Barclays’ model shows AI spending cycle is far from peak. That means Nvidia shares are too cheap
Investors appear to be underestimating hyperscalers’ artificial intelligence spending plans, and therefore missing out on the true value of Nvidia , according to Barclays. Nvidia has been trading sideways since December after the stock’s blockbuster performance in recent years, with the chipmaker’s shares down less than 1% year to date. A big reason behind Nvidia’s underperformance this year is due to the broader rotation away from megacap technology stocks, amid concerns about high valuations and the longer-term viability of AI infrastructure spending. Investors largely ignored Nvidia’s blowout earnings report and strong guidance it gave in late February. Barclays believes the flat performance in Nvidia is a buying opportunity, however. Based on their analysis of financials from AI leaders OpenAI and Anthropic published in The Information, Barclays analysts believe that the market is significantly underestimating how much tech hyperscalers, like Microsoft and Google parent Alphabet , will need to spend in the next few years. Consensus hyperscale capex is at least $225 billion “too low” in 2027 and 2028, analyst Tom O’Malley wrote in a Wednesday note to clients, saying that “the capex up-cycle lasts into at least 2028 and could also be magnitudes larger vs. consensus.” Nvidia is, however, currently trading as if capex levels were to hit a ceiling in 2027, he said. “When cycles peak, multiples reflect the potential for lower future earnings power, which suggests a reason for NVDA trading at ~17.5x CY27E P/E today,” he said. “We would argue that even when assuming NVDA EPS growth is in line with cloud capex from our model +44%/+11% for CY27/CY28, which we see as conservative, the name is trading at ~14.5x CY28E numbers (or peak in this exercise) meaning that you would have taken nearly 15x turns off a traditional growth multiple at 30x. We believe this is simply too great a penalty, even in this scenario,” he wrote in the note. NVDA 1Y mountain NVDA stock performance over the past year. Moreover, O’Malley pointed out that hyperscalers’ total AI spending could climb even further as tech giants move to next-generation hardware. Barclays’ current estimates assume only Nvidia’s Blackwell-architecture GPUs, but the firm said that Nvidia’s newer solutions — including the Vera Rubin, Vera Rubin Ultra, and Feynman chip families — could lead total capex to increase. The average selling price, or ASP, of chips is also exceeding the standard pricing of older models, the analyst said. “We view this as a material positive for AI semis stocks,” the note reads. “Capex growth should naturally slow at some point, but we believe it will be materially higher in the near term and likely to peak later than what leading AI semis names are currently pricing in.” To be sure, the semiconductor industry is quite varied, and some stocks in the group have fared significantly better than Nvidia year to date. The iShares Semiconductor ETF , which tracks an index of U.S.-listed semiconductor companies including Micron , Nvidia and Advanced Micro Devices , is up 13.6% this year, while the S & P 500 is down about 1%. Micron has rallied more than 46% as it manufactures dynamic random-access memory, or DRAM, chips, which are heavily used in Nvidia’s graphics processing units.
<
