Bank of America Global Research reiterated a Buy rating on Nvidia (NVDA) this week with a $350 price target, designating the chipmaker its top sector pick ahead of fiscal second-quarter earnings scheduled for release after the market close on Wednesday, August 26. As the dominant supplier of AI accelerators to hyperscale cloud providers — including Microsoft, Amazon, Google, and Meta, which collectively account for the bulk of Nvidia's data center revenue — the company's results are widely treated as a real-time barometer of enterprise AI infrastructure spending.

Nvidia shares opened at $223.96 on Friday, up 2.3%, trading near their 12-month high of $236.54. Despite that rally, BofA argued in a research note that the stock remains inexpensive by its own historical standards. The bank pointed to a forward price-to-earnings ratio of approximately 16 times projected 2027 earnings per share — Nvidia's lowest forward multiple in a decade.

Q2 Revenue Beat and Strong Q3 Outlook Expected

BofA projects Nvidia will report fiscal second-quarter revenue between $94 billion and $95 billion, exceeding the company's own guidance of $91 billion. For the third quarter, the bank forecasts revenue of $107 billion to $108 billion, above the Street consensus of roughly $104 billion.

Part of the bullish thesis rests on the initial shipments of Nvidia's next-generation Vera Rubin chips, which BofA said marks the beginning of a multi-quarter upgrade cycle. Vera Rubin succeeds the current Blackwell platform, and its ramp will be closely watched as an indicator of whether Nvidia can maintain its accelerated product cadence against emerging alternatives such as AMD's MI400 series and in-house custom silicon from Google and Amazon. Spot GPU rental prices for B200 chips have risen to $5.66 per hour, near record highs, signaling continued supply tightness. The bank expects Nvidia to retain 65% to 70% of the AI accelerator market through 2030.

Memory Cost Concerns Called "Overblown"

A key investor concern heading into earnings is rising memory costs. DRAM now accounts for 40% to 50% of total build costs, up from a historical range of 15% to 20%. BofA pushed back on this worry, arguing that Nvidia's pricing power and sourcing relationships with suppliers such as SK Hynix limit the impact. On Vera Rubin NVL compute racks, memory costs represent just 60 basis points of margin headwind compared to the current Blackwell Ultra.

Overall gross margins are expected to stabilize in a range of 73% to 74% over time, down modestly from roughly 75% today.

Venture Commitments Viewed as Manageable

BofA also addressed questions surrounding Nvidia's venture investments. The bank calculated that Nvidia has deployed approximately $70 billion in direct equity into ecosystem partners, including $30 billion into OpenAI and up to $10 billion into Anthropic. These commitments effectively deepen Nvidia's ties to its largest customers, tying model developers more closely to its hardware and software stack. With $469 billion in projected free cash flow across 2026 and 2027, BofA said the balance sheet can absorb those commitments.

Prior Quarter Results and Analyst Consensus

Nvidia's most recent earnings report, released on May 20, showed first-quarter revenue of $81.61 billion, up 85.2% year over year, with EPS of $1.87 beating estimates of $1.76. The company also authorized an $80 billion share buyback program and raised its quarterly dividend to $0.25.

BofA projects Nvidia EPS to exceed $25 by 2030. Analyst consensus currently stands at a Buy rating with an average price target of $304.26, making BofA's $350 target among the most aggressive on the Street. Beyond headline revenue and EPS, investors will be watching Nvidia's data center segment growth, commentary on Blackwell Ultra adoption, and any updates on the Vera Rubin production timeline for further signals on the trajectory of AI compute demand.