July 2026 US stock “crazy split” play-by-play: software crashes on one side, AI hardware mania on the other

If you’ve been scrolling through US stocks recently, you’re probably as baffled as everyone else by this surreal market. This market is no longer the old script where “tech stocks all rise together and blue chips all stay steady.” Instead, it’s an extreme split—like ice and fire. Today, let’s break down the hottest themes in plain talk, no fancy buzzwords—just the grounded trading reality.

First, look at the recent market roller-coaster over the past two days: on July 13, with geopolitical tensions and rate-hike panic, the market collectively sold off— the Dow fell 0.26%, the S&P 500 dropped 0.79%, and the Nasdaq plunged 1.55%; in just two days, the semiconductor sector fell more than 11%. Then on July 14, after the CPI print exploded, all three major indices reversed and turned sharply green. The Dow edged up 0.02%, the S&P 500 jumped 0.38%, and the Nasdaq surged 0.9%, staging the classic “bad news runs out and turns into good news” drama. Most bizarre of all was the level of sector fragmentation: on one side, IBM plunged 25.21% in a single day, marking its biggest drop in nearly 40 years; on the other side, SK Hynix’s ADR went from a slump of 9% to a rebound surge of 27%. Micron, AMD, and Applied Materials—AI hardware names—collectively roared back up by 8%, and capital was moving like crazy.

Why did AI compute infrastructure hardware suddenly flip from a “sell-off wave” to a “mania wave”? The core is two things: first, June CPI came in with a month-over-month -0.4% and a year-over-year 3.5%, cooling far more than expected. The probability of a July rate hike was cut from 50% to 15%, and US Treasury yields promptly fell back—loosening the leash on high-valuation tech stocks. Second, Meta clarified there’s no compute oversupply; SK Hynix’s ADR + options listing also lit up the storage trade. On top of that, HBM high-bandwidth memory is in short supply relative to demand—so AI capital expenditures really did shift from “software services” to hard-core infrastructure like “servers, storage, and wafer foundry production.” SanDisk, Western Digital, ARM, and Corning optical communications all charged higher. The Philadelphia Semiconductor Index nearly pushed toward 8000 intraday. Even with institutional warnings that valuations are too high and short positions hitting a three-year high, funds still doubled down on the AI hardware main theme.

Meanwhile, on the other side, traditional software stocks are basically a “real-time AI bear market warning site.” IBM’s Q2 revenue and EPS both missed expectations, and the CEO laid it out plainly: customers are shifting IT budgets from software and traditional IT toward AI infrastructure, and orders are getting delayed. This one shock dragged down software names like ServiceNow, Adobe, and Accenture— even Microsoft and Oracle were pulled under. Goldman Sachs even directly called for a “cyclical adjustment” coming for the software sector. Capital then went on a frantic retreat from “old tech,” clustering instead into the compute/AI hardware track.

And the “mythical battle” during earnings season is also the biggest current hotspot: JPMorgan, Bank of America, and Goldman Sachs all kicked off reporting. Goldman’s Q2 revenue hit an all-time high, and net profit surged 78%; Bank of America’s stock-trading revenue rose 70% year over year; JPMorgan’s net profit grew 41%. While still speaking, Dimon also reminded investors about geopolitical and inflation risks. On the other side, IBM’s blow-up directly confirms the harsh reality that the “AI dividend” goes to the infrastructure end. Institutional disagreement is also at maximum intensity: hedge funds were net buyers of US stocks for the first time in four weeks, and bullish options sentiment hit the highest level since 2020. Yet Goldman and Bank of America are still issuing warnings, saying the market is getting overheated—so the long-vs-short game is especially fierce.

Apple’s independent momentum and Nvidia stabilizing its AI leadership by standing on $41.1 billion of data center revenue also show that US stocks have long moved on from the era of “buy AI blindly and you make money.” The main line for the second half of 2026 is especially clear: hardware companies that can actually fulfill compute orders and have real AI revenue are favored, while companies that only tell stories with traditional software are being abandoned by capital. Inflation and the Fed are still the market’s “nerve switch”—rate-hike expectations for September remain a disturbance on the tape. Geopolitical tensions and oil price volatility will also keep launching occasional surprise attacks.

For ordinary people like us, you don’t need to chase rallies or get yanked around by short-term up-and-down moves. In today’s split-and-differentiate US stock market, stock selection matters more than timing by a million times. If you want to set up an allocation, prioritize legitimate channels like QDII or compliant cross-border wealth management—avoid illegal speculation. Keep US stock exposure to 10%-20% of household assets, learn more about market logic, and stay disciplined and rational on position sizing—that’s the key to making it a long-term journey.

The stock market always carries both risk and opportunity. All market sharing is just idle conversation for reference and does not constitute any investment advice. Rationally observe the market and invest steadily—this is the core to understanding this round of US stock split热点～#PreIPOs第二期OpenAI认购 $BTC ‌