Bank of America lifts its AI data center market forecast to $2.2 trillion and names five chip stocks as must-own positions for Q4.

Bank of America named Nvidia, Intel, Micron Technology, Marvell Technology and Lam Research as its top five semiconductor picks for the fourth quarter, citing seasonal strength, attractive valuations and specific catalysts for each name. The call coincided with a substantial upward revision to the bank's long-term AI infrastructure outlook. Analyst Vivek Arya raised BofA's forecast for the AI data center systems market to $2.2 trillion by 2030, up from a prior estimate of $1.8 trillion, implying roughly 40% annual growth.

The bank's thesis rests on three pillars. Historically, the fourth and first calendar quarters have delivered the strongest returns for semiconductor equities; between 2010 and 2025, chip stocks posted median outperformance of 300 to 500 basis points relative to the S&P 500 during those periods. Valuations offer a second advantage: the PHLX Semiconductor Index trades at approximately 21 times forward earnings, roughly 12% below its median multiple of 24 times since ChatGPT launched, despite projected earnings growth exceeding 40%.

The third pillar is structural AI demand. Arya wrote that "the virtuous flywheel of agentic tool adoption by consumers and enterprises, competition between public and private AI labs, and limited chip supply will likely keep AI spending robust for the next several years." A single agent workflow can generate 10 to 1,000 times more tokens than a chat exchange, and any slowdown in frontier model development or addition of safety guardrails would likely increase rather than reduce computing requirements, shifting demand toward faster-growing inference workloads. The bank now forecasts combined capital expenditure by major U.S. and Chinese cloud providers will reach roughly $1 trillion in 2026—about double last year's level—climbing to $1.4 trillion by 2027 and reaching $2 trillion to $3 trillion by the end of the decade.

Each of BofA's five picks carries a specific near-term catalyst. Nvidia is expected to benefit from enhanced share buybacks and a series of GTC developer events, with BofA projecting the company will retain 70% or more of the AI accelerator market over time. Intel stands to gain from what Arya termed "agentic CPU strength" and anticipation of a foundry customer win, as AI agents revive demand for general-purpose processors that coordinate AI workloads; the bank sees the server CPU market expanding from about $61 billion in 2026 to $210 billion by 2030. Micron's catalyst is the initiation of share repurchases beginning December 9. Marvell holds its analyst day on October 6 with focus on the ramp of its custom XPU processors. Lam Research, which makes wafer fabrication equipment, is positioned for share gains across both memory and logic segments as industry wafer equipment spending rises from $156 billion in 2026 to $360 billion by 2030.

Beyond the top five, BofA continues to hold favorable views on Advanced Micro Devices, Applied Materials, Analog Devices and ON Semiconductor, and maintains a buy rating on Broadcom, citing attractive valuation and potential for increased buybacks when the company reports fiscal fourth-quarter results in December.

The BofA call landed on the same day Micron reported fiscal fourth-quarter results that exceeded Wall Street expectations across nearly every metric. The memory-chip maker posted adjusted earnings of $33.42 per share, well above the $31.61 consensus from analysts polled by LSEG, on revenue of $54.23 billion that beat the $51.07 billion estimate—nearly four times the level of the year-earlier period. For the full fiscal year, revenue reached $133.2 billion, approximately 3.5 times the prior year's record, with data center revenue up fourfold, DRAM revenue surpassing $100 billion for the first time and adjusted earnings per share of $75.52, up 811% year over year.

Micron's guidance for the fiscal first quarter also came in ahead of projections, with the company forecasting adjusted earnings of $38.15 per share on approximately $61.5 billion in revenue. Yet the stock barely moved in premarket trading, a stark contrast to its 273% surge since the start of 2026. The muted reaction reflects a market that has already priced in extraordinary expectations. Two details in the outlook reinforced caution: Micron guided fiscal first-quarter gross margin to about 86.25%, below the 87% just reported, and said the quarter would be the floor for fiscal 2027 gross margins, citing roughly $1 billion in higher costs—mostly increased incentive compensation, along with startup and other expenses. The company also plans to raise capital spending, with about $25 billion in the first half of fiscal 2027 alone and construction taking a growing share of the mix.

Analyst commentary reflected the disconnect between results and market reaction. Deutsche Bank's Melissa Weathers, who maintains a buy rating and $1,550 price target, acknowledged it: "We remain highly constructive on the memory sector and MU in particular, with industry fundamentals seeming to be structurally sound for the foreseeable future. While we expect volatility in memory narratives to persist in this AI environment, we believe MU's co-specific drivers and robust supply-demand backdrop can support through-cycle earnings power well above historical trends." Her target implies nearly 46% upside from Wednesday's close.

UBS holds the Street's most aggressive stance, with a $1,625 price target. The firm argued that Micron's gross margins are unlikely to fall below 65% to 70% through the cycle, and that tangible book value per share should approach $360 by calendar year-end 2028. "Ultimately, we think the market will reward this stock with a much higher P/E multiple," UBS analysts wrote.

Goldman Sachs offered a more tempered view, maintaining a neutral rating while raising its price target to $1,250 from $1,100. The firm noted that Micron has signed 10 additional long-term agreements, bringing total coverage to about 35% of expected revenue through fiscal 2030, with roughly 75% of those containing structured pricing frameworks. Goldman believes these agreements are positive for the stock, though elevated investor expectations and modest gross margin pressure may limit near-term upside.

Bank of America's own analysts carry a buy rating and a $1,550 target on Micron, JPMorgan is overweight at $1,540, and Citi is at $1,300. The bank argues that memory is capturing a growing share of AI system costs—a "memory tax" it now puts at more than half of its data center systems market estimate, equivalent to roughly 35% of AI capital spending—and expects memory to capture additional value in 2027 and beyond.

Micron's strategic customer agreements expanded from 16 in the prior quarter to 26, with customer prepayments rising from $22 billion to $32 billion. Remaining performance obligations reached approximately $150 billion, up from about $100 billion. Chief Financial Officer Mark Murphy said the company generated $44 billion in operating cash flow and $33.2 billion in free cash flow during the quarter, and committed to returning 100% of excess cash to shareholders.