Nvidia (Nvidia) announced better-than-expected quarterly financial results on Wednesday, August 26, outperforming market expectations. The company also provided a rare outlook for performance one year ahead, preliminarily estimating a revenue growth of about 70% for the next fiscal year ending in January 2028, far exceeding Wall Street’s previous estimate of 44%.
Despite the company’s already sizable business scale, Nvidia stated that the demand for AI computing is still accelerating. However, supply constraints and rising memory costs will be major limiting factors in the coming quarters.
In the second quarter of this fiscal year, Nvidia achieved revenue of $96.22 billion, a 106% increase from the same period last year, surpassing analysts’ forecast of $92.17 billion. Data center revenue reached $89 billion, up 117% year-on-year. Adjusted earnings per share (EPS) were $2.22, also higher than the market’s estimate of $2.10.
The company expects third-quarter revenue to be $108 billion, with a variance of 2%, exceeding analysts’ average estimate of $104.19 billion. It is worth noting that Nvidia did not include any data center computing revenue from China in this outlook.
Nvidia’s founder and CEO Jensen Huang said, “AI has reached a turning point. It is doing useful work. Its tokens are creating productivity and bringing profits. Nowadays, computing power equals revenue.”
Compared to the quarterly performance forecast, Nvidia’s outlook for the fiscal year 2028 garnered more attention from the market.
Nvidia’s Chief Financial Officer Colette Kress stated in an analyst conference call that the company expects revenue for the next fiscal year ending in January 2028 to grow by 70%, while analysts had previously expected 44%.
Huang said, “We have never predicted or provided guidance on performance one year ahead.”
Kress noted that even at the company’s current scale, demand is still accelerating. She mentioned that customer forecasts “indicate that our growth next year will double,” and added that the guidance provided by the company already reflects restricted supply.
Demand sources are further expanding beyond traditional large cloud service providers. Nvidia expects that AI labs will account for about a quarter of the company’s overall business next year. Additionally, new AI cloud operators like CoreWeave and Nebius are expected to have over 8 gigawatts (GW) of Nvidia GPU computing capacity by the end of this year, up from 3 GW at the end of last year.
The next-generation Vera Rubin platform has also begun shipping to customers. Nvidia anticipates that in the current quarter ending in October, Vera Rubin will account for approximately one-fifth of the company’s data center revenue.
Nvidia announced on the same day an expanded partnership with Amazon’s cloud service (AWS). Both parties plan to deploy an additional 2 million Nvidia GPUs in Amazon’s global infrastructure in 2027 and 2028.
Huang added that aside from these 2 million GPUs, Amazon could also purchase “millions of CPUs.” Nvidia is currently expanding the supply of its new Vera CPU, with some products to be integrated with the Rubin platform.
Kress stated that the capital expenditures of the “big five hyperscale cloud service providers” are expected to increase from around $800 billion in 2026 to $1.3 trillion next year, indicating that major tech companies are continuing to expand their AI infrastructure.
However, strong demand also brings new supply pressure. Nvidia mentioned that memory shortages and rising component costs will weigh on future gross margins in the coming quarters.
The company expects the gross margin for the third quarter to be around 74% and to decrease to a low point of 71% to 72% in the fourth quarter of the 2027 fiscal year (ending in January next year).
Kress said, “We wanted to address this directly rather than have it linger as an unresolved issue. The current memory shortage is largely driven by the construction of AI infrastructure itself.”
The Chinese market still faces uncertainties. In May of this year, the US Department of Commerce approved around 10 Chinese companies including Alibaba, Tencent, and ByteDance to purchase Nvidia’s H200 chips, but related deliveries were at a standstill.
US Department of Commerce officials stated in July that shipments had begun, but the quantity was still “very limited.”
Kress mentioned that due to ongoing geopolitical uncertainties, Nvidia did not include revenue from Chinese data center computing in the third-quarter forecast and forward-looking outlook.