The popularity of ChatGPT's Studio Ghibli-inspired images has led to significant challenges for the company. According to the CEO, over 1 million users joined within just five days of the launch of GPT-4o, the model powering these images. Remarkably, the company hit the milestone of 1 million users within an hour on March 31. This surge in traffic has strained the company's resources, resulting in delays in new model releases, slower response times, and even occasional service outages. Sam Altman shared updates via X (formerly Twitter), assuring users that the team is actively working to tackle the current situation. However, he warned that users should expect delays in new releases from OpenAI.
Altman emphasised the importance of understanding that some features may not work as intended, and service may be slower while they navigate these ongoing capacity challenges. He acknowledged that these issues are a natural part of their efforts to enhance their systems.
GPU requirements
Regarding GPU requirements, Altman mentioned their ongoing efforts to scale their AI infrastructure. In a follow-up post, he stated that the team is pushing to get things back on track as quickly as possible. He also encouraged anyone with GPU capacity available in increments of 1,00,000 to reach out immediately.
GPT-5 launch
Originally, the company hinted at launching GPT-5 in February, but Altman’s recent comments suggest that this release may be postponed. The ongoing GPU scarcity raises concerns about whether this will be a long-term issue for the company, which now boasts 500 million weekly users of ChatGPT.
On a positive note, OpenAI announced this week that it has secured an additional $40 billion in funding, even though the San Francisco lab has spent billions on GPUs, which consume immense amounts of electricity.