They are so coveted that they are transported in armored trucks with security measures equivalent to those used for diamonds. The global digital industry and its 40,000 clients are lining up to get them, and the waiting list only grows longer. These are the famous GPUs or accelerated computing chips from the company Nvidia, which today dominates about 80% of the market and is key for the complex mathematical calculations of AI. These same chips have turned this American company—now the most valuable in the world and one of the seven valued at over $1 trillion—into the cornerstone of this new revolution.

 

Unlike other Big Tech companies, Nvidia sought the key to success for 30 years—facing several near-bankruptcies. Its ability to adapt is attributed to the unique management style of its founder, the Taiwanese Jensen Huang, who preaches the value of resilience and a horizontal organization with strategic communication at every level. In his speech to Stanford graduates, his alma mater, he said that to achieve “greatness,” one needs character and “abundant doses of pain and suffering.”

 

What is impressive about this story is that until recently, Nvidia was associated with the video game industry, to which it provided graphics cards that revolutionized gamers’ user experience. Realizing that semiconductors were useful for training AI systems, the company leaped from a niche market to meet a global demand with a great need to increase “computing power,” with clients like Google, Microsoft, and Amazon requiring them to power their data centers around the planet.

 

But Nvidia’s success also coexists with a great threat because its main semiconductor factory is in Taiwan, an island constantly threatened by China. The so-called “chip war”—a diplomatic, political, and economic battle—suggests that whoever controls semiconductor manufacturing will hold military supremacy and future industrial development. The United States—displaced by Asia in production—is seeking to diversify production locations, even mentioning Latin America as an option for setting up nearshore industries. However, the shortage of specialized human capital and the very low level of investment in enabling infrastructure to make this an unrealistic bet, although some analysts see opportunities in research or design centers.

 

When it comes to ‘Chilean computing power,’ the local infrastructure is precarious and more traditional because AI models require intensive use of GPUs, and in our case, we only have access through the cloud. While we still have some potential opportunities compared to the region, the key is to secure this advantage with strong investment in more and better infrastructure—something that might be considered in the National Data Centers Plan that the government will present in the second half of this year, and which could be the master key to boosting more massive AI development in the country.

 

Mónica Retamal F.

Kodea’s Executive Director