TL;DR - The AI trade is rotating. Memory, storage, space, quantum, mining and parts of nuclear have already peaked and retraced. Semiconductors, networking, software and data-center infrastructure remain stronger. Overview Weekly performance through 7 August 2026. The percentages show the latest weekly change, not the total retracement from the peak. --------- 1. Already Peaked And Retraced: Memory and storage: SNDK (-0.2%), MU (+6.0%), WDC (-20.3%), STX (-5.1%) Photonics: AAOI (+43.8%), COHR (+44.2%) Mining and AI-HPC: IREN (+12.0%), WGMI (-5.9%) Space: RKLB (+27.6%), ASTS (+22.0%) Quantum: IONQ (+21.9%), RGTI (+20.0%), QBTS (+14.8%) Nuclear: OKLO (+24.7%), CEG (+2.7%), CCJ (+12.7%) Semiconductors: AMD (+1.5%) --------- 2. Retraced But Recovering: PLTR (+39.8%), COHR (+44.2%), RKLB (+27.6%), ASTS (+22.0%), AAOI (+43.8%), IONQ (+21.9%), IREN (+12.0%) --------- 3. Yet To Show A Confirmed Major Peak: AI Compute and Semiconductors: NVDA (+11.6%), TSM (+3.9%), AVGO (+9.9%), ARM (+17.9%), ALAB (+7.4%) Networking: ANET (+4.6%), MRVL (+16.6%), CRDO (+20.7%), CSCO (+4.7%) Software and Cybersecurity: NOW (+12.3%), ORCL (+13.2%), CRWD (+12.3%), PANW (+9.7%), FTNT (-1.4%), MSFT (+7.6%), GOOGL (-0.5%), AMZN (+1.1%), CRM (+4.7%), ADBE (+5.9%), SNOW (+12.7%), DDOG (+2.0%), CDNS (+0.2%), SNPS (+2.5%) Data Centers and Power: VRT, PWR, ETN, HUBB, EQIX, DLR, AMT, CCI, SBAC, APLD (+1.2%), CORZ Physical AI and Defense: TSLA, ISRG, SYM, KTOS, AVAV, RTX, NOC, LMT, GD, LHX Copper and Critical Materials: FCX, SCCO, TECK, RIO, BHP, MP, ALB, SQM, LAC --------- 4. ETF Highlights Structurally Strong: AIQ, SMH (+7.8%), IGV, CIBR, DTCR, BOTZ, ARKX, NLR, QTUM Clearly Retraced: WGMI (-5.9%) Too new for a full-cycle assessment: DRAM --------- 5. Summary The most heavily retraced areas are memory, storage, space, quantum and mining. The strongest ongoing momentum is in semiconductors, networking, software and AI infrastructure. The next phase may favor companies solving the bottlenecks beneath the AI narrative, including power, cooling, connectivity, materials and data-center capacity. NFA / powered by @elfa_ai