Nvidia (NASDAQ: NVDA) continues to lead in the AI infrastructure market as demand for its GPUs grows significantly. Recent forecasts indicate that AI models, such as xAI’s Grok-3, will require training on 100,000 GPUs, sharply higher than previous models. Meta Platforms also anticipates its Llama 4 model needing ten times the computing power of its predecessor, Llama 3, which was trained on 16,000 GPUs.
CEO comments from Alphabet and Meta emphasize the risk of underinvesting in AI, with spending on AI projects expected to rise over the next five to ten years. Currently trading at a forward P/E ratio of about 29 and a PEG ratio below 0.8, Nvidia is considered undervalued, presenting a potential buy opportunity for investors.
However, concerns persist about the future demand for GPUs. As the focus shifts from AI training to AI inference, spending could decline if AI models reach adequate performance levels. Despite this, Nvidia is well-positioned to maintain its growth trajectory, supported by its dominant market presence and ongoing innovation.
