Bank of America (BofA) reaffirmed its buy rating on Nvidia and a $350 price target. If that target is reached based on the current share price, Nvidia's market value could rise by more than $3.1 trillion.

On Aug. 18 (local time), blockchain media outlet Decrypt reported that Nvidia shares traded at around $220.12 on the morning of Aug. 18. On that basis, the $350 target implies about 59 percent upside. Applying about 24.2 billion shares outstanding would put Nvidia's value at about $8.47 trillion.

The announcement is a reaffirmation of the existing outlook rather than a target increase. Vivek Arya (비벡 아리아), a Bank of America analyst, kept the $350 target in a report on Aug. 17. The target has been unchanged since it was raised to $350 from $320 in May.

Arya described Nvidia as an "unrivaled and sustainable growth franchise." He said Nvidia trades at about 18 times expected earnings, which he viewed as a valuation at a seven-year low.

Bank of America's core argument is that concerns about rising memory costs are overstated. It said investors are overemphasising the burden of high-bandwidth memory costs while not fully reflecting Nvidia's pricing power, economies of scale and about $119 billion in supply-chain commitments.

The report estimated that as Nvidia moves from Blackwell to the Rubin platform, the cost of high-bandwidth memory per server rack could rise by $200,000 to $300,000. It said the total rack price could rise by $2 million to $3 million to about $6 million to $7 million. In that case, it calculated that Nvidia could absorb higher memory costs while keeping gross margin in the mid-70 percent range.

Its view on competition is also optimistic. It judged that although Google, Amazon and Meta are developing their own AI chips, they have not been able to stop Nvidia's expansion. It said Nvidia revenue from hyperscale cloud customers rose 115 percent from a year earlier. That is almost twice the pace of overall cloud capital expenditure growth.

It also kept its long-term outlook. Bank of America said Nvidia could capture more than 65 to 70 percent of long-term AI infrastructure spending. In May, it raised the AI infrastructure market opportunity to $3 trillion in 2030 from $1.7 trillion. At the time, it raised its Nvidia estimates after fiscal first-quarter results beat market expectations.

Nvidia posted fiscal first-quarter revenue of $81.6 billion. That was up 85 percent from a year earlier. Data center revenue rose 92 percent to $75.2 billion. Non-GAAP gross margin held at 75 percent. The company guided for second-quarter revenue of $91 billion, with a margin of error of plus or minus 2 percent. The forecast did not include China data center computing revenue.

The market's focus is now shifting to earnings due on Aug. 26. Investors are expected to check whether second-quarter revenue beats company guidance, how higher memory costs affected margins, whether hyperscale customer demand remains strong, and how far preparations have progressed for the Vera Rubin launch.

The share-price trend remains volatile. Nvidia shares fell about 2.2 percent on the morning of Aug. 18 as semiconductor and technology stocks weakened broadly amid rising U.S. Treasury yields. That reduced its market value to about $5.33 trillion. On a current-value basis, reaching the $350 target would require additional value gains of more than $3.1 trillion.

Risks remain. Factors cited include competition from custom AI accelerators, rising memory and infrastructure costs, China-related export restrictions and the possibility of slower spending by major cloud operators. Bank of America's reaffirmation signals it is maintaining an optimistic view, but the stage for retesting the $350 thesis presented in May is likely to be the Aug. 26 earnings report rather than the reaffirmation itself.

Nvidia tokenised shares traded in the crypto market, NVDAB, are expected to broadly track moves in the Nasdaq-listed underlying stock. But because NVDAB trades outside U.S. market hours and circulates through crypto market infrastructure, temporary price dislocations could occur depending on liquidity, quote differences and the method of calculating 24-hour returns.