The uncomfortable truth first: Intel has now surged more than 300% from its 2025 low of $17.67, hit an all-time high of $142.32 in June, pulled back to the low $80s in July, and is now at $89 with this week's 11% move adding another leg. At every step of that run, the same debate has repeated. Is this a genuine foundry turnaround or a series of headline-driven pops on a business still losing money? Thursday's catalyst, the expansion of Intel's foundry toolchain for AI chip customers, is the latest entry in that debate. It is also one of the more substantive ones. Here is what actually happened. Intel expanded its purpose-built silicon business beyond networking and IPUs through the Fortinet Security Processor 6 collaboration. The company confirmed 18A-P has entered risk production on sched

Stop calling it a bounce. A bounce is what happens when nothing changed. What happened Thursday is different. Four separate entities, Samsung, Microsoft, Amazon, and Apple, each independently confirmed within 24 hours that the AI memory shortage is real, worsening, and extending further than the market had priced. When the buy-side was selling memory stocks on CXMT IPO fears and AI capex peak anxiety, the sell-side of that trade just had its thesis demolished by the four largest technology companies on Earth. Here is the full picture. Samsung reported its highest-ever quarterly revenue with operating profit of 89.5 trillion won, ahead of expectations. More importantly, it warned that memory supply constraints may persist into 2028 and signed multi-year supply agreements with major data cen

Tonight is the most loaded single session of the 2026 earnings season. At 2pm ET the Fed announces. At 2:30pm Warsh speaks. After the close, Microsoft and Meta report. These are not independent events. They form a chain: the Fed sets the discount rate applied to every future dollar of AI spending, and then Microsoft and Meta immediately show whether those future dollars are materialising fast enough to justify the capex. If the Fed holds and sounds balanced, and both MSFT and META beat, the AI trade gets its first clean confirmation in months. If any one of the three goes wrong, the dominoes fall in sequence. Test One — The Fed at 2pm ET The base case is a hold at 3.50 to 3.75%. CME FedWatch currently assigns 64% probability to no change. But this is the second-most uncertain FOMC meeting

The uncomfortable answer first: this is probably both, and deciding which one dominates depends entirely on what August 5 delivers. SanDisk has now fallen roughly 31% in a month from its June all-time high, including Friday's 10.79% single-session drop that led the entire memory complex lower. The stock is at $1,471 premarket, up 2.47% as buyers step back in. Options traders are pricing a 25% move in either direction on August 5 earnings. Goldman Sachs has a $2,200 target with a Buy rating. The bear case targets $1,027 if NAND pricing rolls over. That is a 24x spread between bull and bear, which means the market has no consensus on what SanDisk actually is: AI infrastructure compounder or cyclical memory stock at the wrong point in the cycle. What Caused Friday's Drop There was no company-

The week of July 21 to 25 is the single most instructive week in the memory supercycle so far. Not because of the Tuesday gains. Because of what happened on Friday. Understanding both moves together tells you more about how to trade this sector than any analyst note written this year. Here is the full sequence. Memory stocks fell 25% across July to their lows. Monday July 20 saw a 4 to 6% snap-back as no fresh Korea headlines appeared over the weekend. Then Tuesday exploded: SNDK surged 14.27%, SKHY ADR jumped 13.75%, MU climbed 12.17%, WDC was up 12%, the DRAM ETF gained 11%. The catalyst was a Morgan Stanley report forecasting a 25% memory price increase from Q2 to Q3, with the firm explicitly stating it was buying the dip. BofA called the slump a summer reset. Semiconductor ETFs absorbe

The setup going into tonight is more loaded than any Intel earnings in recent memory, and not just because the stock is up 278% year to date. Two days ago Intel landed its first publicly named external foundry customer under CEO Lip-Bu Tan, Fortinet's SP6 security chip on Intel 4 process, and the stock surged 8.64%. It is now sitting at $102 to $105, down 2.68% today as the market takes profits ahead of the print. Options traders are pricing a 13 to 15% move in either direction. That is the 94th percentile of implied volatility over the past year. The market is treating this as a genuine binary, and it is right to do so. The reason is simple. Intel's transformation under Lip-Bu Tan is either real and beginning to compound, or it is still a restructuring story without foundry economics to b

Stop thinking about the 480,126 deliveries. The market has had three weeks to process that number. TSLA is still trading below its pre-delivery report levels at $376 to $379. That one fact tells you everything about what tonight's print actually needs to deliver: not more evidence that Tesla can sell cars, but evidence that selling 480,000 cars at $25 billion in capex does not permanently destroy the margin structure that justifies trading at 175 to 180 times forward earnings. Tonight's call is entirely about three numbers. Automotive gross margin. Free cash flow. Cybercab production rate. The first two tell you whether the car business is being sacrificed to fund the future. The third tells you whether the future is arriving on any recognisable timeline. What the Estimates Actually Say Th

The most important sentence published this week about SpaceX did not come from JPMorgan. It came from Morgan Stanley, which issued a Buy rating with a $300 price target and simultaneously disclosed that SpaceX faces a $672 billion funding gap with no projected free cash flow until 2035. Jim Chanos called that disclosure "truly glorious." He is correct. When your own bull case requires acknowledging a $672 billion capital hole and a decade without positive cash flow, the bull case needs more than a price target. It needs a compelling argument for why the market should fund that gap at current multiples. SPCX has fallen 35.6% from its all-time high of $225.64. It breached $150 this week, briefly traded below the $135 IPO price on July 15, and is currently trading around $123 to $139 dependin

The honest framing first: what happened on July 16 and 17 is not the same thing as what happened in early July. Last week's selloff was profit-taking, SK Hynix listing rotation, and macro noise. This week's selloff has a specific, identifiable cause, and that cause has direct implications for the memory thesis that the prior selloffs did not. The Bank of Korea raised its benchmark rate 25 basis points to 2.75% on July 16, its first hike since January 2023. That triggered a 6.48% KOSPI collapse, a circuit breaker halt, SK Hynix down 11.53%, Samsung Electronics down 8.23%, SKHY down 9% in US trading, MU down 8% below $910, and SNDK down 8% to $1,487. What the Bank of Korea Actually Did and Why The BOK's decision was not a surprise to economists. South Korea's CPI hit 3.1% in May 2026, meanin

What actually happened this week is three separate things colliding at once, and they need to be separated before you make any positioning decision. What Actually Caused the Drop The BiCS10 announcement had almost nothing to do with the selloff. The new 10th-generation 3D NAND chip launched on the same day the stock fell 14%. The market did not sell SanDisk because the product is bad. It sold because a stock up 858% year to date has no margin for error when sentiment shifts, regardless of what is on the press release. Three things hit simultaneously. The June jobs report printed 57,000, well below expectations, with prior months revised down 74,000 combined. Weak jobs data raises the question of whether AI capex cycles slow. That question, even when premature, is enough to trigger profit-t

The uncomfortable framing first: the 7.5% drop is not a market error. It is the correct read of a company that just delivered its best quarter ever and then reminded investors that 93% of its current market cap is priced on businesses that still lose money. Tesla delivered 480,126 vehicles in Q2 2026, crushing Wall Street consensus of 406,024 by nearly 18%. Up 25% year over year. Up 34% from Q1. Its strongest second quarter ever and its first year-over-year delivery growth after two consecutive years of declines. Energy storage deployments hit 13.5 GWh against an estimate of 13.3. European markets grew 108% year over year. The car business is recovering. The market does not care about the car business. That is the entire story. The Two-Company Problem Apply a traditional auto sector multip

The question Tiger SG is asking, what did you miss in H1, is more uncomfortable than it looks. Because the answer for most investors is not a single stock. It is a structural misread of how the entire market was rotating underneath the headline numbers. The S&P 500 rose 9.5% in H1 2026, slightly behind its historical annual average of 12.8%. That sounds orderly. It was anything but. Beneath the index, the old winners became the laggards. The hyperscalers, Microsoft, Alphabet, Amazon, Meta, and Oracle, solidly underperformed the market. Microsoft was on track for its worst monthly loss since 2008, down 20% in June alone. Oracle fell 30%. Meanwhile, investors piled into memory chip companies whose products help power AI. Samsung, Micron, and SK Hynix are now the 10th-, 13th-, and 14th-mo

The number nobody is leading with: reserve interest is 99% of Circle's revenue. Not most of it. Not a lot of it. Ninety-nine percent. That single fact reframes every other sentence in this story. On June 30, Open Standard announced Open USD, or OUSD, backed by over 140 companies including Stripe, Visa, Mastercard, BlackRock, BNY, Coinbase, Google, Shopify, American Express, Standard Chartered, DBS, and Ripple. CRCL fell 17.55% on the day, extended losses into Wednesday, and is currently sitting around $65 to $66, down 39% from its IPO high and having just been dropped from five major Russell Growth indexes. The question is not whether OUSD will overtake USDC next year. It almost certainly will not. The question is whether a company that generates 99% of revenue from one mechanism, keeping

MSFT fell another 3.46% today, extending a multi-day losing streak that has now pushed the stock down more than 24% in 2026, potentially its steepest June drop in company history. This is not one bad headline. It is three separate fears compounding on top of each other, and untangling which ones are real versus which are noise is exactly the work that matters right now. Fear One: Capex Is Eating Free Cash Flow Microsoft is on track to spend roughly $190 billion on AI infrastructure in fiscal 2026, up from an earlier estimate near $165 billion. Q3 capex alone came in at $31.9 billion, with Q4 guided above $40 billion. The consequence shows up directly in free cash flow. FCF fell to $15.8 billion in the most recent quarter, down from $20.3 billion a year earlier, against reported net income

Same story. Two completely different stocks. On June 25, Apple raised prices across nearly its entire hardware lineup, citing a memory shortage CEO Tim Cook called a "hundred-year flood." The stock fell 6.12% to close at $275.15, its worst single day since April 2025 and its sharpest fall since the "Liberation Day" tariff shock. Roughly $200 billion in market cap evaporated in one session. The same day, Micron reported the most profitable quarter in its history. That is not a coincidence. It is the same supply shock, hitting two companies on opposite ends of the same chain. What Actually Happened The price hikes were sweeping and immediate. The MacBook Air 13-inch jumped from $1,099 to $1,299. The base MacBook Pro climbed from $1,699 to $1,999. The entry-level MacBook Neo rose from $599 to

This was not a beat. This was a number that rewrote the script. Micron reported fiscal Q3 2026 revenue of $41.5 billion against a Street estimate of $35.7 billion, a 16% beat. EPS came in at $25.11 versus $20.49 expected. Revenue nearly quadrupled year over year from $9 billion to $42 billion. Gross margin jumped to 84.9%. Cloud memory revenue alone rose over 300% to $13.77 billion. CEO Sanjay Mehrotra delivered the line that moved the entire sector: there is no line of sight to AI memory supply catching up with demand, with shortages expected to persist beyond 2027. The stock closed at $1,048.51 the day before earnings and gapped to $1,182 to $1,196 in the immediate aftermath, a move of 14 to 17%. Reuters tallied the read-through into a roughly $400 billion single-session rally across AI

Tonight, history prints. SPCX begins trading on the Nasdaq at a fixed price of $135 per share, raising $75 billion at a $1.77 trillion valuation. That makes it the largest IPO ever, more than double Saudi Aramco's 2019 record. SpaceX debuts as roughly the seventh-largest US company, bigger than Tesla on day one. The demand numbers are absurd. Over $250 billion in subscriptions locked up. Retail alone contributed $70 billion. Allocation rates expected at just 20 to 30%. Over 1,000 institutions fought for shares. Polymarket estimates the IPO creates roughly 4,000 new millionaires inside the company. Oppenheimer says $190. Morningstar says $63. The gap between those two numbers is the entire story. Let's break it down properly. What You Are Actually Buying This is no longer a rocket company.

Here is the most uncomfortable chart in markets right now. The US and Iran are exchanging strikes. The Strait of Hormuz is disrupted. Oil is above $90. By every textbook, gold should be screaming higher. Instead, gold futures touched an intraday low of $4,047, pressing against the psychologically critical $4,000 level, down roughly 27% from its January all-time high of $5,589. Gold's worst monthly drop since 2008 happened in March, during the most serious Middle East escalation in decades. The safe haven failed exactly when it was supposed to work. So before anyone buys this dip, you need to understand why it failed. Because the answer determines whether $4,000 is a generational entry or a trapdoor. Why the Safe Haven Failed The mechanism is counterintuitive but logical once you see it. Wh