NVIDIA’s 13th Revenue Record: AI Demand Turns Compute Into the New Growth Currency

A record that still managed to surprise Wall Street

NVIDIA’s latest earnings report was not merely another strong print from the dominant AI chipmaker; it was a signal that the AI infrastructure cycle is still expanding faster than many market models had assumed. The company reported revenue of $96.2 billion for the second quarter of fiscal 2027, ended July 26, 2026, up 18% from the previous quarter and 106% from a year earlier [1]. That figure marked NVIDIA’s 13th consecutive quarter of record revenue, according to Korean financial reporting on the results, and it came in more than $4 billion above the LSEG consensus of $92.17 billion cited in that report [4].

The earnings beat was broad enough to matter. NVIDIA reported GAAP diluted earnings per share of $2.46 and non-GAAP diluted earnings per share of $2.22, while Associated Press reported that the adjusted result was above the FactSet consensus forecast of $2.09 per share [1][2]. Net income reached $59.69 billion for the May-to-July period, compared with $26.42 billion in the same quarter a year earlier, underscoring how much operating leverage remains in the AI accelerator business [2].

The result also reframed what counts as “high expectations” for NVIDIA. Revenue more than doubled year over year, data-center sales more than doubled, and management still guided above consensus for the current quarter [1][2]. In most industries, that would be a late-cycle blowout. In NVIDIA’s case, it is now the benchmark investors use to test whether the AI trade remains intact.

Data centers are the company

The clearest message from the quarter is that NVIDIA is, for market purposes, now overwhelmingly a data-center company. Data Center revenue reached $89.0 billion, up 18% sequentially and 117% year over year [1]. That segment represented roughly 92% of total quarterly revenue, according to Seoul Economic Daily’s breakdown [4].

The mix also matters. Seoul Economic Daily reported hyperscaler sales of $48.7 billion, up from $24.2 billion a year earlier, while revenue from AI clouds, industrial and enterprise customers reached $40.3 billion, up from $16.9 billion [4]. That suggests the AI demand base is no longer only a story about a handful of mega-cap cloud providers racing to train frontier models. It now includes AI-native clouds, sovereign AI projects, industrial deployments and enterprise infrastructure [3][4].

NVIDIA’s own language reflects that change. CEO Jensen Huang said AI has reached an “inflection point” and argued that compute has become revenue-generating infrastructure rather than speculative capacity [1]. On the call, CFO Colette Kress described demand as coming from hyperscalers, AI labs, AI-native companies, enterprises and sovereign customers [3]. The phrase “AI factory,” once more of a marketing frame than a financial category, now describes the capital-intensive facilities driving the majority of the company’s sales [1][2].

The $108 billion question

The forward guidance may be as important as the quarter already reported. NVIDIA said it expects fiscal third-quarter revenue of $108.0 billion, plus or minus 2%, with GAAP and non-GAAP gross margins expected at 74.0%, plus or minus 50 basis points [1]. Associated Press reported that analysts had been forecasting $104.86 billion for the current quarter [2]. Axios noted that reaching $108 billion would be the first time NVIDIA exceeds $100 billion in a single quarter [3].

That outlook is especially notable because NVIDIA said it is not assuming any data-center compute revenue from China in its fiscal third-quarter guidance [1]. In other words, the forecast above consensus does not depend on a reopening of a politically uncertain Chinese market for advanced AI chips [1][2]. This is an important detail because export controls, licensing rules and China-specific product strategies have repeatedly complicated semiconductor forecasts.

The company’s confidence goes beyond the next quarter. Kress said on the earnings call that NVIDIA expects revenue to grow by about 70% in fiscal 2028 and described that outlook as supply constrained [3]. Associated Press reported that Kress said the company’s growth outlook would be closer to double, based on customer forecasts, if supply were not limiting production [2]. Axios similarly reported that the stock rose after Kress delivered the fiscal 2028 projection and said the company remained supply constrained [3].

Supply, memory and the margin line

The most revealing constraint is no longer demand. It is supply. Huang told analysts that NVIDIA’s “entire supply chain” is challenged and said the company has supply for 70% growth even though demand is much higher [2]. Seoul Economic Daily reported that memory bottlenecks and higher memory prices are weighing on the company, particularly because high-bandwidth memory and DRAM remain essential to AI servers [4].

That pressure appears in the margin outlook. NVIDIA reported GAAP and non-GAAP gross margins of 75.0% for the second quarter, but guided for 74.0% in the third quarter [1]. Seoul Economic Daily reported that NVIDIA expects margin pressure to continue and that higher memory costs were a reason for the lower margin outlook [4]. For a company trading as both a growth story and a profitability story, even a modest margin step-down is material.

Still, the headline remains that margins are exceptionally high for a hardware-centered business at this scale. NVIDIA’s second-quarter GAAP operating income was $63.7 billion, up 124% year over year, on revenue of $96.2 billion [1]. That means the company is not merely selling scarce chips; it is capturing a very large share of the economic value created by the rush to build AI infrastructure.

AWS, financing and the circularity debate

One of the quarter’s most important strategic developments was NVIDIA’s expanded relationship with Amazon Web Services. NVIDIA said AWS will deploy an additional 2 million GPUs beginning in the current quarter through the second quarter of fiscal 2029, along with Vera CPUs, and that AWS will use NVIDIA’s Nemotron models on Amazon Bedrock and SageMaker [3]. Associated Press also reported the AWS plan and said the chips will be incorporated into Amazon’s warehouse robotics efforts [2].

This kind of deal illustrates why NVIDIA’s opportunity is not limited to selling one generation of accelerators. The company is selling GPUs, CPUs, networking, systems software, model infrastructure and robotics-adjacent platforms into the same customer base [1][3]. Each new deployment potentially reinforces CUDA, NVIDIA networking and the company’s role as a full-stack AI infrastructure provider.

But the financing side is drawing more scrutiny. Axios reported that NVIDIA has defended capital solutions and investment structures that support AI customers, including arrangements that critics see as circular financing [3]. Kress acknowledged that some observers would call this circular financing but argued that NVIDIA views the support differently because the companies involved have proven technology leadership and rapidly rising usage [3]. Huang, according to Axios, said he was delighted to invest in frontier AI labs and described the opportunity as generational [3].

The risk is straightforward: if chip suppliers, cloud providers and AI labs finance one another in ways that accelerate purchases, revenue growth can look stronger than end-user monetization alone would justify. The counterargument is also clear: AI infrastructure requires enormous upfront capital, and a supplier with NVIDIA’s balance sheet may rationally help finance platforms that expand its long-term market.

Shareholder returns and market reaction

NVIDIA is also returning cash while funding expansion. During the second quarter, the company returned about $26.0 billion to shareholders through share repurchases and dividends, and it ended the quarter with about $99.0 billion remaining under its share repurchase authorization [1]. The company also said it will pay a quarterly cash dividend of $0.25 per share on October 1, 2026, to shareholders of record on September 10, 2026 [1].

The stock reaction showed how high the bar remains. Associated Press reported that NVIDIA shares ended the regular session down 1.6% before rising 4.1% after hours following the earnings call [2]. Axios reported a similar after-hours rebound and said shares rose after the company provided its fiscal 2028 growth projection [3]. For investors, the print was strong, but the call’s longer-term supply-constrained outlook appears to have been the bigger catalyst.

The bigger read-through

NVIDIA’s 13th straight revenue record confirms that the AI infrastructure cycle has not yet peaked, at least from the supplier’s side [4]. The key demand signal is not only that hyperscalers continue to buy; it is that AI clouds, enterprises, governments and industrial customers are becoming large enough to matter [3][4].

The key risk is that the same scale that makes NVIDIA powerful also makes the AI cycle more fragile. A $108 billion quarterly guide, a 70% fiscal 2028 growth outlook and multiyear GPU commitments point to extraordinary demand [1][3]. They also require power, data-center capacity, memory supply, advanced packaging, customer financing and sustained returns from AI applications.

For now, NVIDIA’s answer is that compute is becoming revenue. The company’s latest results make that argument more credible, but they also raise the burden of proof for the next quarter. A business that just crossed $96 billion in quarterly revenue is now being valued on whether $100 billion-plus quarters can become routine [1][3].

Sources from the last 72 hours

1. [1]NVIDIA Announces Financial Results for Second Quarter Fiscal 2027Aug 26, 2026, 8:20 PM UTC
2. [2]Nvidia projects 70% revenue growth in 2028Aug 26, 2026, 9:51 PM UTC
3. [3]Nvidia CEO Sees 70% Revenue Growth Next Year, Blames Memory BottleneckAug 26, 2026, 11:07 PM UTC
4. [4]Strong AI chip demand powers Nvidia's Q2 results past Wall Street's expectationsAug 26, 2026, 8:40 PM UTC

AI-generated article based on recent web research, then preserved as a dated editorial snapshot.