Whether a two-day surge in chip stocks marks a genuine trend change or just a dead cat bounce cannot be judged from price action alone. The Philadelphia Semiconductor Index (SOX) hit an all-time high of 14,655 on June 22, 2026, then slid to around 10,896 within two weeks — roughly 26% off its peak, entering a technical bear market — before rebounding on bargain hunting.
This rebound could be a real bottom, or simply a pause in selling. The signals that separate the two lie in data, not intuition. According to TradingKey, after short-term oversold conditions the SOX has historically rebounded an average of over 3% within five trading days and over 6% within a month. In other words, a sharp bounce by itself is no proof of a turn.
The truth of a rebound is decided not by price but by capex guidance: if the numbers hold, it is a dip; if they crack, it is a retracement.
Where the SOX stands — start by calculating the drawdown
The starting point is distinguishing a simple dip from broken structure. Conventionally, a 10% drop from the high is a correction and a 20% drop is a bear market. Applying −20% to the June 22 high of 14,655 gives a bear-market threshold of 11,724 — yet the two-week low of 10,896 sat below even that line.
| Metric | SOX level | vs. peak |
|---|---|---|
| All-time high (Jun 22) | 14,655 | — |
| Bear-market threshold (−20%) | 11,724 | −20.0% |
| Recent two-week low | 10,896 | −25.7% |
| Typical oversold bounce (5 days) | — | avg. +3% |
The math shows this decline pushed more than five percentage points past the bear-market threshold. A sharp bounce from this zone is statistically common and does not, on its own, confirm direction. That is why the three confirmation signals below matter more than the index level.
Dead cat bounce vs. trend change: how to tell them apart
The way price rises reveals its character. A dead cat bounce tends to lift only the most beaten-down names on thin volume and gives back the gains within days. A true trend change comes with rising volume and builds zones that hold even when bad news lands. The order of checks:
- Volume: does the rebound carry heavier-than-usual volume, or is it a thin-volume pop?
- Post-earnings reaction: repeated "sell on the news" — a stock falling despite good results — signals a heavy ceiling, because expectations are already priced in.
- Positioning: if speculative money such as hedge funds has already cut positions substantially, selling pressure may be spent.
- Fundamental confirmation: does the rebound line up with improving earnings and investment metrics? Here, capex guidance is decisive.
Only when all four align in one direction can a trend change be discussed. If even one diverges, guarding against a retracement is the reasonable stance. The higher the valuation, the colder the earnings reaction — the same structure covered in when multiples betray you.
Why capex guidance is the key
The ceiling of the chip earnings cycle ultimately depends on how much Big Tech pours into AI data centers. What the market now watches in earnings is not the revenue or profit surprise itself, but whether hyperscalers keep or raise their capex plans. According to Value Add Pulse, combined 2026 capex guidance from Amazon, Microsoft, Alphabet and Meta totals roughly $725 billion — about 77% higher than last year's $410 billion.
| Company | 2026 capex guidance | Note |
|---|---|---|
| Amazon | ~$200B | largest single commitment |
| Microsoft | ~$190B | about +61% YoY |
| Alphabet | $180–190B | raised guidance |
| Meta | $125–145B | raised guidance |
| Four-company total | ~$725B | $410B → +77% |
If these numbers hold or grow, revenue expectations across the AI value chain — chips, power infrastructure, optical networking — are supported. The upward revision of the 2026 global semiconductor sales forecast to $1.3 trillion fits the same picture. If capex growth shows signs of cracking, the recent rebound could reverse. The HBM angle continues in the AI capex cycle check.
What to watch
These indicators point to direction earlier than the size of the bounce. A checklist to cross-reference through earnings season:
- Whether Big Tech keeps or raises capex guidance — any downgrade lowers the bar across the value chain.
- Whether stocks rise after strong results or fall on "sell the news".
- Whether the rebound carries volume — thin-volume surges are less reliable.
- Whether the SOX recovers above the bear-market threshold (about 11,724) and holds.
- Whether rate moves pressure valuations through the discount rate — read alongside rates and the discount rate.
Writing down the conditions under which these signals would flip gives you a reference when conviction wavers. For instance, if capex holds yet chip stocks keep falling, the variables to re-check are rates and positioning, not earnings.