The Week Seven Walls Named Themselves

Last week named six walls. This week named seven more.

The week opened with Samsung and SK Hynix committing $500 billion to a Korean chip hub. It closed with Michael Burry calling that same commitment 'the beginning of the end.'

In between, the global chip rout spread from Wall Street to Asia, the AI model approval framework went industry-wide, SoftBank had to guarantee a $10 billion loan its OpenAI shares alone could not secure, and the jobs report landed 50% below consensus.

For 50 years, technology got cheaper while everything that actually matters - your doctor, your house, your kid's tuition - went up 4,000%. Nobody could fix it. Until NOW.

A 41-year stock market legend says AI is about to do to healthcare, housing, and education what it already did to your TV.

"Prices won't just drop. They will COLLAPSE. And investors who understand why could make a fortune." For the full story, click here.

The profit-taking started Wednesday. The VanEck Semiconductor ETF dropped 5.2%. Teradyne (TER) and KLA (KLAC) fell 13%. Sandisk (SNDK) shed more than 10%. By Thursday's Asian open, the selloff had gone global. Samsung fell more than 7%. SK Hynix sank more than 9%. Kioxia tumbled 13%. South Korea's Kospi dropped nearly 8% and triggered a trading halt. The Roundhill Memory ETF headed for its worst week since inception.

Jefferies then published the concentration data. Ten stocks drove 78% of the S&P 500's first-half return. The chip complex dominated the list. When the top ten contributors concentrate in one sector and that sector sells off, the index math moves against itself simultaneously.

Guggenheim upgraded Salesforce (CRM) and ServiceNow (NOW). Capital moved from chip leaders toward software laggards in real time.

The Signal

Run your most chip-heavy position against a documented 5.2% single-session SMH drop. If it clears your hurdle, you hold a chip bet with known parameters. If it needs the rally to continue, you hold concentration.

Michael Burry posted Tuesday. He added Tesla (TSLA), Caterpillar (CAT), Applied Materials (AMAT), and the SOXX ETF to his existing Nvidia (NVDA) and Palantir (PLTR) shorts. He explicitly named Samsung and SK Hynix's $500 billion Korean chip hub as the trigger.

The Caterpillar short is the most analytically precise. Caterpillar climbed 86% in the first half and now trades at 39 times forward earnings against 13 times three years ago. Burry is betting the same AI infrastructure thesis that re-rated Caterpillar breaks it. His SOXX puts expire in March, requiring roughly a one-third decline from peak. He is sizing for a cycle.

Palantir is down approximately 40% since he announced that short.

The Signal

Pull your chip-equipment exposure specifically. Applied Materials and Lam Research (LRCX) are both inside Burry's frame. Size them accordingly.

Confirmed by satellites 300 miles above the Earth's surface... Elon Musk is rolling out a breakthrough technology that could replace our need for foreign oil and ignite a $10 trillion boom a small group of stocks. Click here to learn how you can invest in this before it becomes mainstream.

The Fable shutdown lasted two and a half weeks. Commerce Secretary Lutnick approved Fable 5 Tuesday after Anthropic agreed to address the Amazon-identified workarounds. Susie Wiles said the two sides worked together "in a way we have never seen before."

Within 48 hours, Washington was drafting voluntary industry-wide standards for frontier model releases. Google (GOOGL) is already at the table. OpenAI publicly thanked the administration for working on "a durable framework." The Fable resolution became the operating template for every US AI lab.

Amazon (AMZN) flagged the Fable vulnerability. Amazon also holds $13 billion in Anthropic equity and could add $20 billion more.

The Signal

Run the exit case for any pre-IPO AI position with the next release slipping one quarter under review. If the return still clears, hold. If it needs a clean release calendar, you are underwriting government permission, not company execution.

SoftBank tried to borrow $10 billion against its OpenAI stake from Goldman Sachs (GS), JPMorgan (JPM), and Mizuho. The banks said the shares were not enough. SoftBank added a corporate guarantee.

Private AI stock failed as standalone loan collateral. SoftBank owes $40 billion on a bridge loan due March 2027. The guarantee turns a valuation loan into a balance-sheet loan.

The Signal

Marks are opinions until someone lends against them. For any position that values private AI stakes, check the structure. A guarantee means the lender does not trust the number either.

Landmark Executive Order 14241 Unleashes

Republican or Democrat – whether you support or oppose Trump’s New Dollar – every American could soon be forced to use it

Discover three critical moves to help you prepare, before it’s too late

Meta Platforms (META) announced it is building a cloud business to sell excess computing power. The stock jumped 10%. CoreWeave (CRWV) fell 14%. Nebius Group (NBIS) fell 17%.

SpaceX (SPCX) preceded it. SpaceX signed Anthropic to $1.25 billion per month and Google to $920 million per month for compute. Meta is following the same playbook at $145 billion in annual capex scale.

Palantir CEO Alex Karp called the Anthropic-OpenAI token model broken on CNBC the same week. Palantir gained 9%.

The Signal

The neocloud framework now carries a fourth hyperscaler competitor. Price-test any CoreWeave or Nebius exposure against a fourth well-capitalized seller entering the market.

The BIS, IMF, ECB, Bank of England, and Bank of Canada all voiced AI concentration concerns at the same venue within three days. The IMF's top financial stability official named the leverage on both sides of the AI trade as his biggest worry.

Apollo's (APO) chief economist presented two Sintra recession scenarios. AI succeeds and consumer spending collapses as jobs disappear. AI fails and $700 billion in investments prove worthless. Macklem named the dot-com bubble directly. AI-linked debt accounts for nearly half of all investment-grade issuance this year. 87% of all venture capital is going to AI.

Warsh's Sintra debut Wednesday was the parallel institutional read. He said prices are too high. He declined to signal July. Kevin Hassett called raising rates a macroeconomic mistake on Fox Business the same day.

The Signal

Watch whether the Fed, ECB, or Bank of England references the BIS annual report in official policy communication. A formal citation converts the Sintra warnings from advisory to binding supervisory reference.

It's not semiconductors, AI chips or quantum computers. But none of those technologies can exist without it. On July 1st, 2026, Trump is expected to ban exports of something every tech company desperately needs—forcing them all to relocate to U.S. soil. See what he's about to ban here…

The jobs print landed 50% below consensus at 57,000. The unemployment rate dropped to 4.2%. The miss took near-term pressure off the July hike case.

JOLTS Tuesday had told the nuanced story. Openings beat at 7.594 million. But hiring fell 45,000. Layoffs rose. The gap between workers who say jobs are plentiful versus hard to get dropped to its lowest since early 2021.

The Signal

ADP missed at 98,000 the same week. A second consecutive ADP miss before the next payrolls print would name the labor market softening as a trend. The July 28 meeting is the next binding event.

Seven sequences. One holiday-shortened week. Each named a wall the second half must now price through.

These are not seven separate stories. They are one story told seven ways. The buyer pool for the largest AI offerings in private-market history is contracting on every structural dimension that was supposed to anchor it. The question is which assumption your position depends on, and whether it survives the wall behind it.