The first AI chip of Openai enters the final design phase, plans for mass production for 2026: Report | Mint

OpenIE is finalizing the design for its first in-house chip over the next few months and wants to send it to the Taiwan Semiconductor Manufacturing Company (TSMC) for fabrication. The purpose of the first generation AI chip construction effort by the CHATGPT manufacturer is to reduce dependence on NVidia.

Allegedly, the process of sending the first design to a chip factory for construction is called ‘taping out’. A tape out process may take about six months to meet millions of expenses and to complete the tens. Additionally, there is no guarantee that the chip will first work in the first tape and if there is any problem, OpenII will have to diagnose it and repeat the entire process.

What will Openai do with your first generation AI chip?

If the initial tape-out goes smoothly, Openai will be able to allegedly produce its AI chips by 2026. Rentally, Openai’s AI chip is being made on TSMC’s 3 nanometer process technology and uses a commonly used systolic array architecture with high- bandwidth memory which is also employed for its chips by Nvidia.

The AI ​​chip will be able to run both training and AI models, but it will be prominently deployed to run a limited scale and AI model. Intily, the chip is likely to play a limited role with the infrastructure of Openai.

Why Openai is building your AI chip?

Generative AI chatbots such as Chatgpt, Gemini and Meta AI require a large number of chips to traditionally train their basic models. The powerful chips required for these functions are currently supplied by most NVidia WHCIH, which is about 80% of the market share.

Efforts by Openai will allow the company to take advantage of their interaction with chip suppliers including NVIDIA. Furthermore, after the success of its first chip, Openai is allegedly planned to develop more advanced processors with more advanced processor with alleged increase in each generation.

Is Openai ahead of other large technology?

Openai’s plan to send its first chip design to TSMC will stand ahead of the curve later this year, other chip designers will take time over years. In particular, other major technical companies like Satya Nadella led Microsoft and Mark Zuckerberg Meta has failed to produce satisfactory chipsets even after years of effort.

Meanwhile, the recent rise of Deepsac, which is built on a fraction of the cost and computing power of CHATGPT and other western AI chatbots, suggests that powerful big language models may require less chips in future to develop models. .