Buy the dip or wait? BofA maps out 3 scenarios for chip stocks and top picks

Bank of America sees a potential recovery in semiconductor stocks in early 2027 as improving U.S. policy clarity, corporate buybacks and relatively attractive valuations help ease concerns over artificial intelligence spending, although investors may need to wait for earnings estimate cuts before the sector finds a durable bottom.

The Philadelphia Semiconductor Index (SOX) has already fallen 37% from its recent peak, compared with an average correction of 32% during previous downturns. The bank said indiscriminate selling driven by fears of slowing AI capital expenditure could begin to run its course by early 2027, with congressional policy developments and share repurchases helping restore confidence.

Semiconductor stocks trade at 13.7 times forward earnings, a 22% discount to the S&P 500's 17.7 times multiple. A return to the historical trough valuation of around 12 times earnings would imply at most another 12% downside, according to BofA. The bank also highlighted buyback support from Lam Research, which has a $25 billion programme, ongoing repurchases by Nvidia and Micron's planned buyback start on December 9.

Previous semiconductor downturns suggest the sector typically declines six to nine months ahead of the fundamental earnings turn, before recovering as earnings estimate reductions become reflected in valuations. The SOX has gained an average of 42% during previous recoveries after those cuts were priced in.

BofA sees the late-June 2026 market peak as a potential signal that earnings estimate reductions could begin in the first quarter of 2027, as companies adjust second-half spending plans. However, the bank cautioned that cheaper valuations alone may not be sufficient to restore confidence, with actual earnings cuts likely needed to establish a sustainable floor.

AI spending remains the biggest swing factor

U.S. cloud capital expenditure is currently running at an annualised rate of about $1.2 trillion, with forecasts pointing to more than $1.5 trillion in 2027. If spending stabilises at current levels rather than accelerating, it would represent a roughly 20% downside against 2027 capital expenditure expectations.

That scenario could reduce planned AI infrastructure deployment to around 30 gigawatts from more than 40 gigawatts in BofA's baseline assumptions. At an estimated $40 billion per gigawatt of investment, the adjustment could create substantial pressure on semiconductor revenue and earnings.

A 20% reduction in projected capital expenditure could translate into more than 20% lower sales and a 35% to 40% earnings-per-share headwind for AI-focused semiconductor companies. Nevertheless, BofA said spending reductions could disproportionately affect future land, power and data centre construction, while newer and more efficient chips, including Nvidia's Vera Rubin platform, remain investment priorities.

BofA outlines three scenarios and top stock picks

Baseline case: The sector faces around 10% further downside before reaching a trough in the first quarter of 2027. The bank favours leading companies across computing, networking, semiconductor equipment, memory and analog chips. Its picks include Nvidia, AMD, Broadcom, Credo Technology, Marvell Technology, Astera Labs, Lam Research, KLA, Micron, Texas Instruments, Analog Devices and ON Semiconductor.

Bear case: A downturn extends into mid-2027 as AI capital expenditure weakens more decisively. Defensive companies, including electronic design automation providers Cadence Design Systems and Synopsys, high-free-cash-flow analog chipmakers Texas Instruments and Analog Devices, and less AI-exposed NXP Semiconductors, could outperform.

Bull case: A sharp recovery in AI investment drives stronger gains in higher-beta memory, optical networking, quantum computing and CPU stocks. BofA highlighted Micron, Credo Technology, Marvell Technology, AMD and Arm among potential beneficiaries.

Despite the recent selloff, BofA believes semiconductor valuations already reflect considerable scepticism about the AI investment cycle. The sector trades at a 22% discount to the broader market, compared with a 15% premium during the AI-driven rally. The key question for investors is whether earnings revisions and greater visibility on capital expenditure will provide the catalyst for a recovery, potentially shifting market attention towards a renewed AI spending cycle in 2028.