Is Arista Networks Stock a Buy on the Dip?

Arista Networks posted Q1 revenue of $2.71 billion, up 25% year over year and above the $2.6 billion forecast, while adjusted EPS rose 32% to $0.87. However, management warned that shortages of wafers, memory, CPUs, and optical components could pressure gross margins; gross margin already fell 180 bps to 61.9% and adjusted gross margin fell 170 bps to 62.4%. Full-year revenue guidance was raised to $11.5 billion, with AI fabric sales now expected at $3.5 billion, but the stock remains expensive at 39.5x 2026 earnings.

Analysis

The key second-order issue is that AI networking is moving from a scale-up story to a procurement-arbitrage story. When a platform vendor like ANET cannot fully pass through input inflation to hyperscalers, margin compression usually lags revenue growth by a quarter or two, which means the market may still be underestimating the next 2-3 reporting periods. That makes the current setup less about one quarter’s miss and more about whether ANET’s earnings power gets mechanically capped just as investors are paying peak multiple for AI infrastructure exposure.

The relative winners are upstream component suppliers with pricing power and constrained capacity, not the assemblers. If shortages persist across opticals, memory, and CPUs, the bottleneck shifts bargaining leverage toward vendors closer to the wafer and photonics layer; that is structurally supportive for NVDA and select semiconductor adjacencies, while being a tax on system integrators. MSFT and META remain the only customers large enough to force roadmap discipline, but they also have the balance sheet to dual-source and pressure pricing, so ANET is stuck in the middle: too important to lose, too commoditized to reprice.

The contrarian takeaway is that the selloff may not be fully about fundamentals; it is also about positioning into an expensive AI winner that is no longer a clean margin expansion story. If supply normalizes in 6-9 months, ANET could rerate quickly because revenue visibility stays strong and the market has already discounted near-term pain. But if shortages persist into 2027 as management suggests, earnings estimates are still too high and the equity deserves a lower multiple than the current forward P/E implies.

Near term, the trade is less about chasing a bottom and more about owning the cleaner AI monetizers versus the squeezed picks-and-shovels layer. A durable catalyst to reverse the weakness would be evidence that gross margin stabilized before revenue growth decelerates, or that large customer additions broaden bargaining power enough to offset component inflation. Absent that, ANET likely remains a range-bound name where rallies should be sold into rather than bought aggressively.

Key Decisions for Investors

• Short ANET vs long NVDA in a 3-6 month pair trade: thesis is that AI infrastructure spend continues, but value capture migrates upstream to the chip/platform layer while ANET absorbs supply-chain inflation; use a 1.0x notional hedge and cut if ANET gross margin stabilizes sequentially.
• Avoid adding to ANET on the initial dip; wait for either a second leg down after the next quarter or evidence of inventory normalization. Risk/reward improves only if forward P/E compresses materially below current levels or if margins stop slipping.
• Maintain or add to MSFT/META over ANET for AI exposure: these balance-sheet leaders can negotiate supplier terms and internalize more of the AI value chain, making them better risk-adjusted owners of AI demand over the next 6-12 months.
• For tactical traders, consider buying 1-2 quarter downside protection on ANET via puts or put spreads into the next earnings window; the catalyst is not revenue collapse, but estimate cuts from margin pressure, which can hit the stock before fundamentals fully deteriorate.