Nvidia has solidified its position as a leader in artificial intelligence (AI) datacentres, reporting a quarterly revenue increase of 16%, which translates to a remarkable 93% rise compared to the same period last year.
Financial Performance and Growth
The company’s datacentre segment generated $35.6 billion in quarterly revenue, contributing to an impressive total of $115 billion for the full year—an increase of 142% year-on-year.
CEO Insights on AI Advancements
During the earnings call, Jensen Huang, the CEO and founder of Nvidia, highlighted the strong demand for the Blackwell AI supercomputers. He noted that enhancing computational power for training leads to smarter AI models. “AI is advancing at light speed as agentic AI and physical AI set the stage for the next wave of AI to revolutionize the largest industries,” Huang stated.
Challenges from Custom Chip Development
Financial analysts raised concerns regarding DeepSeek, an alternative that requires less powerful graphics processing units (GPUs). Major cloud service providers (CSPs) like Microsoft are also developing their own custom chips tailored for AI workloads. According to a transcript from the earnings call available on Seeking Alpha, these CSPs generate approximately half of Nvidia’s revenue. However, Huang believes that expanding enterprise customer demand presents a significant long-term opportunity for Nvidia’s GPUs.
Future Demand for AI Models
Huang expressed confidence that new AI models would drive increased demand, despite advancements that may lead to greater computational efficiency. He explained, “The more the model thinks, the smarter the answer.” Reasoning models such as OpenAI, Grok-3, and DeepSeek-R1 can require up to 100 times more processing power, and future iterations may require even more.
Complexity of the Technology Ecosystem
In response to concerns about CSPs opting for application-specific integrated circuits (ASICs) instead of GPUs, Huang emphasized the intricacy of the technology stack involved. He remarked that constructing an ASIC mirrors the complexity of developing new architectures. “The software stack is incredibly hard,” he noted. “That’s fairly obvious, because the amount of software that the world is building on top of architecture is growing exponentially.”
Market Analyst Perspectives
Alvin Nguyen, a senior analyst at Forrester, commented on Nvidia’s record earnings, stating that such performance has become expected given the ongoing demand for their AI products. He acknowledged that Huang’s remarks about reasoning models were a strong retort to concerns regarding DeepSeek’s impact on demand.
However, Nguyen critiqued Huang’s responses to inquiries about custom chips, describing them as “dismissive.” He pointed out that ignoring the necessity for companies like Amazon, Microsoft, and Google to consider alternatives to Nvidia’s GPUs, alongside the need for semiconductors specifically designed for their AI applications, could be detrimental.