Company map from the 24h scan:
$MRVL: cleanest catalyst setup. S&P 500 inclusion on Jun 22 creates forced passive demand, while AI infra demand supports the core story: 800G/1.6T optics, AI switching, custom XPUs, and raised FY27/FY28 outlook.
$LITE/$AAOI: optical bottleneck trade. AI clusters need more lasers, transceivers, and interconnects as bandwidth demand scales.
$LITE is the cleaner optics expression;
$AAOI is the higher-beta version.
$NVDA/$SMH: broad “paid by capex” exposure. The video takeaway was to own the semiconductor/input layer where AI spend becomes revenue, not necessarily every company writing the capex checks.
$NVT/$CECO /
#HPS.A.TO/$ORA /$VRT : power, grid, thermal, and always-on energy. Data centers are becoming a transformer, electrical distribution, cooling, ventilation, and power-generation problem.
$VRT is obvious; the edge may be one layer below.
$FCEL : speculative carbon-capture angle. If onsite gas generation becomes part of AI data-center power, carbon capture can attach to that stack.
$OUST: Physical AI sensors/perception. If the next AI cycle moves into robots, autonomy, and industrial systems, perception hardware becomes a real infrastructure layer.
$SPCX/$RKLB: space-flow trade.
$SPCX is driven by index/flow mechanics post-IPO;
$RKLB is the more accessible public proxy if attention keeps moving into space infrastructure.