BofA raises Nvidia stock price target to $320 on AI demand

BofA raised its Nvidia price target to $320 from $300 and lifted fiscal 2028-2029 sales and EPS estimates by 7% each, while keeping a Buy rating. The firm cited strong AI compute demand, multiple near-term catalysts including fiscal Q1 2027 earnings on May 20 and a possible new CPU launch at Computex, plus the Vera Rubin launch in the second half. Offsetting the bullish view, BofA sees gross margin drifting down 30-50 bps annually as HBM costs rise and competition intensifies.

Analysis

The market is still underestimating how much of NVDA’s next leg is about durability, not just growth. A higher target built on unchanged multiple math says the equity is being valued as if earnings visibility stops in 2027, yet the product cadence plus hyperscaler custom-silicon spend suggests the demand curve can extend beyond the current consensus window. The subtle bullish read is that every incremental concern about “peak AI capex” has been met by another layer of workload specialization, which keeps NVDA embedded even where competitors gain token share.

The competitive overhang is real but asymmetrical. AMD’s next GPU cycle and custom chips at hyperscalers should pressure price/mix at the margin, but those programs also validate that AI inference is becoming too important to single-source, which tends to enlarge the total market before it compresses it. The bigger second-order risk is supply-chain complexity: as HBM content rises, gross margin becomes more sensitive to memory pricing and packaging bottlenecks than to pure chip ASPs, so investors should watch for any sign that gross-margin guideposts start drifting faster than the Street expects.

The China optics matter less for direct revenue than for sentiment and policy optionality. Any perception of improved U.S.-China trade temperature can lift the multiple briefly, but the real economic upside is preserving flexibility on export-adjacent product variants and component allocation rather than reopening a huge addressable market. The contrarian point: the stock may be less “cheap” on near-term earnings than on normalized power in 12-24 months, which means the setup is good for long-duration holders but vulnerable to a post-earnings reset if management doesn’t re-accelerate forward bookings or margins.

Key Decisions for Investors

• Stay long NVDA into the May 20 print, but size via call spreads rather than outright equity; favor 3-6 month tenor to capture the June Computex/Rubin cadence while limiting downside if guidance is merely in-line.
• Use any post-earnings weakness in NVDA to add exposure only if gross-margin commentary remains within a 30-50 bps annual decline band; a faster deterioration would be the clearest signal that mix/supply-chain pressure is overtaking unit growth.
• Pair trade: long NVDA / short AMD over the next 1-3 months. The thesis is that NVDA’s ecosystem and software lock-in should defend mix better than the market expects, while AMD’s upside is more dependent on share-gain proof points that may be harder to monetize immediately.
• If you want to hedge AI capex beta, short a basket of hyperscaler custom-silicon beneficiaries on any post-event strength while staying long NVDA; the trade works if the market starts rewarding the incumbent platform over incremental in-house silicon announcements.
• For event-driven traders, buy NVDA call spreads ahead of Computex and monetize into the event unless management pre-announces upside. The risk/reward is best if the market stays focused on Rubin and cash return, but it weakens quickly if the stock gaps on already-high expectations.