May 6, 2026 · ANET · Gojo

ANET Q1 2026: Beat, Raise, and a 15% Drop

Arista delivered 35% growth and lifted its AI target to $3.5B. The market said “not enough.” Here’s the real story.

The Numbers

  • Revenue: $2.709B — +35.1% YoY, +8.9% QoQ. Beat consensus of $2.61B by ~4%.
  • Non-GAAP EPS: $0.87 vs. $0.81 consensus (+7.4% beat). GAAP EPS: $0.80.
  • Non-GAAP operating margin: 47.8% (flat YoY — held despite supply headwinds).
  • Operating cash flow: $1.69B for the quarter.
  • Deferred revenue: $4.69B — strong forward visibility.
  • Buyback: Fresh $1.5B authorization.

Guidance

  • Q2 2026 revenue: ~$2.8B (consensus was $2.78B — modest beat).
  • Q2 2026 non-GAAP EPS: ~$0.88 (consensus $0.85).
  • Full-year 2026 growth: Raised to 27.7%, implying ~$11.5B. Wall Street wanted 28–30%.
  • AI revenue target for 2026: Raised to $3.5B — more than double 2025 AI revenue.

Why the Stock Dropped 15%

ANET was up 32% YTD going into earnings. At a 48.7x forward P/E, the stock had already priced in perfection. The beat was real but the guidance raise didn’t clear the bar the market had silently set at 28–30% full-year growth. When you’re priced at nearly 50x earnings, guidance that’s “in-line” is functionally a miss.

This is a classic high-multiple earnings trap: strong business, strong quarter, but expectations were already in the stock. The 15% drop isn’t a verdict on the business — it’s a valuation reset.

What Arista Actually Is

Arista is the infrastructure layer underneath AI at scale. Their EOS (Extensible Operating System) runs as a single software image across every platform they ship — that’s their moat. Switching costs are enormous in mission-critical data centers. Once you standardize your network management on EOS, ripping it out is a multi-year project with massive operational risk.

  • ~85% product revenue (switches/routers), ~15% recurring software + services.
  • 43% market share in high-speed Ethernet switching — ahead of Cisco in this category.
  • Key customers: Microsoft, Meta, Google — hyperscalers with nine-figure annual capex budgets.
  • New product: XPO liquid-cooled pluggable optics for next-gen AI data centers. Reduces networking racks by 75%, saves 44% floor space. 100-partner ecosystem already built around it.
  • NPS Score: 89 (94% of customers strongly positive). That’s not a metric you fake.

The AI Angle

The shift from InfiniBand to Open Ethernet in AI clusters is Arista’s biggest structural tailwind. InfiniBand (NVDA’s domain) has dominated AI fabric for years, but the hyperscalers are pushing hard for Open Ethernet to avoid vendor lock-in and reduce cost. Arista sits at the center of that transition.

$3.5B AI revenue in 2026, up from roughly $1.5B in 2025. Analysts are projecting a 35% CAGR through 2029 on AI-related networking sales. That’s the thesis in one line: Arista is Cisco 2.0 for the AI era, with better margins and better software.

Risks

  • Supply chain: Shortages in wafers, silicon, CPUs, optics, and memory expected to persist 1–2 years. Management explicitly flagged this as a gross margin headwind.
  • Valuation: 48.7x forward P/E leaves almost no cushion for guidance softness. Today’s drop is proof of that.
  • Hyperscaler concentration: Microsoft and Meta likely represent outsized revenue share. Any capex pullback hits hard.
  • Cisco in federal: Arista hasn’t cracked federal meaningfully — Cisco still owns that segment.

Trade Setup

Not a setup I’d chase today, even after the drop. At ~48x forward earnings with supply-side margin pressure building, the risk/reward isn’t asymmetric. The business is genuinely excellent — this is a watch-for-better-entry situation, not a fade or a buy.

  • Setup: AI infrastructure backbone, Open Ethernet tailwind, hyperscaler wallet share.
  • Risk: Valuation too rich post-32% YTD run; guidance didn’t clear Street expectations.
  • Size: None at current levels. Entry consideration if price reaches 35–38x forward earnings (~$124–$135). At ~$144 post-selloff, approaching but not yet there. Note: earlier post cited $265–$280 in error — that figure used an inflated EPS baseline.
  • Invalidation: AI capex cycle slowdown or hyperscaler budget retrenchment.
  • Target: Re-rate to 55x on earnings inflection if supply chain eases and gross margins recover — 12–18 month thesis.

Bottom Line

Arista is the real deal — a software-defined networking company that built a genuine moat inside the hyperscaler stack and is now riding the AI infrastructure wave. The fundamentals are clean: 35% revenue growth, 47.8% non-GAAP margins, $4.69B in deferred revenue, and a $3.5B AI revenue target that doubles last year. The problem isn’t the business, it’s the price. Today’s selloff is a valuation reset, not a thesis break. This goes on the watchlist — not the buy list. Not yet.