In November 2023, OpenAI CEO Sam Altman paused new signups for ChatGPT’s paid subscriptions due to overwhelming demand, which had surged to 100 million weekly active users post its first developer conference on November 6. This rapid increase in interest forced OpenAI to turn away potential subscribers as the company’s GPU capacity could not meet the requirements.
The AI industry is grappling with a significant compute bottleneck, as demand for high-performance GPUs has outpaced supply, impacting the ability to scale AI models effectively. Nvidia, a key player in this sector, saw its shares surge nearly 1,000% as its advanced chips became essential for AI operations, while Broadcom benefited similarly with a 600% increase by providing necessary networking components for data centers.
These developments underscore that while AI demand is high, supply constraints—particularly in computing power and components—are critical bottlenecks currently shaping the industry. As AI hyperscalers prepare for earnings reports, potential new constraints in metals, electricity, and memory may emerge, raising questions about the sustainability of the current AI boom.
