NVIDIA, the dominant leader in the global artificial intelligence (AI) semiconductor market, has once again shattered financial records while securing a massive new multi-billion-dollar deal. Amid lingering market debates regarding over-investment and the durability of the AI boom, NVIDIA CEO Jensen Huang has firmly dismissed skepticism, declaring that the global AI ecosystem has officially reached a definitive "inflection point". 

On August 26 (local time), NVIDIA announced a landmark expansion of its strategic cooperation with Amazon Web Services (AWS), the cloud computing subsidiary of Amazon. Under the new agreement, NVIDIA will supply an additional 2 million advanced AI graphics processing units (GPUs) to AWS’s global infrastructure, slated for deployment primarily between 2027 and 2028. This massive influx of hardware builds directly upon Amazon’s earlier announcement from earlier this year to install 1 million NVIDIA chips within AWS data centers, underscoring an unprecedented acceleration in enterprise and hyperscaler demand. 

While the exact financial terms of the contract were not officially disclosed by either company, industry analysts estimate the total value to reach tens of billions of dollars, given that list prices for these enterprise-grade components run in the tens of thousands per unit. The expanded supply agreement encompasses NVIDIA's most cutting-edge hardware architectures, including Blackwell Ultra, Rubin, and Rubin Ultra GPUs, alongside next-generation networking solutions and NVIDIA Vera CPUs tailored for agentic AI workloads. 

Record-Breaking Financial Performance and Hyperscaler Growth

The announcement of the Amazon deal coincided with NVIDIA’s release of its financial results for the second quarter of its fiscal year (spanning May through July), which demonstrated staggering corporate momentum. NVIDIA reported a quarterly revenue of $96.22 billion, representing a phenomenal 106% increase compared to the same period last year. This milestone marks the 13th consecutive quarter in which NVIDIA has broken its own revenue records. 

A primary driver of this meteoric growth is the surging expenditure by "hyperscalers"—large-scale data center operators and cloud giants. Revenue generated specifically from hyperscale customers surged from $24.2 billion to $48.7 billion year-over-year, effectively doubling as corporations race to build out robust infrastructure capable of training and running massive language models (LLMs). Although major tech corporations such as Amazon, Microsoft, and Google continue to develop their own proprietary chips (such as Amazon’s in-house Trainium and Annapurna Labs innovations) to optimize internal costs, the sheer computational requirements of generative AI necessitate continuous, large-scale procurement of NVIDIA's superior hardware ecosystem. 

Jensen Huang: "AI Has Reached an Inflection Point; Compute is Revenue"

Addressing shareholders and industry observers during the earnings announcement, CEO Jensen Huang delivered an optimistic assessment of the macroeconomic AI landscape. Pushing back against recurring market anxiety regarding "circular financing" or speculative over-investment, Huang emphasized that artificial intelligence has transitioned from a speculative technological novelty into a productive, value-generating utility. 

"AI is now at an inflection point, and it is actively doing useful work," Huang stated in an official corporate statement. Referencing 'tokens'—the fundamental units of data processed during AI model training and inference—he noted, "Tokens are productive and profitable, and compute has now become revenue." 

Huang further elaborated on the shifting dynamics of global demand, noting that while infrastructure buildouts were previously concentrated in a handful of isolated research laboratories, the market has expanded into a golden age characterized by diverse global participants. "Multiple frontier research labs are scaling simultaneously, the open-model ecosystem is thriving, and physical AI is beginning to operate in earnest, creating strong momentum across the United States and worldwide," he explained. 

Highlighting NVIDIA's manufacturing readiness, Huang pointed to the company's latest Vera Rubin platform, noting that mass production is already underway to support the unprecedented demand for agentic and physical AI applications. As global enterprises, sovereign governments, and specialized startups continue to heavily invest in automated workflows and scientific discoveries, the strategic alliance between NVIDIA and AWS is positioned to serve as one of the primary growth engines of the modern digital economy.