-
Nvidia has dominated the early stages of the AI boom.
-
Most of the growth in AI has come from the cloud so far, but that could change.
-
Nvidia-partner Arm is poised to be a top chip supplier in the edge.
Nearly three years into the AI boom, there's no question who the big winner has been.
Nvidia (NASDAQ: NVDA) has grabbed much of the spoils from the new technology as the chipmaker has seen its market cap jump by roughly $4 trillion. No other company can come close to that claim.
Nvidia has provided the building blocks in its GPUs and superchips to power generative AI tools like ChatGPT, and its technological advantage in the area only seems to have strengthened since OpenAI launched its trademark chatbot.
Whatever happens next in AI, Nvidia is sure to play a major role, especially as the company is ambitious enough to roll out a new platform every two years, but Nvidia has risen to AI dominance through the data center, catering to cloud hyperscalers and generative AI start-ups like OpenAI. However, Nvidia doesn't make chips for the so-called edge, which refers to the devices that people actually interact with, like smartphones, computers, and appliances.
One semiconductor company that serves the edge segment as well as the data center is Arm Holdings (NASDAQ: ARM), and it looks well positioned to be a leader in the next generation of smart devices.
Arm is already in the AI race. It has more than a 99% share of the smartphone market thanks to its battery-efficient CPU technology, which is also helping to drive its growth in the data center, where power efficiency is at a premium as well.
Arm has partnered with Nvidia and top cloud infrastructure companies like Microsoft and Alphabet, which license Arm's architecture for some of their chips.
The company, which has traditionally licensed its technology and collected money from royalties and the sales of products with its architecture, is moving to be a stand-alone chip designer, becoming a competitor of customers like Nvidia.
Last month, Arm hired Amazon's AI chip director, Rami Sinno, who led the development of its AI chips Trainium and Inferentia.
Arm has not made a formal announcement, but according to media reports, the company is preparing to design and sell its own chips, securing Meta Platforms as one of its first customers for a new data center chip.
In addition to the move into chip design, Arm has also advanced its licensing, moving from CPUs to a more complete package called a system-on-chip (SoC), which makes it easier and faster for its customers to move from the design phase to manufacturing those chips.