The AI Supercycle Just Entered Phase 2. Most Investors Aren't Ready.

Keysight reported FYQ1 revenue of $1.6B (+23% YoY) and record orders of $1.65B, with wireline surpassing wireless for the first time and management guiding full-year revenue growth "just above 20%"; the stock jumped ~23% after earnings and is up ~80% over 52 weeks. Calix launched Calix One in Feb 2026 (built on Google Cloud), had ~1/3 of its 1,200 customers on the platform by mid-March, and posted Q3 2025 revenue of $265M with a record gross margin of 57.7% and 20 consecutive quarters of positive free cash flow; BEAD federal funding is driving multi-year fiber deals. Risks: Keysight's hardware-heavy mix yields mid-60% gross margins below pure software peers and is sensitive to hyperscaler capex, while Calix's growth depends on the timing and pace of BEAD rollout.

Analysis

The AI cycle is shifting from concentrated GPU scarcity to distributed, physical bottlenecks — testing, interconnects, and field-deployed management. That dynamic favors vendors who control validation workflows and recurring software layers because order timing becomes a function of lab capacity and integration complexity rather than silicon tapeouts; lead times lengthen and convert into durable pricing power for test and service providers.

Second-order beneficiaries include OSATs, optical-module and connector OEMs, lab automation software, and managed-service platforms that turn one-off installs into subscription revenue. Incumbent consumer broadband incumbents face localized share risk where government-funded fiber projects accelerate replacement economics; conversely, companies that productize field operations (agentic agents, automated provisioning) will compound ARPU with low marginal cost.

Key near-term catalysts are tranche releases of public funding and hyperscaler capex cadence; both create 6–18 month visibility windows for revenue inflection but deliver lumpy bookings. Tail risks: a hyperscaler pause or a technological leap (e.g., integrated optics reducing external test needs) could compress the thesis quickly, while regulatory/program delays shift outcomes out multiple years.

Consensus is overweighting obvious chip winners and underweighting dispersion risk inside the supply chain. A tactical rotation into infrastructure providers with recurring software exposure offers asymmetric upside if you buy around meaningful pullbacks and use time-limited option structures to cap downside while capturing multi-quarter delivery cycles.

Key Decisions for Investors

• Long KEYS (6–12 months): establish position via a 9–12 month call spread to limit capital outlay and define downside (sell a nearer-term call to finance). Target 30–60% upside if testing backlog monetizes; hard stop at 20–25% drawdown. Monitor hyperscaler order cadence and lab-capacity expansion announcements as entry/scale triggers.
• Long CALX (12–18 months): buy LEAP calls or accumulate stock on BEAD-tranche confirmations (scale 25% on initial announcement, add at execution milestones). Risk/reward centered on 40–80% upside if platform adoption converts customers to recurring ARR; downside ~30% if public funding stalls — hedge with OTM puts or size accordingly.
• Pair trade (3–9 months): short NVDA exposure (smaller notional, e.g., 0.25–0.5x dollar of NVDA vs long KEYS) to express mean-reversion from chip multiple concentration into infrastructure. Protect tail risk by purchasing OTM calls on NVDA; target capture of rotation alpha if market shifts budgeting from GPUs to systems-level spend.
• Tactical pair (6–18 months): long CALX / short CMCSA on selective size — trade reflects rural fiber substitution risk to cable incumbents. Size conservatively (1–3% net portfolio) and tie add-ins to BEAD deployment milestones; expect multi-quarter realization and use quarterly newsflow to re-evaluate exposure.