Text size
New Street Research believes Nvidia is in a different situation than Cisco Systems was in before its stock plunged at the end of the internet bubble.
Justin Sullivan/Getty Images
Wall Street is worrying that gains linked to artificial intelligence could be part of a market bubble. The last thing anyone wants to do is pile into a hot trend at the peak and then ride a stock down into a thematic bust.
Nvidia
(ticker: NVDA), as the biggest winner from AI, is at the center of those doubts. Analysts see a chance that the stock could suffer the same fate as
Cisco Systems
shares (CSCO), which collapsed as the dot-com boom ended. Stock in the maker of networking hardware has yet to return to the peak it hit in 2000.
But New Street Research believes the circumstances surrounding Nvidia will differ from Cisco’s internet bubble selloff. On Monday, analyst Pierre Ferragu reiterated a Buy rating on Nvidia shares and reaffirmed his price target of $635.
“Is there ‘Cisco risk’ for Nvidia? We don’t think so,” he wrote. Nvidia isn’t trading at a “bubble valuation, and the downside risk to the stock is more limited.”
In Monday trading, Nvidia shares rose by 3% to $447.93.
The analyst said that during the dot-com boom, Cisco’s revenue rose by 15 times, with its share price rising to 121 times the earnings expected for the next 12 months and 22 times forward revenue. In contrast, Nvidia’s revenue is up about seven times, and the shares have a valuation of 27 times forward earnings and 14 times sales.
Ferragu estimates Nvidia’s stock could fall about 20% if its revenue in 2025 disappoints Wall Street, assuming expectations for profits are cut by 33%, sending the valuation down to the low of 30 times forward earnings it hit the last time results badly disappointed investors.
At current valuations, “investors already factor in a significant risk of pull back, mixed with a very likely deceleration beyond 2024,” he wrote. “If concerns about 2025 recede, the stock has significant appreciation potential.”
Nvidia dominates the market for chips used for AI projects.
Any comparison to Cisco understates Nvidia’s strong technology leadership in AI. What Nvidia has to offer is a more advanced multilayered infrastructure with hardware and software components, rather than a simpler router of internet data.
Start-ups and corporations prefer Nvidia’s chips because of its robust software programming ecosystem, CUDA. Developers have been building and sharing AI-related tools and software libraries for over a decade on Nvidia’s proprietary platform, making it easier to build AI applications rapidly.
And the shift toward AI infrastructure may be just getting started. Earlier this year, BofA Global Research said the technology industry is still in the early stages of a spending shift toward AI, with only 10% of cloud servers enabled with chips suited to AI projects.
Ultimately, it all means Nvidia has more headroom to grow. It has a bigger opportunity to increase its chips’ performance and AI software capabilities than Cisco has had with networking hardware.
Write to Tae Kim at tae.kim@barrons.com