OpenAI Teams with AMD to Boost AI Power, Challenge Nvidia

OpenAI has formed a five-year strategic partnership with Advanced Micro Devices (AMD) to supply processors for its AI data centers. The agreement, set to begin in 2H26, will equip OpenAI with thousands of Instinct MI450 GPU units. This move will bolster the infrastructure for its Stargate project and represents a notable development in the competitive AI hardware market.

The collaboration is designed to accelerate the development and deployment of large-scale AI systems. "AMD's leadership in high-performance chips will allow us to accelerate progress and bring the benefits of advanced AI to everyone faster," says Sam Altman, CEO, OpenAI.

The agreement occurs amid unprecedented demand for computing power in the AI industry. This partnership allows OpenAI to diversify its hardware suppliers, mitigating supply chain risks and fostering a more competitive market. For AMD, it represents a direct challenge to Nvidia's established dominance in the market for AI processors.

OpenAI is set to receive the equivalent of 6GW of computing power, an amount sufficient to supply 4.5 million homes. The market reacted immediately to the news, with AMD's stock price increasing by 24% in pre-market trading, reflecting investor confidence in the collaboration's potential.

The financial structure grants OpenAI an option to acquire up to 160 million AMD shares, representing about 10% of the company, at US$0.01 per share. This option is contingent upon the achievement of specific, pre-established objectives. 

The technical rationale for the decision is clear. Lisa Su, Chief Executive Officer, AMD, said previously that the MI450 series is engineered to outperform competing offerings from Nvidia. She also announced that the 2025 generation of processors, the Instinct MI350 series, already surpasses Nvidia's Blackwell chips in performance.

This performance data attracts partners like OpenAI and introduces greater variety to the AI sector, reducing its dependency on a single hardware provider. In June, Altman said that his team is conducting work on the MI300X chip, which AMD launched last year. "But we are extremely excited for the MI450, the memory architecture is excellent for inference. It can also be an incredible option for training," Altman said.

This alliance is a key component of OpenAI's broader strategy to secure the necessary infrastructure for its Stargate project. The deal complements a previously announced investment from Nvidia of up to US$100 billion, which includes a commitment to supply at least 10GW of computing power.

Throughout the year, OpenAI has secured other agreements with corporations like Oracle, SoftBank, and Broadcom to expand its facilities in the United States. It has also formed alliances with Samsung and SK Hynix to establish Stargate locations in South Korea.