February 27, 2025 – OpenAI’s latest AI model, GPT-4.5, is facing a delayed launch due to ongoing GPU shortages. CEO Sam Altman recently acknowledged that the demand for AI computing power has exceeded expectations, making it difficult to roll out the model at scale.
GPT-4.5: A More Human-Like AI Model
Breakthrough Conversational Abilities
Sam Altman described GPT-4.5 as the first AI model that genuinely feels like talking to a thinking being.
He emphasized that the AI delivers surprisingly insightful responses, unlike any previous model.
Not Designed for Benchmark Domination
GPT-4.5 is not a model optimized for test scores, meaning it may not significantly outperform previous models in standard AI benchmarks.
Instead, OpenAI focused on a more nuanced, engaging, and intelligent user experience.
GPU Shortages Delay Rollout
Limited Availability for Pro Users
GPT-4.5 is currently restricted to Pro-tier members, who pay $200 per month.
OpenAI initially wanted to launch GPT-4.5 for both Plus and Pro users, but supply chain constraints forced them to prioritize Pro members.
Scaling Challenges
Altman acknowledged that AI demand has grown faster than anticipated, making it difficult to accurately predict computing needs and prevent GPU shortages.
OpenAI expects to add tens of thousands of GPUs in the coming weeks and hundreds of thousands more in the near future.
AI Industry’s Growing GPU Crisis
The AI boom has exacerbated the global GPU shortage, particularly for high-end chips like NVIDIA’s H100 and upcoming Blackwell B200.
Tech giants such as Meta, Microsoft, Google, and Amazon are aggressively expanding their AI infrastructure, further straining GPU supply.
xAI, Elon Musk’s AI startup, is also stockpiling GPUs, planning to scale up to 1 million units.
With limited supply and skyrocketing demand, access to high-performance AI compute has become a critical bottleneck for the industry.
