NEWS

Nvidia reports its first quarterly earnings since Jensen Huang stepped down as CEO

Nvidia reports its first quarterly earnings since Jensen Huang stepped down as CEO
Image: Redação iMasters
News
Reading
5 min
In short

For the first time without Jensen Huang as CEO, Nvidia reports its third fiscal quarter results this Wednesday, with the market watching the production ramp of the Vera Rubin chip and the new US export rules.

A Nvidia reports its third fiscal quarter results this Wednesday at 5:20am in Hong Kong (6:20pm on Tuesday, Brasília time). It is the company's first quarterly report since co-founder Jensen Huang resigned as CEO, and the market treats the event as a stability test for the world's most valuable chipmaker, according to SCMP. The earnings call with analysts begins at 6am, also Hong Kong time.

Executive Colette Kress took over as interim CEO and arrives at the earnings release under pressure to prove the company can scale its next hardware generation and keep customer orders flowing without the visionary who ran the company for decades. Nvidia shares fell 4.1%, to US$156.14, on Monday, the eve of the earnings report.

The numbers the market will nail down

According to FactSet estimates cited in a Morgan Stanley note on Monday, Wall Street expects revenue of US$108.9 billion for the quarter, up about 91% from a year earlier, and adjusted earnings of US$2.47 per share. Nvidia itself had guided the market, in August, toward revenue of US$108 billion, with a margin of error of 2 percentage points up or down.

The most sensitive question for Kress is whether the company will stick to the revenue growth projection of about 70% for the fiscal year ending in January 2028, made in August. Morgan Stanley is already working with a much more conservative number: 52% growth, after revising its estimate downward last week, citing tight memory supply and increasingly restrictive export rules toward China.

A more immediate gauge is Nvidia's own guidance for the January quarter. Morgan Stanley projects revenue of US$120 billion for that period, below the FactSet consensus of US$125.1 billion.

Vera Rubin: the bet that needs to take off

Hitting that target depends largely on how quickly Nvidia can ramp up production of Vera Rubin, its artificial intelligence computing platform that began shipping in August. Morgan Stanley estimates that Rubin generated about US$20 billion, close to a fifth of data center revenue, in the quarter ended in October, with a growing share in the following two quarters.

Nvidia's board chairman, Harvey Jones, said last week that the company expects to honor commitments with customers and keep its product roadmap without major changes. For analysts, this is the kind of signal that tries to separate the company's governance from the figure of Huang, who concentrated much of these decisions.

A clean quarter and guide.

Dan Ives, partner and senior managing director at the investment bank Yorkville Ives & Co.

For Ives, this would be the decisive step to stabilize market perception of the company, according to a note released on Monday.

Competition, supply chain and China

Arthur Lai, head of Asia technology research at Macquarie Capital, believes that redistributing decisions previously concentrated in Huang's hands could delay price approvals and manufacturing capacity commitments. According to him, Rubin's production is already advanced enough to protect short-term deliveries, but cloud providers could accelerate the qualification of rival chips from Advanced Micro Devices or invest in their own processors, which would give them more control over specifications and deployment costs.

Citi analyst Laura Chen says potential delays in future platform updates could hurt Taiwanese hardware suppliers starting in 2028. Even so, she expects Taiwan Semiconductor Manufacturing Company (TSMC), the world's largest contract chipmaker, to remain relatively resilient, since competing AI accelerators depend on the same advanced manufacturing nodes.

Financing and export controls tighten the squeeze

Beyond operational execution, Kress faces growing scrutiny over how Nvidia finances customer demand and handles increasingly strict US trade restrictions. In August, the executive indicated that AI companies financed through Nvidia's own balance sheet could account for about a quarter of the business in the next fiscal year. JPMorgan credit analysts believe the company will likely keep financing customers, even if new deals take longer or become more expensive should lenders demand additional guarantees on the chipmaker's role in these contracts.

Export controls are another concrete obstacle. On Monday, the US government announced new licensing requirements for shipments of advanced chips to overseas data centers that give China-based companies remote access to significant processing capacity. Nvidia's August guidance already assumed no computing sales to data centers in China. The US also imposed licensing rules for shipments to five countries, including Singapore and Malaysia, subject to exceptions.

Industry sentiment has also cooled. In a Friday report, consultancy SemiAnalysis cut its projection for global AI accelerator spending in 2028 by 28%, to US$758 billion, and lowered its estimate for Nvidia's own computing revenue that year by 39%, citing tighter financing, lower-than-expected shipments and weaker chip prices. Even so, the revised projection still puts industry spending in 2028 well above the US$392 billion expected for this year.

What this means for AI developers in Brazil

Brazil is not among the five countries affected by the new licensing rules, but the country isn't immune to the squeeze. Brazilian companies that rent Nvidia GPU instances from providers such as AWS, Google Cloud and Azure feel the same memory and capacity bottleneck that is pressuring Nvidia's numbers globally, which tends to translate into longer queues and higher prices for top-tier instances, on top of the dollar's exchange rate swings.

Vera Rubin's production pace matters directly to anyone planning to train or fine-tune models locally: the slower the production ramp, the longer Brazilian teams remain dependent on previous-generation hardware, with a worse cost per operation than what the new platform promises. If global cloud providers, as Macquarie points out, start qualifying AMD chips or their own silicon at greater scale, that could, over time, open up price and availability alternatives for contracts in Brazil as well, which today are almost entirely dependent on Nvidia's catalog.

Finally, it's worth watching whether Kress will update the growth guidance for the fiscal year: a downward revision, combined with SemiAnalysis's cut to the sector's projection, would signal a slower AI investment cycle than what was announced in 2025 and 2026, which directly affects investor appetite for AI startups in Brazil that depend on that optimism to raise funds.

Translated from the Brazilian Portuguese original · Read the original