Market Cap Breaks Through $5.5 Trillion! As NVIDIA Hits New Highs, Bank of America Reveals: Five Top Picks for AI's Next Phase

On Wednesday,$NVIDIA (NVDA.US)$The stock price surged significantly again and hit a new high.all-time highs.This strong performance not only solidified its position as the 'world's largest stock by market cap,' but also powerfully drove both the S&P 500 and Nasdaq indices to break historical records.

NVIDIA’s stock price achieved a six-day winning streak, closing up 2.3%. During trading, it hit a peak of $227.84 at one point.Its total market capitalization surpassed the $5.5 trillion mark for the first time in history, making NVIDIA the first company ever to achieve this milestone.This remarkable achievement once again indisputably highlights NVIDIA’s unshakable leading position in the chip industry.

As NVIDIA's earnings release approaches (after market close on May 20th Eastern Time), bullish sentiment continues to build, with major Wall Street firms raising their price targets successively. Among them, Bank of America increased its target price from $300 to $320 and reiterated it as the 'top pick' in the industry. Wells Fargo & Co, expressing strong optimism about the growth potential of NVIDIA’s data center business, raised its target price significantly from $265 to $315.

At this critical juncture in the construction of AI infrastructure,How big is the "pie" of the data center market? Besides NVIDIA, which other companies are worth focusing on?Regarding these key areas, Bank of America has provided clarity in its latest research report, offering the market a detailed "navigation map" for the semiconductor sector.

How big is the "pie" of the data center market?

Bank of America raised its forecast for the total addressable market (TAM) of AI data center systems by 2030,from the previously estimated $1.4 trillion to approximately $1.7 trillion, with a compound annual growth rate (CAGR) as high as 45%.

This growth is not driven by a single component but rather by incremental gains from a diversified overall architecture:

AI accelerators (GPUs/ASICs, etc.) lead the growth: The TAM for AI accelerators is expected to grow from the initially projected $1.0 trillion to approximately $1.2 trillion by 2030.

Network and CPU segments expand in tandem: The expected scale of AI networking equipment has been revised upward to $316 billion; the market size for data center server CPUs is projected to reach about $110 billion (with AI CPUs accounting for $88 billion).

"Incremental" rather than "replacement" logic: The diversification of computing power and memory components is the core driver of TAM expansion. For instance, standalone CPU racks will work in tandem with existing GPU-CPU racks; ultra-low latency memory racks based on SRAM will also coexist with GPU racks based on HBM.

Faced with tech giants' capital expenditures often reaching hundreds of billions, the market can't help but worry whether this is a bubble where costs might not be recouped. In response, Bank of America believes2026 will be the turning point for accelerated growth in AI sales and return on investment。

First, the "profitability" of top-tier large models is being realized,For example, Anthropic's annualized revenue run rate has surged to $19 billion.On the other hand, cloud computing giants have massive order backlogs,Microsoft’s commercial Remaining Performance Obligation (RPO) stands at nearly $627 billion, while Amazon AWS holds $364 billion in backlog orders.These long-term contracts, backed by real cash, are the strongest support for continued investment in AI infrastructure.

Which other companies are worth watching?

In this highly predictable sector, aside from NVIDIA, core supply chain hubs capable of reliable delivery have become the focus of investor attention. Bank of America explicitly highlighted five top industry picks: $NVIDIA (NVDA.US)$ 、 $Broadcom (AVGO.US)$ 、 $Micron Technology (MU.US)$ 、 $Advanced Micro Devices (AMD.US)$ and $Marvell Technology (MRVL.US)$ 。

First of all, $NVIDIA (NVDA.US)$ It remains the absolute leader in its sector, with Bank of America raising its price target from $300 to $320.The valuation support behind this is its 28x forward P/E ratio for 2027 and a comprehensive product lineup capable of meeting every细分需求 of AI workloads.

Bank of America expects NVIDIA to maintain approximately 65% to 70% or more of the AI accelerator market share over the long term.The market is closely watching its series of catalysts:First-quarter earnings on May 20, potential new CPU releases at the Computex conference in early June, the Vera Rubin architecture in the second half, and a possible cash return plan later this year.

The most eye-catching aspect of this report is undoubtedly $Micron Technology (MU.US)$ , with Bank of America aggressively raising its price target from $500 to $950.The core logic lies in the structural reduction in memory supply elasticity due to capital constraints, advanced packaging limitations, and power shortages. Driven by AI, memory demand will continue to outstrip supply. Bank of America applied a sum-of-the-parts valuation for Micron: just its highly sustainable AI/HBM business alone, valued at 27x P/E, supports a value of $240/share, with an additional $710/share (based on a 3.1x P/B ratio) coming from its traditional DRAM/NAND cyclical business, forming this doubled high expectation.

On the other side of the computing power battlefield, $Advanced Micro Devices (AMD.US)$ the price target was raised from $450 to $500.This corresponds to a 42x forward price-to-earnings (P/E) ratio for 2027. Bank of America highlights its robust strength in the CPU sector and significant potential to capture market share in the AI CPU and GPU markets in the future. The potential for an earnings per share (EPS) compound annual growth rate of over 50% provides solid support for its valuation. Additionally, the analyst day event in July and the possibility of securing additional gigawatt (GW)-scale data center projects without warrant requirements are key points to watch moving forward.

In the networking and custom chip space, $Marvell Technology (MRVL.US)$ the target price has seen a significant jump from $125 to $200,with the valuation benchmark now shifting to a 30x forward P/E ratio for 2028. Marvell’s confidence stems from the sustained ramp-up of 800G and 1.6T optical modules, where it is expected to maintain a dominant market share of 60% to 70% in these generations of DSP (digital signal processor) markets. Moreover, demand for custom chips remains strong; for instance, Microsoft's project is expected to scale up by 2028, surpassing previous conservative management estimates.

Finally, as solid as a rock, $Broadcom (AVGO.US)$ Broadcom was given a $450 target price by Bank of America,based on a 26x forward P/E ratio for 2027. Broadcom's advantage lies in its unmatched earnings certainty, with recent framework contracts signed with Google and Meta directly securing visibility into fiscal year 2027. With top-tier profitability, free cash flow generation, and return on investment within the semiconductor sector, Broadcom's AI-related revenues even have potential upside beyond the market consensus expectation of $110 billion.

Summary

The Bank of America report notes that the development of AI infrastructure is entering a longer-term expansion phase. Supported by high backlog orders from cloud providers and gradually materializing revenue, hardware investments have evolved from the initial focus on 'single compute power expansion' to a diversified development pattern featuring 'comprehensive synergy between compute power, memory, and network communication.'

338K Views