Bank of America: Sustained AI Demand Positions NVIDIA and AMD as Top Chip Picks

Bank of America Securities has reaffirmed its optimistic stance on artificial intelligence (AI) chip stocks in a recent research report, stating that AI-related expenditures are expected to remain "stronger and more enduring." The firm continues to favor NVIDIA, Advanced Micro Devices (AMD), and Broadcom as its top selections within the semiconductor sector.

The report indicates a significant upward revision in market expectations for AI. Analysts at the bank pointed to robust capital expenditure outlooks from major hyperscale cloud providers, as evidenced in recent first-quarter earnings. Consequently, they have raised their total addressable market (TAM) forecast for AI data center systems in 2030 from a previous estimate of $1.4 trillion to $1.7 trillion. Within this, the AI accelerator market is projected to reach $1.2 trillion, data center CPUs around $110 billion, and the AI networking market approximately $316 billion.

Analysts identified clear demand-side drivers. They anticipate 2026 will be a year of accelerated AI sales and return on investment, with potential initial public offerings (IPOs) from major AI companies potentially further improving the sector's outlook. By 2027, efficiency gains are expected from the introduction of new architecture computing and memory systems. While market concerns persist regarding the sustainability of capital expenditures, the strong free cash flow of major cloud service providers is seen as a mitigating factor.

Based on this assessment, Bank of America has substantially increased its price targets for several AI chip stocks. The target for NVIDIA was raised to $320, maintaining its status as the top industry pick. The target for Advanced Micro Devices was raised to $500, for Micron to $950, and for Marvell Technology to $200.

The bank maintains its bullish stance despite current supply chain bottlenecks that may limit shipments of cutting-edge components. It argues that well-positioned key suppliers in AI computing, networking, and memory should continue to deliver and achieve outperformance. The ongoing diversification of computing and memory components is viewed as additive to the overall market, rather than a replacement for existing demand.