OpenAI Will Develop Its Own AI Chips

With the rise of AI technologies, some chip manufacturing companies had a very profitable year. Still, OpenAI is considering developing its own AI chips because the market is already facing a shortage. In addition to the AI chatbot race, a new one arises. In this article, let’s discuss the possibilities that will be given to OpenAI with unlimited artificial intelligence chips and the challenges if they don’t do so. We will also speak about the AI chips field and the most profitable year for NVIDIA.

‍

Addressing the Chip Shortage: OpenAI's Vision

The idea of developing AI chips was born last year. Various options existed for solving the problem of a shortage of such expensive parts. Instead of buying them directly from NVIDIA, the best option included creating its chip, working more closely with other chipmakers, and diversifying suppliers. 

OpenAI CEO Sam Altman has made chip acquisition a priority for the company. He has publicly complained about the shortage of GPUs in a market dominated by NVIDIA, which controls more than 80% of the chips suitable for AI. Speaking about them, NVIDIA earned a record $13.5 billion and $6.2 billion in net income for the quarter, up 101% year-over-year, all thanks to the AI boom.

There are two main problems with trying to get more chips: chip shortages and high costs. Running ChatGPT is very expensive for the company - each request costs about 4 cents. If ChatGPT queries grow to at least a tenth of the number of Google search queries, it will require approximately $48.1 billion to purchase GPUs and about $16 billion in running costs per year to keep it running. Additionally, let’s not forget about us, the customers. If you have faced some troubles with the ChatGPT network, long answers, or unpredictable errors, you may be sure - chips are the reason.

The acquisition of a chipmaker could speed up the process of creating its processor. According to the insider, OpenAI was considering steps to evaluate and verify a potential acquisition target. It is not specified which company OpenAI was going to acquire. According to the latest news about AMD purchases on Nod.ai, there is no time to wait.

OpenAI's main sponsor, Microsoft, is also developing a special AI chip that OpenAI is testing. It is unclear whether OpenAI plans to continue working on the chip. Even if the company allocated all the necessary resources for this task, it would not guarantee success. In the case of success with chips, there will be no limits for running the current versions of ChatGPT-3 and 4, but also for developing ChatGPT-5.

To Strengthen Its Ecosystem of AI Tools, AMD Purchases Nod.AI

As part of its ongoing efforts to create an ecosystem of AI development tools, libraries, and models around its hardware, AMD purchased Nod.ai, a provider of open-source AI software.

‍

‍

According to CNBC, the deal is anticipated to be finalised this quarter. The specifics of the transaction were kept under wraps by AMD. The acquisition of Nod.ai, according to AMD SVP Vamsi Boppana, would "significantly" improve AMD's capacity to offer clients "software that allows them to easily deploy highly performant AI models tuned for AMD hardware."

Nod.ai's first goal was to create motion-tracking and gesture-recognition hardware for video games. However, it eventually shifted to providing modules for AI model tools, reducing the requirement for AI developers to manually tune and deploy AI models to run across the data center and edge computers, including AMD-powered machines.

Nod.ai will join the AMD AI department, which the company established earlier this year and currently employs 1,500 developers. By the end of the year, the group is anticipated to grow by about 300 personnel as AMD tries to overtake Nvidia in the rapidly expanding AI chip industry.

The Alternative Way to Solve the Lack of AI Chips Issue

Some readers may wonder: What is the issue about? They’re earning billions, just make more chips, and that’s all. The true issue is that the lack isn’t something new. First, the U.S. has restricted China's ability to produce semiconductors amid tensions over trade, security, and high-tech dominance. Probably the only thing that unites AI competitors on one side is the lack of chips. The availability of graphics processing units (GPUs) is mentioned as a potential risk issue for investors in Microsoft's annual report. In May, OpenAI CEO Sam Altman testified before the US Senate, claiming that the company's chatbot tool was having trouble with the volume of requests being made of it by users. The fewer individuals who utilize the tool, the better, according to Altman, because we are so short on GPUs. Still, we have an alternative: analogue computer chips, which provide a solution to the massive and expanding energy demand of AI, by running artificial intelligence (AI) voice recognition models 14 times more effectively than conventional chips. The company that created the equipment, IBM Research, declined to speak for an interview and made no comments. However, researchers assert in a report describing the work that the analogue chip can lessen AI development obstacles. 

Conclusion

The AI race began a long time ago. Nowadays, we can see its impact on our world and other fields. With the most profitable year in its history, NVIDIA isn’t going to stop dominating the market for AI chips. AMD, on the other hand, will use every possibility to bite this piece of cake. With the idea of OpenAI developing their chips, we will see an interesting situation in the AI field. Microsoft is actively developing their own AI tools, and it seems, even with the recent problems, they’re breathing in the back of Chat-GPT.

‍