OpenAI CEO Sam Altman is currently in negotiations to secure a multi-billion dollar investment for creating an international Artificial Intelligence (AI) chip production network. This significant investment will address key challenges in AI development such as the shortage of high-performance processors for AI models like ChatGPT and DALL-E. In this ambitious endeavor, Altman has collaborated with leading global chip manufacturers whose identities remain confidential.
In the context of current industry trends, the need to develop and manufacture powerful AI chips for models like ChatGPT and DALL-E, capable of responding to requests and creating images, is undeniable. NVIDIA, with a market capitalization reaching $1 trillion for the first time last year, essentially monopolizes the market thanks to its popular H100 graphic processors. These processors are used in training advanced AI models such as GPT-4, Gemini, Llama 2.
Given the intensifying competition for the production of high-performance AI chips, securing necessary production capabilities ahead of time is crucial. In this light, Altman and his team are behind the scenes negotiating with investment giants like SoftBank Group and Abu Dhabi’s G42 for financing this gigantic project.
![New Microsoft Azure Maia 100 AI processor (Image source: Microsoft)](http://davesipaq.com/wp-content/uploads/2024/01/1705765784_383_sem-altman-planiruet-sozdat-mezhdunarodnuyu-set-po-proizvodstvu-ii-chipov.jpg)
![New Microsoft Azure Maia 100 AI processor (Image source: Microsoft)](http://davesipaq.com/wp-content/uploads/2024/01/1705765784_383_sem-altman-planiruet-sozdat-mezhdunarodnuyu-set-po-proizvodstvu-ii-chipov.jpg)
Not only OpenAI, but other IT industry heavyweights such as Microsoft, Amazon, and Google too, are actively working on creating their AI chips. OpenAI investor Microsoft announced in November last year the development of its first proprietary AI chip for training AI models. Following this, Amazon debuted an updated version of its Trainium chip. Google’s chip development team uses AI DeepMind on Google Cloud servers for designing their AI processors, including Tensor Processing Units (TPU).
AWC, Azure and Google also actively use NVIDIA’s H100 processors. In a recent interview with The Verge, Meta✴ CEO Mark Zuckerberg informed journalist Alex Heath that by year’s end, Meta✴ plans to possess more than 340,000 H100 processors in its strategy of developing potent AI (Artificial General Intelligence — AGI).
![NVIDIA GH200 Grace Hopper Superchip (Image source: NVIDIA)](http://davesipaq.com/wp-content/uploads/2024/01/1705765785_166_sem-altman-planiruet-sozdat-mezhdunarodnuyu-set-po-proizvodstvu-ii-chipov.jpg)
![NVIDIA GH200 Grace Hopper Superchip (Image source: NVIDIA)](http://davesipaq.com/wp-content/uploads/2024/01/1705765785_166_sem-altman-planiruet-sozdat-mezhdunarodnuyu-set-po-proizvodstvu-ii-chipov.jpg)
Amid skyrocketing capitalization, NVIDIA remains in the race for innovations, having already announced their new GH200 Grace Hopper chips. This move will solidify its already dominant market position. However, AMD, Qualcomm, and Intel are also keeping pace by releasing processors designed to support AI models on laptops, smartphones, and other devices.