September 14: Software Eats the Hardware Trade

Market regime

The index told you nothing again — SPY is 1.1% lower over a month while the average stock is down 3.8% and only 28.9% of names are positive over that window. Today the tape was a near-perfect split (1,500 advancers vs 1,527 decliners, average stock -0.25%), but the internals keep deteriorating underneath the calm: just 38.6% of stocks hold their 50-day, 3,073 names sit an average of 22.6% below their 52-week highs, and breakdowns outnumber breakouts 37 to 9. VIX at 14.6 is not complacency about the market — it is complacency about the index, which no longer represents the market.

Mega forces

1. Software is decoupling from the AI hardware complex — violently. Information Technology fell 0.71% on average today, but that number is a lie of averages: 74 IT names gained more than 5% while 118 lost more than 5%. On one side, cybersecurity and data software ripped — Zscaler +16.5%, Rubrik +15.6%, CrowdStrike +13.9% (a new 52-week high), Palo Alto +13.1%, Okta +12.0%, Samsara +11.8%, Samsara and CrowdStrike closing at highs. On the other, the picks-and-shovels suppliers were taken apart: Silicon Motion -16.8%, Corning -13.7%, Teradyne -13.3%, Nokia -13.3%, MACOM -12.8%, Coherent -12.7%. This is the market repricing the AI capex chain after a quarter of hyperscaler caution on frontier model pacing (Microsoft's self-imposed limits, Broadcom's Anthropic commentary) — while rewarding the software layer that monetizes AI without owning the fab.

2. Commodity scarcity is the only inflation trade that still works. WTI +21.7% over a month and +107% over a year, broad commodities +10.1%, agriculture +3.9%. Energy is the lone sector with a positive 1-month return (+5.1%, +50.8% over a year), 77% of energy names above their 50-day and only 9% of the sector's 1-year return given back. Oil & gas E&P (+9.8%), integrated oil (+10.8%), royalty/land (+6.7%) and marine shipping (+7.1%) — the latter amplified by a 108% one-month move in tanker equities — are all tariff-proof, rate-insensitive cash machines.

3. Long-duration, rate-sensitive and capex-heavy industrials are being liquidated. Industrials -8.4% over a month with only 21.4% of names above their 50-day, utilities -5.7% and real estate -4.8% with just ~13% above their 50-day. The 10-year at 4.63% with a term premium that strategists keep flagging is doing the work: anything whose cash flows are back-end loaded or bond-proxy-like is being funded out of.

What's working

Energy is the single sector where the trend, breadth and fundamental backdrop align. Beneath it: biopharma and life-science tools (health care is the best 3-month sector at +7.8% with 65.5% above the 150-day — Moderna +130% in a month, 10x Genomics, Natera, IQVIA, Avantor all 45%+ over three months), marine transportation, and select digital-asset infrastructure (Crypto & Digital Asset Platforms +23% 1-month, COIN +9.2% today). Gold is flat-to-lower while oil rips — this is an industrial-demand and supply-scarcity cycle, not a monetary-debasement one.

What to avoid

The entire speculative duration complex: lidar (-24%), SPACs (-18%), quantum (-17%), solar (-16%), space/satellites (-16%), industrial energy storage (-15%), semiconductor capital equipment (-15%) and aerospace & defense components (-15%). The same force — a higher-for-longer cost of capital meeting a capex digestion phase — is hitting each of them. Also avoid the defensives investors instinctively reach for: utilities and REITs are down 3-6% over a month and are the weakest breadth groups in the market, so they are not a hiding place.

Strategy

Stop treating the S&P's 3.9% quarterly gain as information; it is being manufactured by a shrinking set of mega caps. Own the cash-flow-now cohort: energy producers and refiners with buybacks and free cash flow at 4.6% yields, biopharma with de-risked pipelines, and the software names that broke to new highs on volume. Fund those positions from the AI capex derivative trade — equipment, optics, memory, power — which is where the damage is compounding. When 37 stocks break down for every 9 that break out, the correct posture is fewer, larger positions in the narrow list of things working, not broader index exposure.