Can Nvidia maintain its AI chip dominance in the face of numerous challenges?

Nvidia Corp. has become the world's most valuable chipmaker thanks to its dominance in Artificial Intelligence (AI) computing.

But as investors turn their attention to the true profitability of AI, the question is whether Nvidia can maintain its dominance in the face of a myriad of challenges, from fierce competition to geopolitical instability.

In the AI gold rush, Nvidia has emerged as the most effective “shovel seller.” Nvidia’s revenue continues to soar, and orders for its AI accelerator chips continue to fill up.

Nvidia's lasting success, however, depends on CEO Jensen Huang's ability to steer the ship through many storms.

Nvidia’s main moneymaker right now is the H100 Hopper chip, named after computer science pioneer Grace Hopper. It’s an upgraded version of the graphics processing unit (GPU) familiar to gamers. The H100 will soon be replaced by the more powerful Blackwell line, named after mathematician David Blackwell.

Both Hopper and Blackwell incorporate technology that allows clusters of Nvidia-powered computers to work as a single unit, crunching massive amounts of data and performing calculations at extremely high speeds. This makes them perfect for the resource-intensive task of training neural networks, the foundation of next-generation AI products.

Nvidia, founded in 1993, pioneered this market with investments more than a decade ago, betting that parallel processing would make its chips valuable beyond gaming. The Blackwell chips were advertised as offering 2.5 times the AI training performance of the Hopper chips. The new design was so complex that it couldn’t be manufactured as a single chip, but was actually two chips tightly connected to act as one.

Nvidia was already the king of graphics chips. Nvidia engineers realized in the early 2000s that they could adapt these graphics accelerators for other applications. At the same time, AI researchers discovered that their work could be made more practical using these chips.

Nvidia now controls about 90% of the data center GPU market, according to market research firm IDC. Nvidia has achieved this dominance by constantly updating its products, including the software that powers its hardware, at a pace no other company can match.

Despite its peak, Nvidia faces a number of challenges. Major cloud service providers and Nvidia customers such as AWS (Amazon), Google Cloud (Alphabet) and Azure (Microsoft) are developing their own chips.

Traditional rivals like AMD and Intel are not left out. AMD, Nvidia’s biggest rival in graphics chips, expects sales of AI accelerator chips to be flat in the first half of the year, while Nvidia’s chip sales are still growing more than 50% each quarter.

Meanwhile, the US government has tightened restrictions on exports of advanced AI chips to China, the world’s largest semiconductor market. Mr. Huang said these restrictions are counterproductive and could create opportunities for other companies, especially China’s Huawei, to rise.

Additionally, there are now predictions that the development of AI data centers is showing signs of "running out of steam" and Microsoft has scaled back some data center projects.

In the face of these challenges, Nvidia is working to prove to its biggest customers that its products are the best. Nvidia's future depends on its ability to maintain its technological edge in the face of increasingly fierce competition.

Although orders continue to pile up and tech giants have pledged hundreds of billions of dollars to AI, Nvidia's dominance in an ever-changing industry is by no means guaranteed.

(TTXVN/Vietnam+)

Source: https://www.vietnamplus.vn/nvidia-co-the-duy-tri-ngoi-vuong-ve-chip-ai-truoc-vo-van-thach-thuc-post1041898.vnp