TECH NEWS – The latest partnership of the company known for ChatGPT will be expensive.
OpenAI‘s partnership with Nvidia focuses on Vera Rubin and the inference capabilities of the upcoming Nvidia-Groq solution. OpenAI is entering financing agreements with infrastructure partners across the AI industry. The AI giant recently announced it will raise $110 billion in fresh capital with support from Nvidia, SoftBank, Amazon, and others. According to OpenAI, these investments are essential to maintaining operations in the AI sector and are one way for the company to secure the necessary computing capacity.
The Wall Street Journal reports that Nvidia will unveil its Groq-focused “processor” at this year’s GTC. OpenAI will be the largest customer for the upcoming solution, an interesting decision. In connection with Nvidia‘s latest investment in OpenAI, it was revealed that OpenAI will use 3 GW of dedicated inference capacity, likely from Nvidia‘s product to be unveiled in March. Earlier rumors suggested that inference has been one of OpenAI‘s main concerns lately, and that the company was dissatisfied with Nvidia‘s solutions for handling inference.
OpenAI reportedly held talks with Cerebras and Groq to ensure optimal performance for latency-sensitive workloads. However, it now appears that OpenAI will stick with Nvidia, indicating that Groq‘s upcoming LPU-based solution is promising enough for the AI giant to commit 3 GW of capacity to it. We expect the agreement between Nvidia and Groq to result in a hybrid computing configuration.
We currently expect NVIDIA to make significant announcements at this year’s GTC. These announcements will focus on Vera Rubin, the potential next-generation Feynman, and a solution built around Groq. These announcements will further advance Nvidia‘s trend. According to their recently announced financial results, the gaming GPU segment accounted for only about one-tenth of data center revenue.
