Arista Networks Is Up 50%-Why I'm Still Buoyant on ANET Despite Rising Competition

Arista's pullback looks unusual for a company posting $3 billion quarters

This looks like a genuine buy window: Arista just posted its first quarter above $3 billion in revenue while the stock is still down 13.1% year to date.

Strong demand is clear, but the debate is not

Bulls see a business that is still scaling rapidly. Arista grew revenue 37.7% year over year in the latest quarter, a sign that demand is still strong even at a large base.

Bears also have a case. Competition is getting more visible in the same AI buildouts Arista is winning, and management has warned that at least 1 and potentially 2 customers could exceed 10% of annual revenue. At this scale, that matters: one large customer slowing spend, reshaping a design, or stretching delivery could make a quarter feel less smooth than the underlying demand.

Why the setup is interesting now

The timing is what stands out. Management has now raised 2026 revenue guidance for the 3rd time even as the stock has softened. That is an unusual mix: stronger operating proof, a higher bar, and a weaker chart.

If Arista keeps clearing that raised outlook, today's pullback may look like one of the cleaner entries before the market stops questioning the size of the business. If concentration or timing starts to spoil the streak, though, this stops being a simple buy-the-dip story.

Arista's moat comes from the full networking stack, not just the hardware

Once you know demand is there, the next question is what keeps customers coming back.

EOS and CloudVision deepen switching economics

Arista's advantage is not just the hardware in the rack. It is the operating system and automation layer around it-EOS and CloudVision-plus networking gear built for data center, campus, and routing environments. Once a large customer builds operations around that stack, replacement becomes more than a price decision.

Breadth is expanding alongside AI demand

Arista is also widening the places its networking can be used. Management introduced 1.6 Tbps AI fabric platforms, including liquid-cooled options, while positioning the company across AI, data center, campus, and routing environments. The earnings read-through said growth should spread across AI scale-out fabrics, scale-across routing, front-end data center switching, campus, enterprise, and routing, not depend on a single product category.

That breadth matters. The more lanes a customer puts Arista into, the more value it gets from one platform, and the harder it becomes to fragment spending across vendors.

Profitability suggests this is more than commodity hardware

The economics support that view. In Q2, Arista generated $3.036 billion in revenue and reported GAAP and non-GAAP operating margin of 45.4% and 49.9%, respectively. Those are not typical figures for a business selling only commodity networking hardware.

The AI story is broadening as well. AI fabric adoption has broadened from 4-5 customers in 2024 to more than 100 cumulative customers. That does not remove concentration risk, but it does suggest adoption is moving beyond a tiny pioneer group.

Customer concentration and revenue timing are the main risks

That durability argument only holds if the revenue behind it stays reasonably broad and clean.

Concentration is the clearest stress point

Arista already flagged where problems could show up. Management said at least 1 and potentially 2 customers could exceed 10% of annual revenue. That is not automatically a deal-breaker, but it does mean one large customer changing plans could have an outsized impact on reported results.

That risk matters more because Arista is no longer being judged as a small fast grower. The shares are down 13.1% year to date after a 49.9% gain over the past year. Investors are now underwriting a mature leader with elevated expectations.

Deferred revenue can blur the quarter-to-quarter picture

The second watchpoint is recognition. Management specifically highlighted deferred hardware revenue as a factor that could affect how results are recognized. If large shipments get delayed or tied to longer project milestones, reported revenue can look lumpier than actual demand.

The buy case still works, but it now depends on execution

The market may still be underestimating the breadth of the story

After management raised 2026 revenue guidance for the 3rd time and pointed to growth across multiple networking lanes, the market is still reacting mainly to AI excitement and concentration concern. If horizontal network spend proves to be a larger part of the buildout than feared, that could support a broader valuation story.

What to watch next

• Guidance: Does management continue to tie growth to multiple lanes, or does AI become the only lane that matters?
• Revenue conversion: Do reported results keep reflecting the fact that the company just achieved its first $3+ billion revenue quarter?
• Concentration and timing: Does the risk that at least 1 and potentially 2 customers could exceed 10% of annual revenue start driving quarter-to-quarter volatility, or does deferred hardware revenue create more lumpy recognition?

What would weaken the thesis

If breadth fades, if big-customer dependence starts dictating results, or if raised guidance stops showing up cleanly in reported revenue, the setup gets harder.

For now, though, the core buy case still stands: strong demand, a defensible platform, and a stock that has pulled back even as the business keeps getting bigger.