Syz the moment

World tech equities are underperforming other stocks by ~7 percentage points, one of the worst starts to a year in HISTORY. Tech relative performance is now tracking in the bottom 10% of years since 1973. Historically, the median outcome by year-end has been +4 percentage points of outperformance for tech. The last time tech underperformed this badly was during the 2000 Dot-Com Bubble. Is the tech bubble starting to pop? Source: Goldman Sachs, Global Markets Investor

Source: zerohedge

MS: "While this is not an exhaustive study, based on 14 events over the last few decades (generally major, as well as some recent smaller Iran related incidents) the S&P 500 is generally range bound following the start of a conflict. The event causing the largest drawdown in this sample was September 11th, 2001." Source: TME, MS

The index has spent almost six months trading mostly within a 200-point range (with a few over- and undershoots). Impressive given the many under-the-hood moves and the latest political chaos. Source: TME, LSEG

That’s the surface story. The real story might be the biggest hidden risk in the AI boom. In 48 hours, the KOSPI fell 17%. $275B wiped out. Circuit breakers triggered. Tech giants took the hit: • Samsung −10% • SK Hynix −12% Most analysts blame rising tensions in the Middle East and oil above $80. But something deeper is being exposed. Samsung + SK Hynix control: • ~67% of global DRAM • ~80% of high-bandwidth memory (HBM) HBM is the critical fuel of AI infrastructure. Every AI datacenter depends on it. NVIDIA chips. Google TPUs. Hyperscaler AI clusters. And almost all of it comes from one country: South Korea. Here’s the vulnerability: South Korea imports 97% of its energy. Much of it flows through the Strait of Hormuz. The same strait currently under geopolitical threat. That means the AI supply chain may have a single hidden chokepoint: Not chips. Not talent. Not capital. Energy. Because semiconductor fabs cannot run without massive power. And global memory inventories are thin: • DRAM: ~2–3 weeks • NAND: ~3–4 weeks If energy flows are disrupted for more than a month, the entire AI infrastructure buildout could face delays. Markets are already reacting. While semiconductors crashed, defense stocks surged. Capital isn’t leaving Korea. It’s rotating into a new thesis: Energy security is the real constraint of the AI era. The market may have just discovered the weakest link in the AI supercycle. And it’s only 21 miles wide. (The Strait of Hormuz.) Source: Shanaka Anslem Perera ⚡

At least for now... Source: TME

Source: Charlie Bilello

Source: Gunjan Banerji