On October 28, media reports revealed that OpenAI, a leading player in the global AI field, is planning to buy AI chips from Google. For many years, OpenAI has been one of the most important buyers of NVIDIA graphics processing units (GPUs), primarily depending on these chips to train massive AI models and perform complex inference tasks.
As industry demand for computing power continues to soar, companies are actively searching for more cost-effective solutions that can efficiently run AI models. Google’s self-developed Tensor Processing Unit (TPU) has steadily gained traction in recent years and began selling directly to select external customers this year. Looking back at the evolution of their collaboration, over the past 16 months there have been three clear milestones: in June 2025, OpenAI first started renting computing resources from Google; in July of the same year, it publicly stated that it was conducting early testing on Google’s TPUs but had no immediate plans for large-scale deployment; and now, the two sides have officially moved forward with concrete procurement plans. In addition, OpenAI is currently recruiting engineers tasked with adapting and optimizing its models for use on TPUs—such low-level model adaptation often serves as an important precursor to large-scale procurement projects.
Looking back at its computing expansion trajectory, Google’s TPU played a significant role during OpenAI’s early startup phase. Subsequently, OpenAI fully transitioned to the NVIDIA ecosystem and gradually integrated chips from various vendors, including AMD, Cerebras, and AWS.
For the broader industry landscape, this shift sends a clear signal: as its most prominent customer explores new computing avenues beyond a single GPU-based approach, NVIDIA’s exclusive dominance in orders is gradually diminishing. According to data from the authoritative research firm TrendForce, NVIDIA’s market share in the AI server processor segment is expected to decline from 69.6% in 2025 to 64% in 2026, aligning closely with current trends toward technological and supply-chain diversification.