Most of the coverage I’ve read frames 28 October as a Fed day with some earnings attached. I don't think the FED has much impact on equities, prediction markets and OIS pricing have a hold at 3.75 to 4% north of 90%, and PIMCO’s base case is a hold with a tightening bias and the next hike in December. Unless the statement surprises, that’s mostly priced. The capex commentary from Microsoft and Google, followed by Amazon on Thursday is the watch. In my view, this matters more than usual because of the semiconductors that are currently in a drawdown. SOX fell about 11.4% in the week to 16 October, and Reuters put the cumulative loss at around $2 trillion in market value, driven by rate hike bets and concerns about the cost of the AI build. Then Thursday’s session saw all 30 index members finish red, even after all that, the index is still up roughly 57% for the year, so this is a correction inside a huge run rather than a collapse. Hyperscaler capex is essentially the revenue line for the AI chip supply chain, Microsoft has already guided fiscal Q1 capex above $50 billion and roughly $175 billion for calendar 2026, and management has said demand still exceeds available supply. Combined 2026 guidance across the big four sits around $730 billion, and Goldman has 2027 approaching $1.14 trillion. If Microsoft, Google and Amazon hold or raise those numbers this week, the chip selloff has priced in a slowdown that isn’t showing up in the order book. Bull case for chips is that the spenders and the suppliers sit on opposite ends of the same number. Meta beat last week and still fell nearly 7% because the market punished the spending, but every dollar Meta spends lands on someone else’s income statement. Retail seems to see it this way too, with tech funds pulling in $10.7 billion during the worst week of the drawdown. Bear case is still very very active, the Fed is hiking, not cutting, and a December move would lift discount rates on the most richly valued names. IDC is forecasting a 13% drop in smartphone volumes this year from the memory supply shock, which hits a part of the chip complex that has nothing to do with data centres. The funding picture is also getting less comfortable, with Goldman estimating more than a third of 2027 hyperscaler capex will be debt funded. At some point the spenders balance sheets become the constraint rather than demand. If at least two of the three raise or reaffirm higher spend this week, I expect the SOX to outperform the Nasdaq 100 from Wednesday’s close through Friday 6 November. If they guide flat or hint at a digestion year in 2027, I’m wrong and the chip correction has further to run. For those who own both sides which I assume is many, which do you think breaks first. The hyperscalers appetite for capex, or the market’s patience with the hyperscalers?
CryptoAlerta — análise de criptomoedas e mercado em tempo real