Nvidia CEO Jensen Huang said in an analyst call that the market for Chinese games looks good, as the market’s gamers continues to grow past 300 million players.

“China looks fine,” said Huang. “I think China has stabilized. The gaming market in China has been vibrant and continues to be vibrant. There are all sorts of positive signs in China.”

He said that Nvidia’s push into real-time ray tracing is pushing game developers to create more games with the high-end graphics features.

Earlier today, Nvidia reported earnings and revenues that were well below last year’s figures, but the graphics and artificial intelligence chips company beat analyst expectations, and it is climbing back up after a couple of weak quarters.

Nvidia reported revenues of $2.22 billion, down 31% from a year ago, for the first fiscal quarter ended April 28. Non-GAAP earnings per share were 88 cents, compared with $2.05 a share a year ago. But those numbers beat expectations.

In after-hours trading, Nvidia’s stock was up 6% to $169.70 a share.

Analysts expected Nvidia to report earnings per share of 81 cents on revenues of $2.19 billion. For the full year, revenues are expected to be $11.05 billion. GAAP earnings are expected to be 56 cents a share.

In the past couple of quarters, the Santa Clara, California-based company was recovering from the slow launch for its new RTX graphics cards, and the aftermath of the cryptocurrency bust hurt its earnings.

Nvidia is in the middle of a refresh cycle for its graphics processing units (GPUs), based on its RTX architecture, which enables real-time ray tracing. Those chips got off to a slow start last fall as games weren’t ready to make use of them.

Nvidia said that major game engine makers at Microsoft, Unity, and Epic Games now support real-time ray tracing.

“It’s fair to say ray tracing will be adopted all over the world,” Huang said.

If Nvidia has a slow spot in its business, it is among data center hyperscale customers who are buying the company’s AI and enterprise chips. Nvidia said its inference AI chips did well.

Patrick Moorhead, analyst at Moor Insights & Strategy, said, “I believe that inference was very strong and the company expected data centers to kick back into buying more training solutions in the back half of the fiscal year. Intel told a very similar story and I think it makes sense.”