Nvidia, which is on a bull run in the US stock markets, recently became the first chipmaker to hit a $1 trillion valuation on Wall Street. This is despite its core gaming chip business slowing down due to a broader slump in computer sales. So what is behind Nvidia's rise?
Having started off in 1993 as a graphics card maker for gamers, Nvidia has transformed into a chip powerhouse, with solutions for big tech companies and startups to power their artificial intelligence workloads.
For the uninitiated, Nvidia makes graphics processing units (GPUs). Though similar to central processing units (CPUs) found in almost all kinds of machines – from smartphones to computers, televisions, and even cars – GPUs are designed specifically for graphics-related workloads. Therefore, there is a GPU along with a CPU in a machine designed for graphics-focused workloads such as gaming consoles.
Graphics and gaming, however, are just one of the many things GPUs are good at. Unlike the CPU, which does serial processing – one process at a time – GPU is capable of parallel processing wherein it can execute multiple calculations and processes simultaneously. It essentially means GPUs are more efficient and powerful than CPUs for specialised workloads such as machine learning and deep learning, where multiple processes are performed simultaneously.
Nvidia pivoted from consumer-facing graphics and gaming hardware to advanced chips for data centres, and now AI chips. It started investing in tools to make GPUs programmable, and opened up the capability of its chips to perform beyond graphics. This allowed Nvidia to open its chips to businesses beyond those in gaming. The demand for specialised chips for the data-centre business and AI workloads helped Nvidia rise up the charts. Now, it features among the top-five semiconductor makers despite a slump in computer sales. This is largely because of the boom in AI enterprises, which require specialised GPUs such as those from Nvidia, to power their solutions.
Nvidia is an early beneficiary of the AI boom, but challengers aren't sitting idle. Big tech companies such as Google and Amazon have chips for AI models. For now, however, Nvidia seems like the dominant player in the chip market. Recently, it announced an AI supercomputer, DGX GH200. It is a single computer with 256 GPUs, and is touted to be the first supercomputer to break the 100-terabyte barrier for memory accessible to GPUs.