUDA."

He rattled off the ways in which CUDA has become a foundational technology:

They [developers] are going to use the deep learning frameworks, the quantum chemistry code, the imaging processing code. I think they're [AMD] missing the point. The key point that is really important is that what's really changed about our company's business is we are not just a GPU chip company. We've spent the last decade building up our platforms. That includes NCCL, TensorRT [etc.]. That was a decade endeavor for us.

But, I replied, AMD's contention would seem to be that Nvidia alone controls CUDA, that some cloud computing companies are concerned the platform is not "open enough."

Replied Huang, "Somebody has to innovate; we can't all sponge off somebody else."