# AI spending hangover hits Asian chipmakers — short the semis as the $2T rally cracks

_AI-generated trading idea · SHORT · AMD, NVDA, SMH_

> Canonical page: https://commonquant.ai/research/for-you/ai-spending-hangover-hits-asian-chipmakers-short-the-semis-a--3b44aaad-cf73-4909-9d40-10b5dea1e955

Asian chip stocks are cratering as the global AI spending euphoria starts to fade and rumors of new competition circulate. After adding $2 trillion in value just last quarter, these stocks are priced for perfection and extremely vulnerable to the current negativity.

## Idea

Chip stocks just had a historic quarter, adding $2 trillion in value as the AI boom expanded beyond Nvidia. But the mood has completely shifted: Samsung and SK Hynix both plunged over 7% overnight as a US tech sell-off spread to Asia, driven by fears that Meta selling computing power signals an oversupply in AI capacity. When a sector gets this crowded and heavily bought, any negative news can trigger a domino effect. The combination of overvaluation after a record quarter and sudden fundamental doubts about AI capacity creates a high probability of further near-term selling pressure.

## Advanced Analysis

### Verdict: The setup is half-built — wait for the washout day before committing

The idea argues the AI chip rally is cracking after a $2 trillion quarter, per the CNBC piece on the record rally, and the tested rule set is built to buy the panic: enter long SMH after a one-day drop greater than 2% while the 14-day RSI holds at or above 55. The strongest point for the trade is the current profile — SMH closed at $574.29 with RSI at 59.7, sitting 14.1% below its range high, exactly the strong-regime-but-stretched shape the setup wants. The strongest point against is the entry economics: over 60 months the completed backtest produced only a 4.2% return across 16 trades with a 43.8% win rate and a 10.1% maximum drawdown, with daily-bar exits that may overstate fill quality. Fundamentals add a second tension: NVDA's latest SEC-reported quarter (period ended July 26, 2026) showed revenue up 17.9% sequentially to $96.2B at a 75.0% gross margin, yet the June 30, 2026 ownership filings show net open-market insider selling of about $565.4M at NVDA and $153.2M at AMD — backward-looking, but a caution flag for the next filing window around end-September 2026. Right now only half the trigger is in place (RSI is met; the one-day move is -0.39% versus the required 2% drop), so the disciplined action is to wait for SMH to close below roughly $562.80 while RSI stays above 55 — and stand down entirely if RSI breaks the floor first.

#### Conviction score breakdown

Composite score computed by the server from the applicable evidence-tier dimensions.

| Measure | Value |
| --- | ---: |
| Thesis support | 62/100 |
| Trade readiness | 45/100 |
| Risk quality | 55/100 |
| Backtest evidence | 40/100 |
| Fundamentals trend | 58/100 |
| Score | 52/100 |
| Composite Score | 52/100 |
| Evidence Tier | backtested |

#### Decision scenarios

Bull, base, and bear cases synthesized from the cited evidence tier. Likelihoods are rounded evidence-weighted judgments, not statistically calibrated forecasts.

| Measure | Value |
| --- | ---: |
| Evidence Tier | backtested |

### Trade now: SMH fade is armed but not triggered — one down day away

The strategy is a contrarian long on SMH: it buys the ETF on a daily close down more than 2% while the 14-day RSI is still at or above 55, then works a 4% stop and a 10% target. As of the latest close of $574.29, half the setup is in place — RSI sits at 59.7, comfortably above the 55 floor. What is missing is the crack itself: the one-day change is just -0.39%, versus the required decline of more than 2%. Another roughly 1.6% down day from here would put both conditions in place.

So waiting means something concrete today: no position yet. Watch for SMH to close down more than 2% (roughly below $562.80 from the current close) on a day when RSI has not already broken below 55. If both conditions are met, the plan is mechanical: stop 4% below entry (about $18 lower at current levels), target 10% above (about $57 higher), roughly 2.5-to-1 reward-to-risk per trade, sized at 2.5% risk with a 25% maximum position.

The completed backtest supports the discipline rather than excitement: over 60 months the strategy traded 16 times, returned 4.19%, won 43.75% of trades, and saw a 10.1% maximum drawdown. The best window (24 months) showed 4.58% over 5 trades at a 60% win rate and only a 2.9% drawdown — the pattern's edge comes from the occasional 10% winner, so missing the entry or chasing after the bounce has started is costly. One caveat on exit quality: backtest stops and targets were filled on daily bars, not intraday, so real fills may be slightly worse.

The idea's macro thesis — fading the AI rally after a $2 trillion quarter of gains — is directionally short semis, but this rule set expresses it as buying sharp single-day washouts in a still-strong regime. Respect the distinction: if RSI collapses below 55 before the 2% down day arrives, the setup expires and you stand down rather than improvise.

#### AMD price and trigger map

Uses the idea timeframe and keeps price levels on the price axis.

| Measure | Value |
| --- | ---: |
| Ticker | AMD |
| Timeframe | 1d |

#### NVDA price and trigger map

Uses the idea timeframe and keeps price levels on the price axis.

| Measure | Value |
| --- | ---: |
| Ticker | NVDA |
| Timeframe | 1d |

#### SMH price and trigger map

Uses the idea timeframe and keeps price levels on the price axis.

| Measure | Value |
| --- | ---: |
| Ticker | SMH |
| Timeframe | 1d |

### Why the bear case has real support

The proprietary long-side rule set — entering after a \>2% one-day drop with RSI at or above 55 — is precisely the setup that gets run over in a genuine momentum reversal, and its own performance record (4.2% return, 43.8% win rate over 16 trades in 60 months) shows it is a low-conviction mean reversion, not a durable long signal. That weakness on the long side is indirect…

#### AMD Return on equity

Return on equity trend from CommonQuant fundamentals/XBRL data; +100.2% from first to latest point.

| Measure | Value |
| --- | ---: |
| 2008-12-27 | \-24.37795275590551% |
| 2009-12-26 | \0.4521604938271605% |
| 2010-06-26 | \0.2845744680851064% |
| 2010-06-26 | \-0.05718085106382979% |
| 2010-09-25 | \0.15635179153094464% |
| 2010-09-25 | \-0.19218241042345272% |
| 2010-12-25 | \0.46495557749259625% |
| 2011-04-02 | \0.32734274711168165% |
| 2011-07-02 | \0.03717245581962218% |
| 2011-10-01 | \0.055619266055045864% |
| Latest Value | \0.055619266055045864% |
| Change Pct | \100.22815396605267% |
| Ticker | AMD |
| Timeframe | reported periods |

#### AMD Debt to equity

Debt to equity trend from CommonQuant fundamentals/XBRL data; -86.3% from first to latest point.

| Measure | Value |
| --- | ---: |
| 2009-12-26 | \6.561728395061729 ratio |
| 2010-06-26 | \3.2154255319148937 ratio |
| 2010-09-25 | \3.5586319218241043 ratio |
| 2010-12-25 | \2.1599210266535045 ratio |
| 2011-04-02 | \1.4069319640564828 ratio |
| 2011-07-02 | \1.3375990249847654 ratio |
| 2011-10-01 | \0.900802752293578 ratio |
| Latest Value | \0.900802752293578 ratio |
| Change Pct | \-86.27186774491442 ratio |
| Ticker | AMD |
| Timeframe | reported periods |

#### AMD sector percentile check

Ranks AMD against 791 companies in its sector using CommonQuant fundamentals.

| Measure | Value |
| --- | ---: |
| Free cash flow | \98.73577749683945th percentile |
| Operating margin | \87.11943793911007th percentile |
| Revenue growth (YoY) | \77.53807106598984th percentile |
| Rnd Intensity | \65.12301013024602th percentile |
| Ticker | AMD |
| Sector | Information Technology |
| Peer Count | 791 |

### Scores

- **Conviction score breakdown:** 52
- **Thesis support:** 62
- **Trade readiness:** 45
- **Risk quality:** 55
- **Backtest evidence:** 40
- **Fundamentals trend:** 58

### Watch items

- **SMH — ROC (1), one-day change**
- **SMH — RSI (14)**
- **SMH — RSI (14) regime break**
- **NVDA — ROC (1), one-day change**
- **NVDA — RSI (14)**
- **NVDA — Insider net open-market selling (13F/ownership filings)**

## Key details

- Symbols: AMD, NVDA, SMH
- Timeframes: D1
- Tags: \#semiconductors, \#mean\_reversion, \#tech

## Community

- Upvotes: 102
- Views: 2,638
- Copies: 0
- Cosigns: 0

## News sources

- [Samsung Electronics, SK Hynix shares tumble over 7% as chip rout spreads from Wall Street](https://www.cnbc.com/2026/07/02/samsung-sk-hynix-shares-slide-kospi-tech-selloff-nasdaq.html) — CNBC
- [Record chip rally adds $2 trillion in combined value to Micron, Intel and AMD in second quarter](https://www.cnbc.com/2026/06/30/ai-chip-rally-in-q2-adds-2-trillion-in-value-to-micron-intel-amd-.html) — CNBC
- [US Tech Rout Sours Mood For Japan, South Korean Chip Stocks](https://www.bloomberg.com/news/articles/2026-07-02/us-tech-rout-sours-mood-for-japan-south-korean-chip-stocks) — Bloomberg

## Discussion (1)

**@rough\_keeper** · 1 upvotes

For AMD, the key is whether the 1d structure stabilizes after 2026-07-02.

## Related ideas

- [Night movers — Sep 24: Tech and telecom gains offset logistics and retail losses](https://commonquant.ai/research/for-you/night-movers-sep-24-tech-and-telecom-gains-offset-logistics--ff39f096-0b7c-5471-b913-5c50e4499796)
- [Night movers — Sep 21: Chip and communication services surge while energy lags](https://commonquant.ai/research/for-you/night-movers-sep-21-chip-and-communication-services-surge-wh--6fcd8852-c59a-5347-863b-a3f51ccb0eff)

## Related

- [AMD trade ideas](https://commonquant.ai/markets/amd)
- [NVDA trade ideas](https://commonquant.ai/markets/nvda)
- [SMH trade ideas](https://commonquant.ai/markets/smh)
- [Latest market news](https://commonquant.ai/news)

## About CommonQuant Research

Ideas and Advanced Analysis are generated with fresh, LLM-selected news headlines and grounded in SEC XBRL fundamentals and peer percentiles. Strategies built from these ideas are automatically backtested (over a timeframe-dependent historical window with walk-forward tuning) and stress-tested before they can go live, then monitored against the news hourly while they run.