Artificial intelligence, popularly known as AI, is at its prime now. It is probably one of the hottest and most talked-about, if not the hottest, technologies right now. One of the components of AI is AI hardware.
As the name implies, AI hardware refers to the hardware used in AI. The simplest examples of AI hardware would probably be the graphical processing units, or GPUs, that have been redeployed to perform machine learning tasks.
According to a recent opinion piece published on MarketWatch, a website that provides financial information, business news, analysis, and stock market data, AI hardware has a projected annualized growth rate of almost 27 percent. In addition, it was a $10 billion worth industry in 2021 and is predicted to be an $89 billion industry by 2030.
And when it comes to AI hardware, the opinion piece, written by Daniel Newman, also said that the “clear winner” here, aside from the software industry, is the semiconductor industry, and NVIDIA leads this.
However, Newman also pointed out that companies like Intel and AMD are trailing just behind.
Intel is known to have built several AI capabilities into its Xeon CPUs, and while the top choice right now is NVIDIA, the former is showing some prowess.
MarketWatch’s Newman said Intel has been “oversold and underestimated” when it comes to AI hardware. The opinion writer added that NVIDIA’s strong position in the AI market resembles Intel’s position in data center CPUs not long ago.
He further noted that under the leadership of chief executive officer Pat Gelsinger, the company has showcased progress through refocusing its efforts on process leadership and aggressively breaking the boundaries to get its operational and process execution shored up.
“With Intel finally launching its first discrete data center GPUs, I expect the company compete aggressively on price and performance. If it can deliver a lower power product, it could be well received. Also, with Nvidia’s hypergrowth, there is more chatter about challenges servicing the massive customer base, and Intel, like AMD, could see attrition from Nvidia looking for greater attention and support,” Newman wrote.
Furthermore, the MarketWatch writer said that the demand for AI hardware on a worldwide scale will be bigger than ever. Thus, enterprises and cloud providers seeking to deploy large language models and other AI-intensive workloads at scale must consider factors such as cost and power implications.
“That said, I believe NVIDIA’s growth will inevitably slow, and the likely beneficiaries will be AMD and Intel, plus Qualcomm QCOM, -0.65 percent for edge computing, AWS with Trainium and Inferentia for public cloud workloads, and startups including Sambanova, Cerebras, and Groq, which build specialized AI chips,” Newman said in the opinion piece.
