Key Takeaways
- Bank of America maintains its Buy rating on NVDA stock with a $350 price objective, suggesting 78% potential gains
- NVDA has climbed only 4% year-to-date compared to the SOX semiconductor index’s nearly 72% rally
- BofA analyst Vivek Arya argues that fears around memory expenses and custom ASIC threats are exaggerated
- Despite ASIC competition since 2015, Nvidia’s GPU accelerator revenue has multiplied 700-fold
- The forthcoming earnings announcement could serve as a positive catalyst to shift investor sentiment
Nvidia has posted just a 4% gain in 2026, a dramatic contrast to the SOX semiconductor benchmark’s nearly 72% surge. The divergence is striking for a company that continues to command the AI chip infrastructure landscape.
Vivek Arya, a Bank of America semiconductor analyst who ranks in the top 2% of Wall Street professionals, believes this underperformance is unwarranted — and he’s backing that conviction with a $350 price objective on NVDA. With shares currently hovering around $197, BofA’s target implies approximately 78% upside potential.
Arya identifies four primary headwinds troubling investors: deteriorating gross margins due to elevated memory expenses, threats from application-specific integrated circuits, concentrated institutional holdings, and questionable capital allocation related to vendor financing arrangements.
His assessment? These worries are either blown out of proportion or fundamentally misinterpreted.
Regarding memory expenses, Arya contends that although HBM costs per rack may increase by $0.2–0.3 million in the transition from Blackwell to Rubin platforms, total rack pricing could simultaneously jump $2–3 million — escalating from approximately $3–4 million to $6–7 million. This pricing leverage, he maintains, should sustain gross margins around the mid-70% level.
Custom Silicon Competition Has Always Existed
The ASIC competitive landscape is genuine, but hardly unprecedented. Google unveiled its TPU back in 2015. Amazon introduced Trainium in 2020. Meta rolled out MTIA in 2023. Throughout this entire period, Nvidia’s GPU accelerator revenue expanded 700-fold.
Nvidia’s recent segment disclosures reveal that hyperscaler sales jumped 115% year-over-year — approximately double the expansion rate of cloud capital expenditure. According to Arya, this demonstrates Nvidia is capturing market share rather than surrendering it.
The fundamental distinction between GPUs and ASICs, he emphasizes, is versatility. ASICs are engineered for particular workloads within individual hyperscaler environments. Nvidia delivers a platform with universal compatibility. Arya projects Nvidia will command 65–70% of AI capital spending over the long term, with the remaining 30–35% distributed among ASICs and alternative suppliers like AMD.
Regarding ownership concentration, approximately 78% of active S&P 500 funds maintain NVDA positions, with a 1.15x weighting. Strategic equity investments total roughly $65 billion. While Arya recognizes these as potential headwinds, he calculates the investments consume under 35% of free cash flow, preserving capacity for share repurchases and dividend payments.
Upcoming Earnings Report Could Shift Momentum
Arya is highlighting the approaching earnings announcement as the catalyst that may reverse sentiment. He anticipates it will “reinforce Nvidia’s competitive advantages in products, pricing, and supply chain management.”
At present valuations, he notes NVDA is trading at an 18x forward price-to-earnings multiple — the lowest level in seven years. Its large-cap technology counterparts — Amazon, Meta, Google, Microsoft, Apple — command an average of 22x–19x forward PE ratios based on CY27/28 projections. That represents a 30–35% premium versus Nvidia, which Arya characterizes as indefensible considering the company’s growth trajectory.
Wall Street consensus supports this bullish stance. Among 37 analyst ratings tracked, 36 recommend Buy while just one suggests Hold. The consensus price target stands at $309.33, indicating 57% upside over the next twelve months.



