Source: Quantum
Nvidia CEO Huang Renxun is as confident as ever about the company's future, and in the latest earnings call on Wednesday, he reiterated that DeepSeek will not affect sales.
Speculation that DeepSeek's R1 model requires far fewer training chips caused Nvidia's stock price to hit a record low last month.
But in this earnings call, Huang praised R1 as an "excellent innovation" and emphasized that it and other "inference" models are good news for Nvidia because they require more computing power.
"Inference models can consume 100 times more computing resources, and the amount of computing for inference models in the future will be even greater," Huang Renxun said. "DeepSeek R1 has ignited global enthusiasm. It is an excellent innovation, but more importantly, it has open-sourced a world-class inference AI model. Almost every AI developer is applying R1."
Nvidia's sales show no signs of slowing down. Nvidia reported another record quarter, with revenue of $39.3 billion (about 285.377 billion yuan), exceeding its own forecast and Wall Street expectations. The company also said it expects revenue to rise again to around $43 billion (about 312.244 billion yuan) in the next quarter.
According to the financial report released by the technology giant, Nvidia's data center sales will almost double to $115 billion (about 835.072 billion yuan) in 2024, an increase of 16% from the previous month.
On the conference call, Huang Renxun touted Nvidia's latest Blackwell chip as designed for inference and said that the demand for it is currently "very large."
"We will achieve strong growth in 2025," Huang Renxun said.
In fact, despite the brief panic caused by DeepSeek last month, the market for artificial intelligence chips shows no signs of cooling down.
Since then, Meta, Google, and Amazon have all announced massive AI infrastructure investments that will collectively invest hundreds of billions of dollars in the coming years.