OpenAI creates its own AI chip to reduce dependence on Nvidia
Kyiv • UNN
OpenAI plans to introduce its own AI processor in 2026. The project is being implemented in cooperation with Broadcom for internal use.
The ChatGPT developer plans to introduce its own artificial intelligence processor next year to meet the rapidly growing demand for its resources and reduce its dependence on Nvidia. This is reported by UNN with reference to Financial Times.
Details
OpenAI has announced the development of its chip for working with AI models. The project is being implemented in cooperation with the American giant Broadcom, and the first deliveries are expected in 2026. According to informed sources, the agreement provides for an order worth about $10 billion, which has become one of the largest contracts in the semiconductor industry.
Although Broadcom does not officially disclose the names of its customers, insiders confirm that OpenAI is the company's fourth major AI chip customer. Broadcom CEO Hock Tan stated that the new contract significantly increases growth forecasts and provides "immediate and massive demand."
OpenAI's decision follows the strategy of Google, Amazon, and Meta, which are already creating their own specialized processors to optimize AI operations. The new chips, called XPUs, are expected to compete with traditional GPUs from Nvidia and AMD and occupy an increasing share of the AI infrastructure market.
It should be noted that OpenAI plans to use these processors exclusively for internal needs - primarily for training and launching its models, including the latest generation GPT-5, the demand for which has already sharply increased the load on the infrastructure.
Recently, the company's CEO Sam Altman emphasized that OpenAI intends to double its computing power within five months to meet new challenges.
Meanwhile, Broadcom is demonstrating record financial results: after the report was published, the company's shares rose by another 4.5%. According to HSBC analysts, its business of producing specialized chips could grow faster than Nvidia's GPU segment by 2026.
The Financial Times notes that despite this, Nvidia still maintains a leading position in the AI hardware market, remaining the main supplier for the largest technology corporations. However, OpenAI's move could be the first signal of large-scale changes in the balance of power in the artificial intelligence market.