Philadelphia Semiconductor Index Forward P/E Drops to Nearly Seven-Year Low
nashnova research
The Philadelphia Semiconductor Index (SOX) plunged 12.2% in a single session, pushing its forward P/E to 13× — the lowest since January 2019 — yet the index is still up 27.2% year-to-date; the real driver of the valuation compression is not falling prices but a 155% surge in AI-fueled earnings forecasts.
Forward P/E at 13× — why isn't this about a price crash?
SOX fell 12.2% on Thursday, dragging its forward P/E to 13× — the lowest since January 23, 2019.
Yet the index is still up 27.2% in 2026. This means → the numerator (price) never actually returned to distressed levels.
The real mover is the denominator: the weighted rolling 12-month EPS forecast for SOX constituents has jumped 155% since the start of the year, driven by AI infrastructure buildout. In plain terms = shares didn't fall that far — Wall Street's profit expectations rose so fast that the price-to-earnings ratio shrank by comparison.
What happened the last time valuations were this low?
The current 13× forward P/E sits right next to the 13.3× recorded on October 14, 2022, when the semiconductor bear market bottomed.
After that trough, SOX roughly tripled over the following two years.
Reflexivity co-founder Giuseppe Sette called that bottom "one of the best semiconductor entry points in a decade" and notes the current valuation is "almost an exact match." This reflects the market anchoring on that historical floor to reassess whether a similar opportunity is forming now.
Compared with the S&P 500, are semis cheap or expensive?
The S&P 500 closed Thursday at a forward P/E of 17.8×; SOX sits at just 13× — a discount of roughly 27%.
Yet SOX constituents carry a projected 2026–2028 sales CAGR (compound annual growth rate — the average yearly increase) of 40.2%, more than four times the S&P 500's projected CAGR of 8.7%.
This means → on a growth-to-valuation basis, the semiconductor sector is in one of its cheapest windows in years.
The core debate: can 155% earnings growth last?
Sette frames it bluntly: "Valuation is no longer the issue — whether 155% EPS growth can persist is the key."
In plain terms = the price tag is already low; the buy-or-pass decision hinges on whether these companies will actually earn what Wall Street projects.
155% EPS growth — sustainable or a mirage?
BULL
Demand locked through 2028
SK Hynix has sold out its 2027 memory capacity and is negotiating 2028 supply — customers locking capacity a year ahead is historically rare.
Bottleneck is supply
The biggest constraint for hyperscalers remains high-end GPU and chip shortages, not energy; power-infrastructure lead times are longer than chip lead times.
Outlook remains solid
Fund manager Hendi Susanto says the semiconductor outlook stays solid through the rest of 2026 and 2027, with a high probability of renewed strength in 2028.
BEAR
Growth rests on assumptions
Bears argue the bull case is built on possibilities and hypothetical scenarios, not confirmed orders.
History doesn't guarantee a repeat
The post-2022 tripling coincided with macro easing; today's conditions may not replicate.
In plain terms = bulls can point to actual orders and locked capacity; bears question whether that growth rate holds two to three years out — the disagreement is not about today but about 2027–2028.
What does this mean for investors?
Valuations have returned to a zone that historically marked major bottoms — but historical similarity does not guarantee a replay.
Whether 155% EPS growth extends into 2027–2028 is the make-or-break test for this valuation thesis.
This means → the core judgment is not "are semis cheap right now?" but "is the AI-driven earnings explosion a cyclical pulse or a multi-year trend?"
市场有风险，内容仅供研究参考，不构成投资建议。