According to a co-founder at CoreWeave, a prominent provider of GPU-accelerated cloud computing services, even graphics processing units (GPUs) that are not the absolute latest generation are experiencing a notable surge in both price and demand. This trend is primarily driven by the escalating need for powerful hardware to support the rapidly expanding artificial intelligence (AI) server infrastructure.
The Nvidia A100 GPU, for instance, despite being succeeded by newer generations like the H100, stands out as a prime example of a ‘previous generation’ chip that remains highly sought after. Its continued relevance underscores the immense pressure on the supply chain for AI-capable hardware, as companies globally scramble to acquire sufficient processing power for developing, training, and deploying advanced AI models.
This market dynamic suggests that the sheer volume of compute power required for modern AI workloads is so vast that older, yet still highly capable, GPUs are being absorbed into the market at elevated prices. It reflects a period of intense investment and expansion within the AI sector, where every available piece of suitable hardware contributes to meeting the insatiable demand for computational resources.