It is reported that NVIDIA Corporation (NVDA.US) is developing AI inference chips, with OpenAI potentially becoming its largest customer.

According to reports quoted by media sources, chip giant NVIDIA Corporation (NVDA.US) plans to release a new processor specifically designed for artificial intelligence (AI) research companies like OpenAI and other clients to help them build faster and more efficient tools. Insiders revealed that NVIDIA Corporation is designing a new inference computing system. This new platform is expected to be released at the NVIDIA Corporation GTC developer conference in San Jose next month, and will integrate chips designed by the startup company Groq. Inference computing is a way for AI models to respond to user questions, and has become a focus of intense competition in the industry. Companies like Alphabet Inc. Class C and Amazon.com, Inc. have designed chips to compete with NVIDIA Corporation's flagship systems. The rapid development of automated programming in the tech industry has also created a demand for new chips that can efficiently handle complex AI-related tasks. Insiders have stated that OpenAI has agreed to become one of the biggest clients for this new processor, which is a significant victory for NVIDIA Corporation. As one of NVIDIA Corporation's largest clients, OpenAI has been seeking more efficient alternatives to NVIDIA Corporation chips in the past few months, and last month signed an agreement with the chip startup company Cerebras to provide itself with more options. Potential Challenges for NVIDIA Corporation GPUs NVIDIA Corporation has long dominated the GPU (graphics processor) market. Analysts estimate that NVIDIA Corporation controls over 90% of the GPU market share. A GPU is a processor that can execute billions of simple tasks simultaneously. NVIDIA Corporation's Hopper, Blackwell, and Rubin series GPUs are considered industry benchmarks for training large AI models, but they are also priced at a premium. However, since the rise of the AI boom, NVIDIA Corporation is facing limitations for its flagship products for the first time. As the market shifts from training to inference, some customers are pressuring NVIDIA Corporation to release chips that are more efficient in driving AI applications. Over the past year, as enterprises deploy AI agents and other tools, the demand for advanced computing power has shifted from training to inference. AI agents are systems that can autonomously perform tasks on behalf of users. Many companies building and operating AI agents have found that GPUs are costly, energy-intensive, and not the optimal choice for running models in practice. With the rapid rise of "agentic AI," NVIDIA Corporation is facing pressure to develop lower-cost, higher-efficiency inference chips. Last month, OpenAI signed a multi-billion-dollar computing collaboration agreement with Cerebras. Cerebras provides chips focused on inference, claiming that its chips are faster than NVIDIA Corporation GPUs. For NVIDIA Corporation, Alphabet Inc. Class C's in-house chip tensor processing unit (TPU) is also a significant competitor. In fact, Alphabet Inc. Class C is advancing plans to make TPUs replace GPUs. To strengthen its moat, NVIDIA Corporation agreed at the end of last year to pay $20 billion to obtain key technology licenses from Groq and hire its senior team, including founder Jonathan Ross. This is one of the largest "acqui-hire" deals in Silicon Valley history. Groq's chips use a different architecture than NVIDIA Corporation called "Language Processing Units," which are highly efficient in inference capabilities. However, NVIDIA Corporation has not yet publicly disclosed how they will utilize Groq's technology. Source: Finance Liaison Society; GMTEight Editor: Wang Qiujia.