Eclipse raises $1.3B for AI infrastructure and defense startups By Investing.com

Eclipse raised $1.3 billion — $720M for early-stage and $591M for later-stage investments — its largest fund to date (vs $1.23B in 2023) to back startups building physical-world technologies. The firm targets AI infrastructure, manufacturing and defense and lists holdings including Cerebras, Wayve and Redwood Materials; Cerebras is expected to go public imminently. The raise highlights continued VC interest in hardware, defense and industrial tech (notable valuations cited: Valar Atomics $2B, Physical Intelligence ~$11B in talks, Anduril >$60B), likely boosting capital availability for these sectors while having limited immediate market-wide impact.

Analysis

The migration of patient, later-stage capital into hardware-heavy, physical-industries startups creates a multi-year re-rating opportunity for the supply chain that actually manufactures and tests AI compute — not just the chip designers. Expect outsized revenue growth for semiconductor equipment and advanced-packaging vendors as a wave of boutique AI accelerators and vertically integrated battery/robotics plays attempt to scale: tooling orders are lumpy but front-loaded, so a visible uptick in bookings in the next 6–12 months will presage durable FY+1 revenue beats.

Second-order winners extend beyond semicap: OSAT/test houses, specialty materials (high-purity copper/graphite/precursors), and domestic labor-intensive assembly lines will capture the bulk of initial unit economics as startups prioritize sovereignty over low-cost offshoring. Conversely, pure-play cloud software services and legacy fabless players that rely on a small set of accelerator partners face margin pressure if customers move to custom silicon or in-house stacks. Geopolitical push for onshoring amplifies this, skewing procurement toward US-listed contractors and equipment makers over time.

Key risks are execution and timing: hardware startups still face multi-year scale and yield curves, so public beneficiaries trade on bookings visibility rather than installed base today. Macro tightening or an IPO/late-stage funding cooldown would pause capex cascades within 3–9 months and compress multiples; an alternative reversal is rapid commoditization of AI accelerators that collapses pricing power and caps semicap cycle length to 12–18 months instead of several years.

Key Decisions for Investors

• Overweight LRCX (Lam Research) — 6–12 month horizon. Buy shares or construct a 6–9 month call spread to limit premium. R/R: target 20–35% upside if bookings accelerate; downside ~15% on cycle disappointment.
• Long AMAT (Applied Materials) — 9–18 month horizon for packaging and materials exposure. Accumulate shares on pullbacks or buy 12-month calls; expect 2:1 upside vs premium paid if advanced-packaging ramps with new AI accelerators.
• Long LMT (Lockheed Martin) — 12–24 month horizon to play defense procurement reorientation toward sovereign tech. Size as a conservative ballast (10–15% of hardware allocation); R/R ~3:1 given steady backlog and higher-margin integrations.
• Pair trade: long LRCX / short QQQ (equal notional) — 3–6 month horizon to express a reallocation from software multiple expansion to hardware capex. This hedges market beta while capturing a relative rotation; watch for NVDA-driven broader market rallies that could hurt the short leg.
• Tactical options: buy a small notional NVDA 3–6 month OTM call spread as a convex play on sustained AI infra spend. Limited capital at risk (premium only) with asymmetric upside if compute demand keeps accelerating; capped loss if the hardware cycle stalls.