Thursday, November 21, 2024

Nvidia Blackwell GPU Production Increases Following Revenue Surge

Nvidia CEO Jensen Huang tackled investors’ worries about the Blackwell GPU overheating by assuring them the data center AI accelerator is in full production, with sales set to ramp up this quarter.

On Wednesday, Huang spoke with Wall Street analysts after Nvidia reported a staggering 94% year-over-year revenue surge, reaching $35.1 billion in the fiscal third quarter. Data center revenue hit an all-time high of $30.8 billion, marking a 112% increase from the previous year. The company anticipates revenue could soar to $37.5 billion in the current quarter, with a slight margin of error.

Recently, some investors were on edge after reports surfaced suggesting Nvidia asked suppliers to redesign server racks due to overheating issues with Blackwell. Huang responded indirectly, emphasizing rising demand for Blackwell that outstrips supply for the next few quarters. He stated, “Production is in full steam,” adding that Blackwell’s status is strong and shipment plans exceed earlier estimates.

Nvidia continues to be the dominant GPU supplier driving massive generative AI models for major cloud platforms like AWS, Google, and Microsoft, and also serves Meta, the parent company of Facebook. This soaring demand has shot the company’s valuation past $3 trillion and contributes to nine straight quarters of revenue exceeding expectations.

Alvin Nguyen, an analyst at Forrester Research, noted, “The first few quarters I covered Nvidia brought a lot of variety in the news, but now it’s almost monotonous in terms of performance.”

However, Blackwell hasn’t been without challenges. Although Nvidia announced the product in March with plans for second-quarter shipments, a design issue with its manufacturing partner, Taiwan Semiconductor Manufacturing Company, pushed back production.

Looking ahead, Nvidia faces a potential trade war between the U.S. and China, which generated $5.4 billion in revenue last quarter. As President-elect Trump considers a 60% tariff on Chinese goods, analysts worry that retaliation could impact U.S. companies. Huang assured analysts, “We will comply with any regulation that comes along fully and support our customers to the best of our abilities.”

In the longer play, Nvidia needs to broaden its AI accelerator market beyond cloud providers and large data centers. Surveys show that over the next couple of years, traditional enterprises will deploy AI on less powerful models than Blackwell or its predecessor, Hopper. This shift might open doors for competitors like AMD and Intel. As Nguyen noted, if smaller models running on just a few servers can meet needs, it could significantly change purchasing habits.

Research from Gartner suggests the move to deploying generative AI in enterprise data centers will kick off next year. Organizations are expected to transition from pilot projects to actual production, with generative AI projected to triple enterprise server sales globally to $332 billion by 2028.