The current stage of ai - infrastructure construction - has been widely discussed. This is an obvious bullish for chip manufacturers and large-scale datacenter providers like other. Large-scale data centers provide the massive cloud computing power required for ai applications, and analysts predict that these facilities will bear the computing power required for a large number of ai workloads as data center demand increases.$NVIDIA (NVDA.US)$Such chip manufacturers and $Amazon (AMZN.US)$, $Microsoft (MSFT.US)$and $Alphabet-A (GOOGL.US)$Please use your Futubull account to access the feature.

However, according to technical analysts, the next bottleneck in artificial intelligence infrastructure, which is also an investment opportunity, is the network. In conventional technical terms, a network refers to a device network that can transmit and share information through physical or wireless communication. However, in the field of artificial intelligence, due to the large language models and other artificial intelligence applications that require very high bandwidth and low latency, the network requirements are higher.

"Although Nvidia and its GPUs dominate in generative artificial intelligence, we believe that the network is essential in supporting the hardware of models and applications such as ChatGPT," Morningstar analysts said in a June 2024 report. "Until now, the main focus has been on graphics processing units (GPUs), which, of course, are the most important part of this puzzle. But we see the network as the place where the next bottleneck will occur."

Clare Pleydell-Bouverie, portfolio manager at Liontrust Asset Management, said in a May talk, "This is because 'large-scale systems' that are about to be launched, such as Nvidia's rack-scale system, will require 'considerably more' infrastructure content, such as networks."

According to Morningstar technology equity analyst William D. Kerwin and technology stock strategist Brian Colello, they believe that the demand for fast networks in generative artificial intelligence will directly translate into 'strong, long-term growth for well-located network suppliers.' The company expects investment in generative artificial intelligence model training and inference to drive AI network spending to grow 34% to $34 billion over the next five years, up from Morningstar's 2023 estimate of $8 billion.

"The network creates performance bottlenecks for generative artificial intelligence model development," Morningstar analysts noted. "Well-positioned network companies are good second derivative investments for investing in generative artificial intelligence."

The preferred stocks in morningstar's generative ai network trend are other, which they say currently has an attractive undervaluation and provides investors with an immediate opportunity to participate in the growing generative ai network investment.$Marvell Technology (MRVL.US)$In this network trend, other key winners include arista networks(ANET.US), according to morningstar.$Arista Networks (ANET.US)$, $NVIDIA (NVDA.US)$ and $Broadcom (AVGO.US)$, according to Morningstar.

However, their analysts believe that the generative artificial intelligence opportunities in these three stocks have already been 'significantly priced in' as their stock prices have experienced 'strong' appreciation. 'However, patient investors can wait for corrections as the long-term fundamental opportunity is strong,' Morningstar added. The company also added that it is optimistic about the adoption of Ethernet in generative artificial intelligence networks, referring to a network standard.

According to morningstar, arista will be the main beneficiary of ethernet's transition. The technology currently widely used is infiniband. "It's rare for a company to really step into the role of infrastructure provider," Pleydell-Bouverie mentioned the network infrastructure. She noted that other and broadcom were stocks that could participate in this trend. Broadcom is a "leader" in network chip field and will benefit greatly from the emergence of ethernet, she added. " Ethernet networks are gradually becoming the de facto standard for extending these ai workloads. And broadcom has the best class of chips to drive this ethernet network."$Meta Platforms (META.US)$According to morningstar, arista will be the main beneficiary of Ethernet's transition. The technology currently widely used is InfiniBand. "It's rare for a company to really step into the role of infrastructure provider," Pleydell-Bouverie mentioned the network infrastructure. She noted that other and broadcom were stocks that could participate in this trend. Broadcom is a "leader" in the network chip field and will benefit greatly from the emergence of Ethernet, she added. " Ethernet networks are gradually becoming the de facto standard for extending these ai workloads. And broadcom has the best class of chips to drive this Ethernet network."

Editor/new