While the market is still debating whether artificial intelligence is an expensive “tech bubble,” BofA Securities points out in its latest research report that AI-related spending will not only avoid a downturn but will instead be “stronger and more sustained.” NVIDIA (NVDA), AMD, and Broadcom (AVGO) continue to be listed as “top picks” in the semiconductor sector, while Micron Technology (MU) and Marvell Technology (MRVL) have also been added to the core list.
Given the capital expenditure outlook presented by major hyperscale cloud service providers such as Microsoft (MSFT), Amazon (AMZN), Alphabet (GOOGL), and Meta Platforms (META) in their recent first-quarter earnings reports, BofA expects that by 2030, the total addressable market (TAM) for AI data center systems will surge to $1.7 trillion, up from its previous forecast of $1 trillion. Approximately $1.2 trillion of this may come from AI accelerators, higher than the previously forecast $1 trillion, with the upward revision driven by increased shipments of custom application-specific integrated circuits (ASICs) from hyperscale cloud providers, such as Google’s TPU or Amazon’s Trainium chips. The TAM for data center CPUs could reach approximately $110 billion, while the AI networking market may expand to about $316 billion. Analysts note that as the tail effect of AI workloads lengthens, compute and memory components are becoming increasingly diverse, but this complements rather than replaces the overall market size.
Over the past two years, Wall Street’s AI investment logic has primarily focused on “infrastructure front-running.” BofA analysts believe that with AI unicorns such as OpenAI and Anthropic about to initiate IPO processes, the market will gain a clearer view of how top-tier model producers convert computing power into commercial value. Once these startups demonstrate a closed-loop profitability model in the secondary market, the trillion-dollar spending of Microsoft, Amazon, Alphabet, and Meta will have a clear financial rationale. Analysts emphasize that 2026 will be a critical juncture for accelerating AI sales and return on investment, while 2027 will require verification of whether “token economic efficiency” can be realized as expected. Amid exuberant expectations, BofA also maintains a level-headed stance, acknowledging that current supply chain bottlenecks may limit short-term shipments of cutting-edge components and that market doubts about the sustainability of capital expenditures persist. Additionally, BofA has raised the following price targets: NVIDIA (from $300 to $320), AMD (from $450 to $500), Marvell Technology (from $125 to $200), and Micron (from $500 to $950), while keeping Broadcom’s target unchanged.
While BofA’s bullish stance is certainly encouraging, investors should remain vigilant about potential risks. Nevertheless, the BofA report sends a clear signal: AI infrastructure investment represents a “structural long-term bull market” that will last at least until 2030. NVIDIA remains the one “selling shovels,” while AMD, Broadcom, Marvell Technology, and Micron are digging for their own gold from different corners.
