Nvidia’s stock has recently seen a 12% decline from its all-time high, largely due to concerns over DeepSeek’s AI efficiency breakthroughs.
The Chinese startup claimed to have trained competitive AI models using far fewer GPUs than industry leaders like OpenAI. This sparked fears that Nvidia’s high-end AI chips could see a drop in demand.
However, recent comments from Alphabet CEO Sundar Pichai suggest that Nvidia’s dominance in AI computing is far from over.
DeepSeek’s AI Innovations and the Market Reaction
DeepSeek, a China-based AI company, has introduced highly efficient software techniques to train AI models at a fraction of the usual cost.
By using methods like knowledge distillation and optimized data processing, DeepSeek managed to train models using older Nvidia GPUs like the H100 instead of the latest AI chips.
This raised concerns that AI developers could reduce their reliance on Nvidia’s newest GPUs, potentially impacting the company’s revenue.
Sundar Pichai’s Key Insight on AI Computing Demand
In a Feb. 4 earnings call, Sundar Pichai addressed the growing debate around AI training versus inference workloads. He noted a major shift in computing power allocation over the past three years, with a greater focus on AI inference rather than training.
Inference refers to how AI models process live data and generate responses, and it is becoming more resource-intensive due to advanced reasoning models like DeepSeek R1 and Alphabet’s Flash Thinking AI.
These models require significantly more computing power, which keeps demand for Nvidia’s GPUs high. Pichai’s perspective was reinforced by Meta CEO Mark Zuckerberg, who stated that a decline in training workloads doesn’t mean fewer GPUs are needed, as demand is shifting toward inference.
Alphabet’s $75 Billion AI Investment Signals Continued GPU Demand
Alphabet is making a massive $75 billion investment in data center infrastructure and AI chips in 2025, up from $52 billion in 2024.
This significant increase suggests that Nvidia’s GPUs remain an essential part of AI development, despite concerns over DeepSeek’s efficiency methods.
Nvidia Prepares for a Record Year in GPU Sales
Nvidia is expected to report a 112% revenue increase for its fiscal 2025, with AI data center sales driving nearly 88% of that growth. Analysts predict $196 billion in revenue for fiscal 2026, setting another record.
With the upcoming GB200 GPU, which can perform AI inference 30 times faster than previous models, Nvidia remains at the forefront of AI hardware innovation.
Despite short-term stock fluctuations, Alphabet’s and Meta’s AI strategies indicate that Nvidia’s long-term growth remains strong.
