Nvidia’s quarterly revenue is forecast to hit US$91.9 billion, a near-doubling year-on-year. While AI demand remains robust, investors are looking for strong Q3 guidance and progress in CPU expansion to sustain the company’s exceptional growth trajectory.
According to Josh Gilbert, etoro Lead Analyst, APAC & Middle East, Nvidia has spent three years making the impossible look routine, and that is exactly the challenge heading into its results this week.
Revenue is expected to reach US$91.9 billion, up 96.5% year-on-year, with adjusted EPS of US$2.08. Doubling revenue at this scale is extraordinary, yet Nvidia’s consistent performance has made exceptional growth increasingly expected by investors.
“Nvidia has spent three years making the impossible look routine, and that is exactly the problem heading into its results this week,” said Gilbert. “Expectations are high going in but, for once, investors aren’t euphoric.”
The stock is up just 1.5% since its last result, modestly trailing the S&P 500 but comfortably outperforming the broader semiconductor index.
AMD provided a preview of how unforgiving the market has become when it reported at the start of the month. Despite beating revenue expectations, softer-than-expected guidance was enough to send shares lower.
“That’s the market we’re in right now, where good numbers aren’t enough if the outlook doesn’t keep pace,” Gilbert added. “For Nvidia, that puts the focus firmly on Q3, where expectations have moved towards US$107 billion to US$110 billion.”
The high bar reflects continued strength in AI infrastructure spending. TSMC’s July sales jumped 44.7% year-on-year, while the company also lifted its 2026 revenue growth outlook, providing another signal that demand across the AI supply chain remains strong. Nvidia’s data centre revenue is expected to more than double, while Rubin is already in full production, with its supply chain roughly twice the size of the Blackwell ramp.
Gilbert noted that one of the more interesting areas is how far Nvidia’s growth is spreading beyond GPUs. Management is targeting around US$20 billion of standalone Vera CPU revenue this year as agentic AI changes the mix of compute needed.
“That takes Nvidia further into territory Intel and AMD have owned for decades, and it shows why the next leg of AI spending may spread across more of the compute stack rather than staying concentrated in GPUs,” Gilbert said.
Nvidia is also selling more of the infrastructure around the chip. Networking was already a standout last quarter, and as AI factories get larger, that part of the business is becoming a growth engine in its own right. Jensen Huang has talked about US$50 billion to US$60 billion of revenue potential for every gigawatt of AI capacity deployed, potentially rising towards US$80 billion to US$100 billion.
“The bigger these AI factories get, the more revenue lands at Nvidia’s door,” Gilbert said.
Investor reaction may ultimately come down to what Huang says about the Rubin ramp and how quickly that capacity comes online. Gross margin will also matter, with Nvidia targeting around 75% as system complexity rises and memory prices continue to climb.
China remains the obvious cloud over the business, while custom ASICs are becoming a more credible competitive threat heading into 2027.
“The real challenge is proving Nvidia can keep raising the revenue base even as the numbers become almost impossibly large,” Gilbert concluded.
