Nvidia Stock Has Underperformed – That's About to Change, Says Bank of America

Despite still dominating the AI space, Nvidia (NASDAQ:NVDA) has significantly underperformed its peers in 2026. The stock is up by a modest 4% year-to-date, while the industry barometer index, the SOX, has gained almost 72%, even when accounting for a recent pullback.

4th of July Sale - 70% Off

Explore NVDS for 2X short leverage on NVDA

So, what is driving the muted sentiment? BofA's Vivek Arya, an analyst ranked among the top 2% on Wall Street, thinks investor concerns are based on several factors, which he has set out to address.

One relates to worries about "gross margin pressure from higher memory costs."

Regarding this, Arya thinks investors are placing too much emphasis on HBM cost headwinds while overlooking Nvidia's ability to maintain pricing, its scale, and its $119 billion in supply-chain commitments. Although HBM content per rack could increase by approximately $0.2-0.3 million from Blackwell to Rubin, rack prices may rise by $2-3 million, moving from roughly $3-4 million to $6-7 million, as a result of improvements in compute, networking, and software offerings. As a result, Arya expects gross margins to stay near the mid-70% range.

Secondly, there are concerns about custom ASIC competition. Here, however, Nvidia has faced competition from custom ASICs since Google introduced its TPU in 2015, followed by Amazon's Trainium launch in 2020 and Meta's MTIA launch in 2023.

Despite this competitive landscape, Nvidia's GPU accelerator sales have grown 700x since 2015. In fact, Arya points out that according to its latest segment disclosure, Nvidia's sales to hyperscalers increased 115% YoY, nearly twice the growth rate of cloud capex, indicating that its share of hyperscaler spending is expanding rather than declining.

Nvidia offers a broadly available and supported platform, whereas ASICs are specialized solutions designed for a narrower set of applications and are only accessible within the respective hyperscaler's ecosystem. "Long-term we expect NVDA's breadth of AI products to help it maintain dominant (65-70%+) share of AI capex, with remaining 30-35% shared across ASICs and other merchant (AMD) vendors," the 5-star analyst said.

Arya also points to "crowded investor ownership," and "unproductive use of cash in vendor financing" compared to stronger buybacks/dividends, as points of contention.

While concentrated ownership (1.15x weighting and 78% ownership among active S&P 500 funds) and strategic investments of approximately $65 billion "remain overhangs," he estimates that the latter accounts for less than 35% of FCF, leaving significant room for dividends and share repurchases.

Meanwhile, at a "high level," Arya believes that Nvidia is exposed to a similar mix of AI-driven opportunities and potential challenges from rising memory costs as its major technology peers, including Amazon, Meta, Google, Microsoft, and Apple. These five companies currently trade at an average of 22x/19x PE based on consensus CY27/28 estimates, representing a premium of approximately 30-35% compared with Nvidia's current valuation. He attributes this valuation gap to concerns around costs and competitive pressures noted above.

However, he thinks those concerns will be laid to rest shortly. "We expect the upcoming earnings call to be a positive catalyst, clarifying NVDA's durable moats across its products, pricing and supply chain," the analyst summed up.

Bottom line, Arya maintained a Buy rating on NVDA shares, backed by a price objective of $350, implying the stock will gain 78% in the year ahead. (To watch Arya's track record, click here)

The Street's average target might not be quite as high, but at $309.33, still offers 12-month upside of 57%. All told, based on 36 Buys vs. 1 lone Hold, the stock claims a Strong Buy consensus rating. (See NVDA stock forecast)