"Why OpenAI Should Acquire Graphcore"

Since the end of 2022, artificial intelligence has been a fiercely competitive field, and OpenAI has been at the forefront. After achieving great success in software, the organization led by Sam Altman is now seeking to enter the hardware industry.

Against the backdrop of the much-discussed chip shortage in the industry, OpenAI has announced its plan to develop its own artificial intelligence chips, following in the footsteps of Microsoft, Google, Amazon, and Apple. However, despite the high demand for AI hardware, most manufacturers have struggled to rise to the occasion, apart from NVIDIA.

A good example is Graphcore. This startup signed an agreement with Microsoft in 2019 to purchase processors and is now in urgent need of funding. In less than four years, Graphcore has become a unicorn, but its revenue in 2022 was $2.7 million, a 46% decrease from 2021. Losses also soared by 11% to $204.6 million, which the company attributed to a "decline in hardware sales to key strategic customers."

New customers on the horizon

One of the "key strategic customers" mentioned by Graphcore is likely Microsoft. Since Microsoft started supporting OpenAI, the company has predominantly used NVIDIA chips for all its work and no longer uses Graphcore chips in its cloud computing systems, resulting in reduced funding for the chip manufacturer.

Since its last funding round in 2020, Graphcore has been valued at $2.5 billion. Since then, the UK-based company has failed to raise any additional funds. The company has also requested the government to include chip manufacturers in the recent Brunel AI project, which aims to build a new AI supercomputer with $1.1 billion in funding. However, the deal has not been finalized. If the government does accept the deal, Graphcore could once again become a competitor to NVIDIA.

On the other hand, there's OpenAI. According to a recent report by The Information, Altman stated that the company's annual revenue will reach $1.3 billion. Now, the company is planning to manufacture its own AI chips, which could be a significant move for OpenAI to realize its dream of manufacturing its own AI chips and potentially save Graphcore in the process.

OpenAI currently has the financial strength to take strategic actions in the AI hardware field, as evidenced by its revenue growth. The company has considered acquiring a chip manufacturing company or manufacturing its own chips. By making an early investment in Graphcore, OpenAI can not only ensure a reliable supply of AI chips but also pave the way for potential transformative partnerships and open up the market for other participants.

What about Microsoft?

As mentioned earlier, Microsoft, one of OpenAI's main supporters, is also developing its own AI chips. It has been reported that Microsoft and OpenAI employees have been secretly testing these chips, codenamed Athena, since April.

In addition, AI chip startup d-Matrix has raised $110 million with support from Microsoft. Although the company is not a direct competitor to NVIDIA chips, it still provides an affordable alternative for the AI model inference part for Microsoft.

Regardless, these news reports indicate that the company behind ChatGPT may be striving to avoid using NVIDIA GPUs to build AI models.

However, when it comes to OpenAI's entry into the hardware market, the company seems to be diverging from Microsoft and NVIDIA in certain aspects. According to multiple reports, OpenAI plans to create its own "iPhone of AI" and raise funds from SoftBank.

While it may not be a phone, a new product is certainly in the works. Therefore, acquiring Graphcore could be the right choice for the company to have its own in-house capabilities in the near future.

Meanwhile, almost every company, including AMD, Intel, and Cerebras, is striving to have what NVIDIA already possesses and is developing their own AI chips. Given that OpenAI is also evaluating many other targets, it cannot be guaranteed that it will proceed with this deal.