OpenAI is no longer "All In" NVIDIA Corporation (NVDA.US) and is instead investing in Alphabet Inc. Class C (GOOGL.US) TPU to break the monopoly of chips!

date
30/06/2025
avatar
GMT Eight
According to reports, OpenAI has begun using AI chips manufactured by Google to build products, including ChatGPT, marking its first major shift away from solely relying on NVIDIA chips.
According to a report from The Information, the artificial intelligence startup company OpenAI, invested in by Microsoft Corporation (MSFT.US), has started using AI chips made by Alphabet Inc. Class C (GOOGL.US) to build products including ChatGPT, marking its first major shift away from solely relying on NVIDIA Corporation (NVDA.US) chips. The collaboration with Alphabet Inc. Class C's tensor processing unit (TPU) signifies OpenAI's substantial adoption of non-NVIDIA Corporation chips for the first time, initiating a diversified supplier strategy. Previously, the company had long relied on NVIDIA Corporation chips for both training AI models and running inference computations (the process of running the model after training), and was also one of NVIDIA Corporation's largest purchasers of graphics processing units (GPU). OpenAI expects that using Alphabet Inc. Class C's cloud-rented TPUs will help lower costs related to inference. The Information suggests that this could drive TPUs to become a cheaper alternative to NVIDIA Corporation GPUs. Earlier this month, OpenAI announced plans to access Alphabet Inc. Class C's cloud services to meet its growing computational needs, marking a surprising collaboration between two major competitors in the field of artificial intelligence. Morgan Stanley also released a research report supporting Alphabet Inc. Class C, stating that if the agreement is confirmed, it will demonstrate Alphabet Inc. Class C's confidence in its long-term search business position and drive the development of Alphabet Inc. Class C's cloud, with a valuation reaching over 18 times. For Alphabet Inc. Class C, this collaboration comes at a time when they are expanding the external use of their in-house tensor processing units (TPU), which were previously used for internal projects and have now attracted tech giants like Apple Inc. (AAPL.US) and ChatGPT competitor companies founded by former core members of OpenAI such as Anthropic and Safe Super intelligence. However, it is reported that Alphabet Inc. Class C did not rent out its most powerful TPU model to OpenAI, indicating their plans to use the most advanced version to support internal projects, including their in-house Gemini large language model.