The Winning Chip

Altimeter Capital founder Brad Gerstner expressed cautious optimism Monday regarding OpenAI’s newly announced collaborations with Nvidia and AMD to bolster its AI infrastructure. Gerstner emphasized that these agreements are currently in the announcement phase, and deployment and tangible impact are yet to be realized.

“Now we will see what gets delivered,” Gerstner told CNBC, highlighting the critical distinction between a deal announcement and actual hardware deployment. “Ultimately, the best chips will win,” he added, underscoring the performance-driven nature of the AI compute market.

OpenAI’s pursuit of advanced silicon from both AMD and Nvidia reflects the escalating demand for computational power required to train and run increasingly sophisticated AI models. This strategic push also underlines the intensifying competition within the AI sector, where access to leading-edge hardware is becoming a key differentiator. The partnerships signal OpenAI’s proactive approach to managing potential bottlenecks in compute resources, which have become a significant constraint for AI development.

Gerstner noted that these deals serve as “more evidence that the world will remain compute-constrained despite best efforts to bring massive supply online.” This observation points to a broader trend in the AI industry: demand for compute is outpacing supply, leading to increased competition for resources and the need for innovative solutions to optimize hardware utilization.

Industry analysts concur that OpenAI’s move is indicative of the escalating AI arms race, with implications extending beyond commercial competition into the geopolitical arena, particularly between the U.S. and China. The race to achieve AI dominance has significant national security implications, driving investment and innovation in both countries.

Last year, OpenAI’s Chinese competitor DeepSeek, generated considerable attention by claiming to have developed a more cost-effective AI model compared to its U.S. counterpart. DeepSeek has continued its innovation trajectory, recently releasing new open-source models leveraging domestically produced AI chips. This demonstrates China’s growing capabilities in AI hardware and software, potentially reducing its reliance on foreign technology.

The dynamics between U.S. and Chinese AI firms have drawn scrutiny from government entities. Last week, a U.S. government report raised concerns about DeepSeek’s potential alignment with the Chinese government from a national security perspective.

According to the report by the National Institute of Standards and Technology’s Center for AI Standards and Innovation, content generated by DeepSeek models reflect Chinese Communist Party viewpoints more frequently than U.S. based models, raising questions about potential bias and influence.

OpenAI’s collaboration with AMD is encouraging, leading to increased production capacity to facilitate the development of more advanced AI models. The ability to access more diverse and scalable compute infrastructure is viewed as essential for OpenAI to maintain its competitive edge.

As OpenAI President articulated on CNBC’s “Squawk on the Street,” the impending “absolute compute scarcity” is not limited to OpenAI’s operations but extends throughout the broader AI ecosystem. He emphasized the necessity of “this whole industry to come together” to address the challenges of growing demand and limited supply.