Rosenblatt reiterates Arista Networks stock rating on strong AI demand

Arista Networks reported Q3 2026 revenue of $3.66B, up 58.5% year over year and about 11% above its $3.3B guidance; non-GAAP EPS was $1.20 versus guidance of $1.06–$1.08. It raised FY2026 revenue guidance for the fourth time to $13.3B and forecast FY2027 revenue of approximately $18.5B, or 38%–40% growth; Rosenblatt reiterated Buy with a $280 price target and raised its FY2027 EPS estimate to $5.90 from $5.09. Shares nevertheless fell 9% over the past week, with no reason for the decline specified in the article.

Analysis

The key issue is not whether AI networking demand is strong, but whether the next leg of growth can keep clearing a very high execution bar. The selloff despite a major beat suggests investors may be discounting forward growth durability, valuation, or customer concentration—not disputing the quarter itself. Flat deferred revenue alongside sharply higher sales is a useful diligence flag, not proof of weakening demand: check remaining performance obligations, backlog conversion, and customer mix before treating the guide as fully secured.

Over the next 1–3 months, the raised outlook and supply reservations support revenue visibility, but also make execution asymmetric: if customer deployments slip, reserved supply can turn from protection into working-capital and inventory exposure. Over 6–18 months, continued AI-cluster buildouts favor Arista, while Cisco, HPE/Juniper, and Nvidia’s networking offerings can contest spend; hyperscaler capex pauses or architecture changes are the principal demand risks. Strong reported margins provide room to invest, but do not remove multiple risk if growth decelerates faster than expected.

Contrarian read: the market may be treating one exceptional quarter as less valuable than the company’s forward conversion evidence. Do not anchor on the analyst target; it is a single broker view, not validation of fair value. Bull case is intact if forward orders convert without margin erosion; bear case is that supply commitments and growth expectations are running ahead of durable customer demand.

Key Decisions for Investors

• Avoid chasing the earnings beat. Consider a staged ANET long only after price action stabilizes and the next operating update confirms forward demand; the thesis is falsified by a material guidance cut, slowing order/backlog conversion, or sustained gross-margin erosion.
• For the next 1–3 months, track deferred revenue together with RPO/backlog, customer concentration, and purchase-commitment conversion. Treat flat deferred revenue alone as an alert, not a short signal; deterioration across these measures would warrant reducing exposure.
• Use Cisco and HPE/Juniper as competitive checks rather than automatic shorts: monitor evidence of share loss or pricing pressure in data-center networking before expressing a relative-value trade.
• Over 6–18 months, stress-test ANET exposure against a hyperscaler capex slowdown and delayed AI-cluster deployments. A pause in customer spending could hit expected growth while leaving supply commitments in place; absent such evidence, no forced bearish trade is warranted.