OpenAI vs. Nvidia: Who Will Dominate the Future of AI Chip Technology?

OpenAI CEO Sam Altman is actively seeking to raise between $5 trillion and $7 trillion in an ambitious effort to enhance the global capacity for AI chip production. According to sources from The Wall Street Journal, Altman is in discussions with the U.A.E government, SoftBank, and several other key investors. The goal is to collaborate with Taiwan Semiconductor Manufacturing Company (TSMC) to manufacture and operate these AI chips.

Recently, Altman emphasized the pressing need for expanded AI infrastructure, stating that the world requires more fabrication capacity, energy solutions, data centers, and other essential components than current plans account for. He highlighted that building extensive AI infrastructure and establishing a resilient supply chain are vital for maintaining economic competitiveness.

However, Jensen Huang, the CEO of Nvidia, expressed skepticism about the enormous amount Altman aims to secure for the construction of AI chip manufacturing facilities. At the World Government Summit in Dubai, Huang noted, "You can’t simply assume that you will buy more computers. You also have to assume that computers will become faster, which implies that the overall requirement may not be as high." Despite his reservations about the fundraising target, Huang forecasted that the number of data centers will double within the next four to five years, increasing the current valuation of data center assets from $1 trillion to an anticipated $2 trillion as they power AI technologies across the globe.

Altman's initiative could pave the way for a formidable competitor to Nvidia, whose advanced AI chips are currently in high demand, causing supply bottlenecks. Presently, OpenAI heavily relies on Nvidia as a primary supplier for its AI computational needs. However, Altman is exploring the possibility of designing and producing proprietary chips, an endeavor paralleled by strategic partner Microsoft through its Athena project. Meta is also developing custom chips for its AI models, such as Llama 2.

Despite Altman’s ambitions, Huang pointed out that experienced chip manufacturers are innovating by creating more cost-efficient hardware, potentially rendering in-house chip development unnecessary. Furthermore, Altman has personally invested in Rain AI, a San Francisco-based startup working on chips that emulate human brain function. Last December, OpenAI secured a $51 million agreement with Rain AI to provide hardware support for their initiatives.

This pursuit for advanced AI chip production not only reflects the growing demand for AI capabilities but also signifies a shift in the competitive landscape as companies explore innovative ways to meet the evolving needs of the technology sector.