Reported from the source

Quick summary: OpenAI is aggressively diversifying its chip suppliers, committing billions to new deals with companies like Cerebras, AMD, and Broadcom, alongside its long-standing relationship with Nvidia. This strategy aims to secure vast processing power for its expanding AI operations, with total infrastructure commitments exceeding $1.4 trillion last year as it pursues a historically aggressive expansion plan.

OpenAI has recently announced a series of significant chip deals to diversify its infrastructure and meet anticipated demand for its AI technology. On Wednesday, the company unveiled a $10 billion agreement with Cerebras to deploy 750 megawatts of its AI chips through 2028. Cerebras’ wafer-scale chips are touted to deliver responses up to 15 times faster than GPU-based systems. This follows earlier major commitments. In 2025, Nvidia pledged $100 billion to support OpenAI in building and deploying at least 10 gigawatts of Nvidia systems, equivalent to 4 million to 5 million GPUs, with the first phase expected in the second half of this year. However, Nvidia noted in November that there is “no assurance” the agreement will progress beyond an announcement to an official contract stage. In October, OpenAI announced plans to deploy six gigawatts of AMD’s GPUs over multiple years. As part of this deal, AMD issued OpenAI a warrant for up to 160 million shares of AMD common stock, potentially a 10% stake. The first gigawatt of AMD chips is slated for rollout in the second half of 2026. Later that month, OpenAI and Broadcom revealed a collaboration to design and deploy 10 gigawatts of custom AI accelerators (XPUs). Broadcom aims to begin deploying these systems by the second half of this year, with project completion targeted for the end of 2029, though Broadcom expects minimal revenue from the partnership in 2026. While these deals solidify new partnerships, some major players have been left out of direct chip supply agreements. Amazon Web Services (AWS) signed a $38 billion cloud deal with OpenAI in November, and Amazon is reportedly in talks to invest over $10 billion, which could potentially include the use of Amazon’s AI chips. Google Cloud also provides computing capacity to OpenAI, but OpenAI stated in June it has no plans to use Google’s in-house Tensor Processing Units (TPUs). Intel, which reportedly declined an early investment opportunity in OpenAI, announced a new data center GPU, “Crescent Island,” in October, with customer sampling expected in the second half of 2026.

Source: www.cnbc.com