Nvidia, a leading name in the artificial intelligence (AI) sector, holds a predominant market share, producing approximately 80% of the AI chips that companies rely on for training and running models.

However, the recent entry of Chinese firm DeepSeek into the market has prompted key players like OpenAI, Meta, Apple, Microsoft, Google, and Amazon to reconsider their dependence on Nvidia's chips, which have led to high costs and limited computing supply.

OpenAI, a central player in the generative AI revolution, is reported to be developing its own AI chip to reduce reliance on Nvidia.

According to a Reuters report, OpenAI has been progressing in the design of this chip, aiming for completion in the upcoming months.

The design will subsequently be sent to Taiwan Semiconductor Manufacturing Company (TSMC) for experimental production and testing, which is estimated to cost tens of millions of dollars and typically takes around six months, although expedited production is an option.

The company aims to have its custom AI chip ready for mass production by 2026. Initially, the chip’s applications will be limited, primarily focusing on running models rather than extensive training.

However, OpenAI plans to develop more advanced processors in the future, viewing these efforts as strategic tools for negotiating with Nvidia and other chip suppliers.

The announcement of DeepSeek's Model R1 in January marked a significant shift in industry assumptions regarding AI model development.

DeepSeek claims this model was developed at a fraction of the cost compared to similar models from OpenAI and Google, prompting many companies to reassess the necessity of relying on Nvidia's high-priced computing power.

This has ignited interest in more affordable alternatives that can still deliver satisfactory performance for developing next-generation models.

In response to these market dynamics, Meta is reportedly in discussions to acquire South Korean AI chip startup FuriosaAI, aiming to establish its own alternative to Nvidia's offerings, with a potential deal expected to finalize soon.

Last year, Meta unveiled a new generation of AI chips known as the Meta Training and Inference Accelerator (MTIA), aimed at traditional AI tasks rather than generative AI.

Microsoft is also engaged in the development of various chip products intended to enhance AI model performance and data center functionalities.

Amazon has been working on its ecosystem of AI chips for its expansive cloud data centers, recently unveiling Trainium3, its latest training and running chip based on technology acquired from Israeli firm Annapurna in 2015.

Google's approach involves integrating AI models into its AI chip development process led by the company's DeepMind division to automate the development procedures.

Apple is also making significant strides in chip development, with plans underway to produce its first server chip aimed at AI applications, collaborating with Broadcom on critical network technology components.

This chip is also anticipated to enter mass production in 2026, with plans already in place for fabrication using advanced TSMC manufacturing technologies.

Nvidia's substantial market share poses questions about the success of these competing efforts.

The tech industry's investments highlight the confidence that some initiatives will yield viable outputs.

The ongoing evolution instigated by DeepSeek may serve to further challenge Nvidia's prevailing position in the near future.