Semis bubble? Software cheap? Why BofA’s Subramanian disagrees By Investing.com

Bank of America raised its 2026 S&P 500 EPS forecast to $335, implying 22% year-over-year growth, but kept a cautious index view and maintained a year-end target of 7,100. The firm argues semiconductors still screen attractively despite AI-driven earnings revisions, while software looks relatively weak, with semiconductor free cash flow yields near historical highs and software near multi-year lows. BofA also warned the S&P 500 trades expensive on 16 of 20 valuation metrics and said elevated long-term growth expectations to 16% could be a contrarian negative for the index.

Analysis

The biggest second-order takeaway is not “AI wins” but that the market is still paying for growth twice: once through higher multiples and again through crowded ownership. If semis are back to the top of BofA’s short-term rank while sitting well below prior peak ownership, the trade is likely not finished yet; the cleaner expression is to stay long the capex beneficiaries and fade the funded source of that spend. Software is the more vulnerable leg because its multiple support depends on a growth reacceleration that is not showing up in estimates, so any further AI-infrastructure digestion should continue to compress relative performance over the next 1-3 quarters.

The uncomfortable contrarian point is that broad index downside may be more a valuation air-pocket than an earnings recession. When long-term growth assumptions are elevated, the market becomes fragile to even modest guidance misses; that creates asymmetric downside in the highest-duration pockets rather than the whole index. This argues for being selective on index exposure and preferring companies with visible free-cash-flow conversion over names that need an uninterrupted narrative to hold their multiple.

Positioning matters here: crowded long-only overweights in both semis and software mean the market is still paying for consensus, not surprise. That makes dispersion trades attractive, because the next leg is likely to come from estimate revisions and flow rotation rather than multiple expansion. If yields remain stable and oil stays soft, the near-term risk is not a macro shock but a rotation away from expensive “quality growth” into sectors with cleaner cash returns and less crowded ownership.

Key Decisions for Investors

• Stay long a semis basket vs software for the next 4-8 weeks: long SMCI and/or broad semiconductor exposure, short a software ETF or high-duration software basket. Risk/reward favors 2:1 if AI capex revisions keep flowing, with stop-loss on any broad capex rollback or margin reset.
• Use any strength in software to add shorts rather than chase longs: prefer a 1-3 month put spread on richly owned software names/ETFs to express downside if estimate revisions remain subdued. The trade works best if relative earnings revisions stay negative for another quarter.
• Maintain or initiate a relative-value pair: long semis, short consumer discretionary for a 2-3 month horizon. The thesis is that capex beneficiaries and ‘needs over wants’ will hold up better than discretionary if the labor market softens further.
• Avoid adding to broad index beta until forward growth expectations stop rising: if the market keeps trading on a high PEG regime, upside becomes capped while downside accelerates on any earnings miss. Use index rallies to trim rather than buy.
• For higher-conviction tactical upside, keep APP as a momentum expression but only via defined-risk call spreads into the next earnings cycle; upside is tied to continued AI-adjacent multiple support, while downside is sharp if sentiment shifts away from crowded growth.