OpenAI Has Received More Nvidia H100 Chips
-
In a recent interview with the Financial Times, OpenAI CEO Sam Altman stated that the company has received the highly anticipated Nvidia H100 chip order and expects to receive more soon. He added, 'Next year already looks better in terms of acquiring more chips.'
OpenAI and other artificial intelligence companies widely use Nvidia's latest model chips to train their models. This year, the level of attention on AI chatbots like OpenAI's ChatGPT and Google's Bard appears comparable to the focus on Nvidia's $40,000 H100 chip.
OpenAI's move to acquire more chips from Nvidia suggests that more complex AI models will be ready in the near future, surpassing the capabilities of current chatbot versions.
Strong Demand for NVIDIA H100 Chips
Generative AI systems require training on massive datasets to generate complex responses, which demands substantial computational power. Nvidia's H100 chip is specifically designed for generative AI and operates significantly faster than previous chip models. Harvard Business School professor Willy Shih notes that the more powerful the chip, the faster it can process queries.
Behind the scenes, startups, chip competitors like AMD, and major tech companies such as Google and Amazon have been dedicated to developing more efficient specialized chips for AI applications to meet market demand—but so far, no company has been able to surpass NVIDIA.
This intense demand for a company's specific chips has sparked a buying frenzy at Nvidia, with not only tech companies rushing to acquire these highly sought-after chips—governments and venture capital firms are also actively participating. But if OpenAI can secure its orders, this may indicate that the situation is finally shifting, and the flow of chips to AI companies is improving.
Although Nvidia still holds a dominant position, SymphonyAI CEO Prateek Kathpal stated last week that while the company's AI applications currently run on Nvidia chips, they are also in discussions with AMD and Arm regarding their technologies.
What Does More Chips Mean for OpenAI?
OpenAI's growing chip inventory signifies several things. H100 chips will support the company's next-generation AI model GPT-5, which Altman stated is currently under development. This new model will require more data for training, sourced from publicly available information and the company's proprietary data. In an interview with the Financial Times, he mentioned that GPT-5 may be more complex than its predecessor, though he added that it remains unclear what it can do that GPT-4 cannot.
Altman did not disclose the release timeline for GPT-5. However, GPT-4 was released just eight months ago, following its predecessor GPT-3 in 2020, highlighting the rapid development cycle.
The increased procurement of chips also indicates that the company is moving closer to creating Artificial General Intelligence (AGI), an AI system that is essentially capable of performing any task a human can do.