OpenAI, the artificial intelligence research organization behind the popular ChatGPT language model, has hit a roadblock in the rollout of its latest model, GPT-4.5. In a recent post on X, CEO Sam Altman revealed that the company is facing an unexpected shortage of graphics processing units (GPUs), forcing them to stagger the release of the new model.

According to Altman, GPT-4.5 is a "giant" and "expensive" model that requires "tens of thousands" more GPUs to support additional ChatGPT users. As a result, the rollout will be limited to subscribers of ChatGPT Pro starting Thursday, followed by ChatGPT Plus customers next week. This phased release is a departure from OpenAI's usual approach, and Altman acknowledged that it's not ideal, but necessary due to the current GPU shortage.

The enormous size of GPT-4.5 is likely a contributing factor to the high cost of using the model. OpenAI is charging $75 per million tokens (~750,000 words) fed into the model and $150 per million tokens generated by the model. This pricing structure is significantly higher than OpenAI's workhorse GPT-4o model, with some critics calling it "unhinged."

Altman attributed the GPU shortage to OpenAI's rapid growth, which has led to unpredicted demand for their products. This is not the first time the company has faced computing capacity issues, with Altman previously citing a lack of computing power as a delay in their product development. To address this, OpenAI is working on developing its own AI chips and building a massive network of datacenters.

The delay in GPT-4.5's rollout highlights the challenges of scaling AI models to meet growing demand. As AI technology continues to advance, companies like OpenAI will need to find innovative solutions to meet the computing demands of their products. The development of custom AI chips and datacenters could be a key step in addressing these challenges and enabling the widespread adoption of AI technology.

In the meantime, OpenAI's GPU shortage serves as a reminder of the infrastructure challenges that underlie the AI industry's rapid growth. As the company works to add tens of thousands of GPUs to its network, users will have to wait patiently for access to GPT-4.5. The success of OpenAI's efforts to scale its computing capacity will have significant implications for the future of AI development and deployment.