Nvidia has passed its first major earnings test without co-founder Jensen Huang but the big strategic test is still to come, according to Saxo chief investment strategist Charu Chanana.
The chip giant reported third-quarter revenue of $115.8bn on Tuesday, up 103% on a year earlier, with data centre revenue of $107.6bn, up 110%. It guided to a fourth-quarter revenue of $130bn, plus or minus 3%.
Management also raised its fiscal 2028 revenue growth outlook to approximately 75% from 70%. Vera Rubin, Nvidia’s new AI data centre platform, generated nearly $20bn in its first quarter of production shipments, which is the fastest product ramp in the company’s history.
“Nvidia’s underlying business momentum remains intact,” Chanana said. “The AI infrastructure cycle is still delivering extraordinary growth, even as investors become more demanding about valuations and future returns.”
Huang resigned as chief executive on 12 November, five days before the results, and left the board following reports of extramarital affairs. Colette Kress, Nvidia’s chief financial officer, is interim chief executive while the board considers internal and external candidates for the permanent role. An independent committee of Nvidia’s board is also conducting an investigation into claims that Huang has been the target of "a foreign intelligence effort".
Chanana said: “The results were impressive, but perhaps the bigger confidence signal was management’s willingness to provide guidance at all, given earlier concerns that the ongoing investigation might constrain its outlook.”
Kress’s first earnings call as interim chief executive reassured investors that Nvidia’s technology roadmap, customer relationships and competitive advantages could survive without Huang. Kress has therefore passed the first test of operational continuity, the Saxo strategist said.
“However, there is an important distinction between executing an established roadmap and setting the company’s future strategic direction,” Chanana added.
Nvidia’s shares gained around 4–5% in after-hours trading, a positive response but, in Chanana’s view, a modest one given the scale of the earnings and upgraded guidance. The shares have fallen close to 30% over the past month.
“The relatively restrained reaction suggests that confidence in the business is recovering faster than confidence in the stock,” Chanana said. “Nvidia’s shares had already fallen substantially from their October peak ahead of the results, compressing its valuation multiple. That correction creates a more favourable starting point for investors, particularly if earnings estimates continue to rise.”
Even so, she argued that “a lower valuation is not necessarily a floor”, because the market may keep demanding a discount for leadership uncertainty, governance risks and questions over the sustainability of AI spending.
In Chanana’s view, investors face two separate risks: the market may stay unwilling to pay Nvidia’s historical premium even as earnings expectations improve and earnings forecasts may fall if AI infrastructure spending slows or profitability weakens. Continued earnings upgrades, by contrast, could support a share-price recovery even without a full return to the previous premium.
“The valuation correction has made Nvidia’s risk-reward balance more interesting, but a durable recovery will depend on sustained earnings growth and greater clarity around its strategic risks, rather than simply a return to its former valuation multiple.”
Nvidia’s adjusted gross margin was 74.9% in the third quarter, and management expects it to moderate to 71–72% in the fourth before recovering to 72–73% in fiscal 2028.
Saxo’s strategist is not overly concerned by some moderation during a major product transition, but competition from rival chipmakers and from hyperscalers developing their own AI chips could create more persistent pricing pressure over time.
“Revenue growth alone is not enough,” Chanana said, adding that investors should watch whether Nvidia can deliver the expected margin recovery while maintaining its pricing power amid intensifying competition.
Nvidia has helped to finance the companies buying its chips, particularly smaller cloud providers known as neoclouds, earning its nickname the ‘central bank of AI’. This has raised concerns about circular financing, where some chip demand is supported by Nvidia’s own financial commitments rather than entirely by independent customer demand.
Kress’s emphasis on being “disciplined and selective” on its neocloud customers could signal a more measured approach to capital allocation, Chanana said, although it is too early to conclude that this represents a fundamental shift in Nvidia’s strategy.
Greater discipline could also reduce speculative overinvestment, helping AI growth become more sustainable. However, it could also slow expansion among smaller, leveraged infrastructure providers reliant on Nvidia for financial backing.
The key test, Chanana said, is whether genuine customer revenues and independent capital can increasingly fund AI companies’ expansion instead of their largest chip supplier.
Nvidia’s fourth-quarter outlook assumes no data centre compute revenue from China, which Chanana said limits the immediate earnings exposure to tighter US export restrictions. The company reported $2.4bn of licensed Hopper data centre sales to China in the quarter but has stopped those shipments, she said.
According to Chanana, the latest US restrictions on exports go further, affecting advanced chip shipments to south-east Asian markets and Chinese companies’ access to computing capacity hosted in overseas data centres.
Chinese technology companies have used overseas data centres to access advanced AI computing power, so restricting that channel could affect demand that previously looked as though it originated outside China.
“Strong demand elsewhere may protect Nvidia’s earnings in the near term, but investors should not assume lost China-linked demand can always be replaced,” Chanana said.
In her view, the larger long-term risk is a smaller addressable market and stronger Chinese competitors, as restrictions could speed up China’s development of domestic AI chips and split the global AI market into separate US and Chinese systems.
The results should support confidence in the demand outlook for major semiconductor suppliers, particularly Asian companies such as TSMC, SK Hynix and Samsung, Chanana added, as they show the AI infrastructure cycle remains strong.
“However, strong Nvidia earnings do not necessarily mean the previous broad-based AI rally can simply resume,” she added.
Chanana said the AI investment cycle has moved from chips to infrastructure and is heading towards adoption, where the focus is likely to fall on businesses generating measurable productivity gains, revenue growth and cost savings from AI.
This transition is proving volatile, she said, as investors question returns on spending, financing and depreciation costs, while higher bond yields and concentration risks make them less willing to chase growth stocks indiscriminately.
Saxo remains constructive on the long-term opportunity created by AI, but thinks the next phase is unlikely to reward all AI-related stocks equally.
Within infrastructure, the better placed companies are those with strong balance sheets, pricing power and sustainable earnings, while opportunities beyond this could emerge among businesses showing tangible economic returns from adopting AI.
“For diversified portfolios, it is increasingly important to distinguish between different AI business models and avoid relying on a single technology theme or capital expenditure cycle to drive returns,” Chanana said.
“For Nvidia, strong earnings have bought management time. For the broader AI trade, they reinforce confidence in infrastructure demand, but the next leg of the rally may depend increasingly on which companies can turn that demand into sustainable profits.”