OpenAI making a new inhouse chip for their AI models

The leading company in AI development and enhancement OpenAI is pushing the plans to create an in-house chip to cut off NVIDIA completely. Let's take a technical look at this.

NVIDIA has been one of the companies that has been benefiting a lot from the recent AI trends. The AI needs fast parallel processing units which GPU can achieve only to get an amazing result. As NVIDIA is the leading company in GPU tech with around 80% market share they have an advantage. This has made their sales skyrocket and their share prices have seen a great all-time high. This has made them the only company to have such great hardware that can run any AI model without any problem. To decrease their operational cost and get into a market that will surely evolve in the future OpenAI is also taking steps. This step can be taken because of being dependent on a single supplier for their hardware supplies or something more radical.

The father of the ChatGPT is finalizing the plans for their in-house chip, a team of 40 people is being led by Richard Ho in collaboration with Broadcom. The team is fairly small when compared to companies like Google. Rumors tell that they have plans to ship to Taiwan Semiconductor Manufacturing Co within a month. It will take around 6 months to finalize the chip and a ton of money. If they become successful in making the chip this will make them a direct competitor to NVIDIA and give the market an alternative to the NVIDIA chips. It will be a very great success for OpenAI if they are successful in doing so because big tech giants like Microsoft have failed to deliver a great chip. OpenAI has a great percentage of success as they have worked with GPU very deeply which can help them achieve good quality hardware. But it is more likely that the chip will be commercially used by OpenAI only and won't be sold for consumers as these chips will be specifically for running their AI models and the architecture of these chips can be custom-designed to only support their AI models.

Moreover, Deepseeks new AI model has also created a lot of hype in the market amid their claim that they don't need as much power as OpenAI claims to run a big language model.

We can get a clear picture of where OpenAI is heading by the end of this year as by then they will receive the results from Taiwan about their chip designs.