Friday, July 24, 2026 ALERT · Level 3 5 min read

The Last Safe Corner of the AI Trade Just Cracked

Chips fell 4.4% while the S&P 500 rose: the bull case that semiconductors are protected by AI spending just lost its evidence.

The risk read for this session

Sector crash

REALISED

chips −20% from peak · within 1 month

Unchanged from the previous read

Deeper crash

31%

chips −30% from peak · within 3 months

Up 1 point from 30%

Market crisis

S&P 500 −20% · within 3 months, AI at the centre

Unchanged from 8%

SOXX

527.01

−19.7% from peak

NVDA

206.84

−12.6% from peak

VIX

18.58

fear gauge

High-yield spread

268 bp

credit stress

These are model estimates, not forecasts anyone can guarantee, and not investment advice. The dot before each probability is its band: 🟢 low · 🟡 moderate · 🟠 high · 🔴 very high, with thresholds set separately for each scenario. SOXX is the main US semiconductor ETF, the basket I track; from peak is the fall from the highest close of this cycle; VIX is the market’s fear gauge; the high-yield spread is the extra interest riskier companies pay to borrow, my early-warning line for funding stress. My internal alert is at Level 3, on a scale where 3 is the highest.

THE ONE-LINE TAKEAWAY

This week the market found a way to sell both sides of the AI trade. Thursday it hit the companies that PAY for AI (the cloud giants, on soaring capital spending). Friday, July 24, it hit the companies that BUILD it — the chipmakers. SOXX, the main U.S. semiconductor ETF (a basket of chip stocks), fell 4.4% to 527.01, its worst session in weeks, led by memory chips (−8%). The bull case that "chips are protected, because AI spending is their revenue" just lost its evidence.

Today I am 🟠 ALERT, and my internal alert stays at LEVEL 3 (its highest) for a 7th day — because a level only steps down after a full recovery checklist, and none of it is met yet.

New here? Start with this. I'm Andrea Pilotti, and this is AI Market Observatory. I invest my own money in AI and semiconductor stocks, and I work with artificial intelligence for a living — I built this because I needed to measure the risk in my own exposure instead of guessing at it. Every trading day I track these stocks with data and turn what I find into a single risk read: a traffic light for the odds of a sharp fall in this corner of the market. This is an observatory, not a recommendation service, and I am not a financial advisor: my job is to notice trouble early and say it in plain language. Every daily read stands on its own, this one included.

The traffic light, once: green = normalyellow = a down day, no change in the pictureorange = alert, tension risingred = critical

Alongside it I run an internal alert from Level 1 to Level 3, where 3 is the highest and only steps down once a full recovery checklist is met. How every gauge and threshold is defined is public in my methodology, and every previous daily read is in the archive.

THE TELL: THIS WASN'T A CRASH

Here is what makes Friday interesting: the S&P 500 (the broad U.S. market) rose +0.05%, the Dow gained 0.5%, Apple jumped 3.5%, and the fear gauge (the VIX, which measures expected market swings) fell to 18.58, below the calm line of 20. This was not a market panic. It was a quiet, orderly rotation OUT of AI and into everything else. That is exactly what my "capital relocation" signal is built to catch — and it flipped to alert today.

WHY IT FELL

The trigger was a memory-chip selloff that started overnight in Korea (positioning before SK Hynix reports earnings on July 29), plus worry that rising memory prices are squeezing margins — Apple and Microsoft have both raised some prices to offset memory costs. Analysts largely read it as profit-taking after this week's rebound, not as a break in demand: data-center demand for chips still outstrips supply. Even Intel (+12% after hours on Thursday's results) gave the whole gain back, −7.9%.

THE SIGNAL BOARD (🚦 my 9 gauges, in 6 families)

🟠 Moving averages: SOXX now sits 7.4% below its 50-day average (the trend's "waterline") for an 8th day, and today on above-average volume — which upgrades this gauge to alert.

🟠 Structure: SOXX broke a minor support (536) but the major floor (521.81 / the −20% line at 524.76) held on the closing basis. This is the level that matters now.

🟠 Momentum: the weekly trend indicator (MACD) has now CONFIRMED a bearish crossover at the weekly close — no longer "forming."

🟠 Breadth: inside my AI universe only 12% of names rose. But the broad market's breadth was fine (equal-weight beat cap-weight) — it is AI specifically that is weak.

🟠 Rotation & canaries: money rotated out of AI (SOXX −4.4% vs the S&P +0.1%); Oracle made a fresh 52-week low. My 5-stock "canary" basket still beats SOXX over 20 days, which keeps this from going red.

🟠 Distribution: another heavy-selling day; 11 such days are now active — a heavy count.

🟢 Volatility: the VIX is calm (18.58) — but calm while chips crash is complacency, and I flag it.

🟢 Credit: the extra yield risky companies pay to borrow (high-yield spread) is steady at 268 basis points. No stress. I am watching Oracle, an indebted AI-linked name.

🟠 Euphoria (out of the score, a fragility flag): SOXX is still 32.6% above its 200-day average — stretched, though less than the +39% of two days ago.

Three families are in tension (trend, participation, volume), one of them a leading indicator — that is my definition of 🟠 ALERT. It is NOT critical: SOXX is still well above its long-term 200-day average, with no "death cross," and credit is calm.

THE LEVELS (📏 where I am)

SOXX 527.01, down 19.66% from its June peak of 655.95 — back at the doorstep of the −20% line (524.76) after the week's rebound fully unwound. Nvidia (the flagship) held up: −0.9% to 206.84, down 12.6% from its peak, still above its long-term average. The broad market (S&P 500) is only 2.6% below its record.

THE ODDS (📊 my estimates; bands: 🟢 low · 🟡 moderate · 🟠 high · 🔴 very high)

🔴 Sector correction — chips −20% from the peak: ALREADY TRIGGERED (first hit July 17, still inside the 30-day window).

🔴 Deeper crash — chips −30% within 3 months: 31%, up from 30%.

🟡 Broad-market crisis — S&P 500 −20% within 3 months, with AI at the center: 8%, unchanged.

Why only a small move up: the drawdown deepened and momentum confirmed lower (that lifts the odds), but the major floor held on modest volume, the cushion above the −30% line is large, credit is calm, oil fell, and demand looks intact (that caps it). My golden rule: bad news + levels that hold on ordinary volume = more noise than signal.

CREDIT & MACRO (💳)

Credit calm (high-yield spread 268 bp). 10-year Treasury yield 4.69%, 2-year 4.33%. Oil fell ~3-4% (Brent near $97-98) on Middle-East de-escalation — a relief for the inflation-and-rates channel. The Federal Reserve meets July 28-29 (markets price ~64% no change, ~36% a hike): the hottest two-way catalyst on the calendar.

NEXT 7-14 DAYS (👀)

Monday July 27: a new low-cost Chinese open AI model (Kimi K3) is expected to release its weights — a possible "DeepSeek-style" trigger. For context, in January 2025 DeepSeek's model wiped ~$600B off Nvidia in a single day (−17%), and chips (SOXX) fell 7.8% on 5-7x normal volume — then largely recovered within two weeks (SOXX's total drawdown was only −9.9%). The lesson from that episode: such shocks are violent for a day, my volume rule tells signal from noise, and discipline on levels beats prediction. This week also brings the Fed (28-29), SK Hynix earnings (29), and Microsoft/Meta/Amazon (29-30) — where cloud capital spending, my single most important variable, gets its decisive test after Alphabet was punished for raising it.

WHAT WE'RE WATCHING TO PROTECT CAPITAL (🛡️)

My first job is protecting capital, not maximizing returns. The one level that matters: SOXX's major floor at 521.81 / 524.76 on a closing basis. While it holds, this is more noise than signal. A close below it — especially on heavy volume (1.5x average or more) — would be the confirming signal. Decisions are yours, ideally with a licensed advisor.

SOURCES (📚)

Market data: Massive/Polygon (adjusted daily series and grouped market data), including the January 2025 DeepSeek series. Index levels (S&P 500, VIX): public end-of-day recaps for July 24, cross-checked (24/7 Wall St., AP, Yahoo Finance, Washington Post). Credit (high-yield spread): FRED. Treasury yields, oil, Fed odds, news: public sources (etftrends, Bloomberg, coingape/centralbank.watch). More about my method: https://aimarketobservatory.com/

Who I am. I'm Andrea Pilotti. I invest my own money in AI and semiconductor stocks, and I work with artificial intelligence for a living. I built this observatory because I needed to measure the risk in my own exposure instead of guessing at it — the agents that run it every morning are something I engineered, not a company standing behind me.

What this is for. One thing only: risk management, not returns. I publish the same measurements every trading day, with the method in the open, so that anyone can see when the picture changes and decide for themselves what, if anything, to do about it.

⚖️ What this is not. I am not a financial advisor and I am not licensed. Nothing here is a recommendation to buy, sell or hold anything, and none of it is investment advice. These are calibrated model estimates, not certainties: over short horizons noise dominates, and this sector can invalidate any estimate in a single session. Every decision is yours, ideally with professional guidance.