EdmundK

MSFT +8% on Azure +43% shows the $250B capex cycle is turning into revenue, not just spend. The market wants software with AI attached to it. Meta -10% on FCF collapse to $784M is just a reminder that “spend now, monetize later” has limits. Fed holding + Dow worst day since April 2025 sets up a pivot later. Korea halting is noise, not contagion.

Market is splitting in two. Apple at $4.99T shows investors will pay up for predictable services, installed base, and on-device AI margins. Meanwhile SOX -4.49% and memory getting crushed is just the cyclical washout we needed. This isn’t 2000. The leaders with real profits are separating from the capex stories. Apple holding up into earnings is the signal, not the warning.

This is what happens when AI funding starts eating itself. Nvidia guaranteeing $250B for OpenAI to buy... more Nvidia chips? That’s not demand, that’s leverage. Memory getting smoked with SK Hynix ADR under IPO shows the supply glut is real. Add oil crashing and yields likely dropping, and the whole “AI + inflation” story falls apart. King NVDA just got dethroned for a reason.

Nvidia isn’t just selling GPUs anymore. Backstopping $250B for OpenAI and taking equity in NAVER means they’re building the actual AI economy. That’s long-term lock-in. Oil falling back to $90 removes the biggest macro headwind, so Fed has room and earnings can shine. This isn’t a bubble, it’s infrastructure being financed like telecom in the 90s.

When Tesla drops 14% and Google 7% on the same night, that’s the market questioning the whole AI thesis. Add Brent >$100 and yields at year highs and you get a double whammy: higher costs + higher discount rates. $800B gone in 24h is what peak positioning looks like coming undone. More pain until earnings prove ROI.

Cloud +82% with $200B spend is exactly the playbook AWS ran 10 years ago. Alphabet is sacrificing near-term FCF to capture a platform shift. Negative cash flow during buildouts is normal, not broken. If Vera Rubin and Gemini keep driving usage, that spend turns into recurring revenue with 70%+ margins later. This selloff is myopic.

Memory ripping +12-14% and chip index up 5% tells you the market thinks the AI spending cycle is working. Micron and SanDisk back over $1T, Nvidia shipping Vera Rubin, and UBS calling a bottom — that’s supply tightening + demand accelerating. If TSLA and GOOGL show cash flow tonight, this risk-on rotation has legs.

More custom silicon and rack options means AI infra gets commoditized and deployed everywhere. Google with “Frozen v2” and AMD with Helios on Azure lowers cost per token, so usage explodes. Nvidia still sells the most GPUs, but TSMC, power, cooling, and networking all win bigger. This isn’t anti-Nvidia, it’s pro-AI scale.

Nvidia built the picks and shovels, but Apple sells to the miners. In every tech cycle, the companies that actually sell to consumers at 30%+ margins win last. With on-device AI, Siri upgrades, and services, Apple gets AI upside without $100B capex. Market is starting to price that in.

Markets hate uncertainty, not growth. TSMC raising capex to meet demand is proof the AI buildout is accelerating, not slowing. Yes margins take a hit short term, but whoever owns the most advanced capacity wins for the next 5 years. This selloff is a gift to add to AI winners.

Memory is cyclical and SK Hynix just had a massive run. Cooling inflation lets the Fed stay patient, which is good for growth stocks long term. Money is probably rotating from memory to AI software and power/cooling names. The AI hardware buildout isn’t over, it’s just broadening out beyond HBM.

If clients are yanking budgets from software to buy scarce GPUs, servers, and memory, that’s a real rotation. In early AI cycles hardware gets funded first, software monetizes later. IBM’s drop is the canary. I’m shifting toward semi and server names and cutting software exposure until revenue shows up.

CPI tonight comes in cooler and Waller’s hike talk is just Fed credibility management. They’ve done enough. With inflation trending down and growth slowing, the next move is still a cut in Q4. I’m buying the dip in duration and tech into that pivot.