AI Capex Panic Meets Bull Case: Street Splits on Meta, Applied Materials as Broadcom Target Implies 60% Upside

The AI trade is no longer a single trade. Independent research published over the past 24 hours drew a sharp line between the compute and semiconductor capital-equipment complex — where earnings visibility is improving — and the derivative beneficiaries, where multiples have run ahead of realized cash flow. On Meta, the argument advanced is that the market's hyperscaler capex panic is a duration mismatch rather than a thesis break: depreciation and cash burn front-load, monetization follows. Applied Materials was flagged as operating at full throughput, consistent with a broadening wafer-fab equipment cycle that extends beyond leading-edge logic into advanced packaging and memory retooling. Broadcom, down over three months, drew the session's most aggressive call, with a Street practitioner arguing for 60%-plus appreciation on custom accelerator and networking share gains. The counterweight was equally explicit. Vicor's power-conversion opportunity in AI racks was described as large but fully discounted at current levels, while Nebius' headline $40bn backlog was characterized as nebulous — a reminder that contracted revenue in neocloud is only as good as the counterparty and the financing behind the GPUs. Mobileye, meanwhile, was framed as an autonomy option the market is not pricing, a rare underowned name in an otherwise crowded theme. For allocators, the actionable read is rotation within the theme rather than exposure reduction: fade the highest-multiple second-derivative exposures, add where capital intensity converts into an installed base.