ANET, Arista Networks, Inc.
Anchored by Arista's state-oriented Extensible Operating System and Network Data Lake, our network-as-a-service platform delivers a seamless, consolidated networking experience regardless of data location.
Read top to bottom, the owner's questions in the order an owner asks them: what the business is, whether the record holds, whether it survives and is any good, and what you would be paying. New to the questions? Start with the Method.
The business in brief
read the 10-K →What this business is and what moves its needle, read from the numbers in its filings. The quantitative detail is in the sections below; the verdict is left to you.
- What it is
- Revenue is Products (84%) and Services (16%).
- What moves the needle
- Volume against price, and shelf position. What decides it: whether it can raise prices without losing the customer, and whether the brand still commands its margin.
- Is it a good business?
- Return on capital has run high across the record (median 31%, above 15% in 10 of 10 years), though buybacks and expensed R&D and brands shrink the capital base, so the figure overstates the underlying economics. The steadier read is owner earnings: roughly 32% of revenue reaches owners as cash, consistently. High, durable returns can mark a moat, but whether this one is real pricing power or an accounting artifact is the judgment the 10-K is for.
Every line here is arithmetic from the company's own filings, not a model's opinion, and each figure appears in full in the sections below.
Where the money comes from
read the 10-K →Products is 84% of revenue, so this is largely a single-line business.
- Products84%$7.6B
- Services16%$1.4B
From the segment footnote of the company's own 10-K. Shares are of total revenue; the profit bar shows each segment's share of segment operating profit, before unallocated corporate costs.
The record, 2016–2025
realized figures from each filing, no estimates| 2016’16 | 2017’17 | 2018’18 | 2019’19 | 2020’20 | 2021’21 | 2022’22 | 2023’23 | 2024’24 | 2025’25 | TTMTTMMar 2026 | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| RevenueRevenue | $1.1B | $1.6B | $2.2B | $2.4B | $2.3B | $2.9B | $4.4B | $5.9B | $7.0B | $9.0B | $9.7B |
| Gross marginGross mgn | 64% | 64% | 64% | 64% | 64% | 64% | 61% | 62% | 64% | 64% | 64% |
| Operating marginOp. mgn | 21.6% | 28.6% | 12.7% | 33.4% | 30.2% | 31.4% | 34.9% | 38.5% | 42.0% | 42.8% | 42.8% |
| Net incomeNet inc. | $184M | $423M | $328M | $860M | $635M | $841M | $1.4B | $2.1B | $2.9B | $3.5B | $3.7B |
| EPS (diluted)EPS | $0.16 | $0.34 | $0.26 | $0.67 | $0.50 | $0.66 | $1.07 | $1.65 | $2.23 | $2.75 | $2.92 |
| Owner earningsOwner earn. | $153M | $616M | $479M | $947M | $720M | $951M | $448M | $2.0B | $3.7B | $4.3B | $5.3B |
| ROICROIC | 34% | 52% | 18% | 45% | 25% | 25% | 31% | 37% | 36% | 31% | 32% |
| Cash & investmentsCash+inv | $568M | $859M | $650M | $1.1B | $893M | $621M | $672M | $1.9B | $2.8B | $2.0B | $2.8B |
| Net debt / (cash)Net debt | ($568M) | ($859M) | ($650M) | ($1.1B) | ($893M) | ($621M) | ($672M) | ($1.9B) | ($2.8B) | ($2.0B) | ($2.8B) |
| Book value / shareBVPS | $0.95 | $1.33 | $1.67 | $2.26 | $2.63 | $3.14 | $3.86 | $5.69 | $7.80 | $9.70 | $10.59 |
Owner’s Scorecard
Will it survive?
- Can it pay its interest? 1427.7×ComfortableOperating income $3.9B ÷ interest expense $3M
Operating profit covers interest with the kind of margin Graham wanted for a defensive holding. Necessary, not sufficient, it says solvent, not cheap.
- ConservativeTotal debt $99M ÷ operating income $3.9B
Years of operating profit it would take to repay all debt. A first read, not a credit rating: it's gross debt (not netted against cash) over EBIT (not EBITDA), and a cyclical year distorts it.
- Debt, net of cash +$1.9BNet cashCash $2.0B − debt $99M
Cash and short-term investments exceed every dollar of debt by $1.9B, on net the company owes nothing, and can act from strength when others can't. Net debt is the leverage figure that matters; the gross ratio above ignores the cash already set against it. Strategic or illiquid investments aren't counted here.
- Capital-hungryDSO 76 + DIO 253 − DPO 73 days
Days cash is tied up between paying suppliers and collecting from customers. Lower is better; a long cycle means growth itself eats cash.
Is it a good business?
- ExceptionalNOPAT $3.2B ÷ invested capital $10.5B (debt + equity − cash)
The rate the business earns on the money tied up in it, Buffett's north star, because over time a stock tracks the ROIC beneath it. Above ~15% sustained hints at a moat; below ~8% the company may destroy value as it grows. Asset-light businesses (R&D expensed, little capital) read artificially high, pair this with Owner Earnings.
- Cash machineOwner Earnings $4.3B = operating cash $4.4B − capex $120M
What an owner could take out without starving the business. That's 47% of revenue. Treating stock comp as the real expense it is (less $439M of SBC) leaves $3.8B. Honest caveat: capex here blends maintenance and growth, so steady-state Owner Earnings may run higher (see capex vs. depreciation).
- Cash-backedCash from ops $4.4B ÷ net income $3.5B
How much of reported profit showed up as operating cash. Above 1× is reassuring; well below suggests earnings lean on accruals. One year is noisy, growth and working-capital swings distort it, and this is operating cash, not free cash. Watch the multi-year trend.
How is the cash used?
- Reinvests most of itDividends + buybacks $1.6B ÷ Owner Earnings $4.3B
Of $4.3B Owner Earnings, $1.6B (38%) went back to shareholders, $0 dividends, $1.6B buybacks. Net of $439M stock comp, the real buyback was about $1.2B. Returning most of it signals a mature cash machine; reinvesting most could mean a long runway, or empire-building. The split doesn't say which; the return earned on it (see ROIC) does.
- Investing or harvesting? 1.65×ExpandingCapex $120M ÷ depreciation $73M
Descriptive, not a grade. Above ~1× means investing faster than assets wear out (growth, or, sustained for years, today's earnings carrying less depreciation than tomorrow's will). Below means spending less than it's wearing out (efficiency, or a melting asset base). The ratio won't tell you which; the filings will.
Durability & moat, 2016–2025
A moat is a high return that doesn’t fade, reinvested at high returns. Here is what the record says, judgments, not another chart of the numbers.
- Profitable years 10 of 10
Never lost money over the record, the earnings stability Graham insisted on.
- Operating margin 22% (FY2016) → 43% (FY2025)
Margins widened over the record, pricing power intact or improving.
- Owner earnings growth +30%/yr
Free cash to owners grew about 30% a year over the record.
- Worst year 2018 · 12.7% op. margin
Stayed profitable even in its hardest year, the resilience that survives recessions.
Solvent is not the same as cheap; growing is not the same as good. These are vital signs, not a verdict, the judgment is yours, and the filing is one click away.
How the cash was used, 2016–2025
Over the record, the business generated $14.6B of operating cash, and how management split it is, as Buffett insists, the job that matters most. Here it reads as a balanced allocator, splitting cash between the business, owners, and the balance sheet.
- Reinvested$387M · 3%
- Buybacks$3.9B · 27%
- Retained (debt / cash)$10.4B · 71%
It reinvested $387M (3%) back into the business and returned $3.9B (27%) to owners, $0 in dividends, $3.9B in buybacks.
Buybacks are gross of stock issued to employees; net of that, the real return to owners is lower (see Management & pay). And the mix alone doesn't grade management, what matters is the return earned on the dollars reinvested (see incremental ROIC in the durability report).
Management & pay
Two questions Buffett actually asks about pay: is stock compensation, a real expense, whatever the income statement pretends, quietly large, and is the top wildly out of line with the floor. He's no populist about it; he just wants pay that's rational and earned, and comp committees that aren't lapdogs.
- Stock-based compensation$439M
The slice of the business handed to employees in shares this year, 5% of revenue, equal to 11% of operating profit. Buffett's oldest accounting fight: this is compensation, compensation is an expense, real whether or not the headline earnings admit it. And note the trap, the cash-flow statement adds SBC back, so the operating cash, and the owner earnings drawn from it, are flattered by exactly this amount; counted as the cost it is, what an owner keeps is lower.
Graham’s defensive-investor test
5 of 6 metGraham gave the defensive investor seven numerical criteria in The Intelligent Investor. Here they are, run mechanically on the filings, his framework, not our verdict. Meeting them is a floor of safety, not a reason to buy; missing one is no veto, since many fine modern businesses fail his strictest liquidity tests by design. The worth is in seeing exactly where a company stands against the canon, every number sourced.
- Adequate size PassRevenue ≥ $2B · $9.0B
Big enough to weather a storm. Graham's 1972 floor was ~$100M of sales (≈ $700M today); we use a $2B revenue line as a conservative modern stand-in.
- Strong liquidity PassCurrent ratio ≥ 2× · 3.05×
Current assets at least twice current liabilities, near-term bills covered without touching the business. Strict by design: many cash-rich modern firms run leaner and miss it, holding their cushion in longer-dated securities.
- Conservative debt PassDebt ≤ working capital · $99M vs $11.0B WC
Graham's rule that borrowings not exceed net current assets. Capital-heavy and buyback-heavy firms routinely fail it, read it next to interest coverage, not alone.
- Earnings stability PassA profit every year (10-yr record) · no losses
Graham wanted earnings in each of the past ten years, the stability a defensive owner leans on.
- Dividend record MissUninterrupted dividends · none paid
An unbroken dividend was Graham's mark of durability. He wanted twenty years; the filings show about ten, and a single suspension breaks the streak. Non-payers, many fine modern compounders, fall outside his defensive net by design.
- Earnings growth PassEarnings +33% over the record · +803%
At least a third more earnings than a decade ago, averaging three years at each end. Net income (not per-share), so stock splits don't distort it, buybacks and dilution show up in the share-count line instead.
- Moderate price —P/E ≤ 15 and P/E × P/B ≤ 22.5 · decided by the price
Graham's valuation gate, the wall he kept between a sound business and a sound investment. Earnings are $2.75/share and book value $9.70/share. Enter a price in “What the price implies” just below for the P/E, P/B, and whether it clears. But this is the rule Buffett outgrew: there's no hard P/E law, and a wonderful business can deserve a far richer multiple if the thesis holds, treat it as the bargain-hunter's floor, not a verdict on the price.
Graham would be the first to say a checklist is a starting point, not an answer. These are his defensive, bargain-hunter's tests, the cigar-butt lens. Buffett and Munger grew past it, paying fair prices for wonderful businesses; that lens lives in the moat and owner-earnings work above, and both still matter. Clearing Graham’s tests earns a closer look; failing them earns harder questions, not a dismissal.
What the price implies
reverse-DCFA price is the one input we don't pull, you bring it. Type today's close (read it off any broker or quote site) and see the owner-earnings growth you'd have to believe to justify it, set beside what Arista Networks, Inc. has actually delivered. Nothing is stored; the number stays in your browser.
Enter a price above to run it.
Graham capped the multiple at 15×; Buffett and Munger let that rule go, a wonderful business can deserve 50× if the thesis holds. Read it as the bargain-hunter's floor, not a ceiling on good sense.
The discount rate is your interest rate, what a dollar years from now is worth today, anchored to the long-term Treasury yield (~4–5% today, plus whatever premium you want for risk). Drag it toward the risk-free rate and watch how much growth the price suddenly “needs”: interest rates are gravity on valuations.
Owner earnings $5.3B on 1274M diluted shares; net cash $2.8B. This is a lens, not a price target, it says what you'd have to believe, not what the company is worth, and it runs on one year of (noisy) owner earnings at assumptions you can see and change.
What the filing emphasizes, FY2025
read the 10-K →Each year a 10-K must name what could go wrong, in the company's own words. Here are the ones Graham and Buffett would stop on, each set against the figure from the same filings that bears on it, anchored to a period you can find in the record above. We point; the judgment is yours.
- Customer concentrationBusiness
Who the revenue leans on. When one buyer is a large slice of sales, that buyer holds the pricing power, and its troubles become the company's.
“Two of our customers accounted for more than 10% of our sales for the year ended December 31, 2025.”
From the recordRevenue exposed (TTM)$9.7B - Pricing power & competitionRisk Factors
Whether the company sets its price or takes it. Durable pricing power is the surest mark of a moat; price competition is the surest mark there isn't one.
“This competition has resulted in increased pricing pressure, which could result in reduced profit margins, increased sales and marketing expenses and the loss of market share, any of which would likely harm our business, financial condition, results of operations and prospects.”
From the recordOperating margin42.8% (TTM), near a 10-yr high - Supplier & input dependenceBusiness
A choke point upstream. A sole or limited supplier can dictate terms, and a single shortage can stop the line.
“Our products rely on key components, including merchant silicon, integrated circuit components and power supplies, which are purchased from a limited number of suppliers, including certain sole source providers.”
From the recordGross-margin cushion (TTM)64% - Concentrated dependenceMD&A
What the whole business leans on, a product, a platform, a partner. Concentration cuts both ways, and the filing is where management has to admit it.
“Historically, a limited number of customers have accounted for a significant portion of our revenue.”
From the recordOwner-earnings margin at stake (TTM)54% - Litigation & contingenciesRisk Factors
Claims an owner inherits. Most disclosure is boilerplate; this fires only on an actual matter, a named suit, a settlement, a contingency, a number.
“If we fail to meet or exceed such guidance or expectations for these or any other reasons, the market price of our common stock could decline substantially, and we could face costly lawsuits, including securities class action suits.”
A judgment, not a number, weigh it against the filing yourself. - DilutionRisk Factors
Whether your slice quietly shrinks. New shares fund the company at the existing owner's expense.
“If we raise additional equity financing, our stockholders may experience significant dilution of their ownership interests, and the market price of our common stock could decline.”
A judgment, not a number, weigh it against the filing yourself. - Cyclicality & demandRisk Factors
How the business behaves when the economy turns. A cyclical earns its keep across the whole cycle, not at the peak.
“Cyclical changes in our customers' demand for our products and services, particularly changes in the demand of our largest customers, could result in fluctuations in our revenue, revenue growth and results of operations.”
From the recordWorst year on record12.7% operating margin (FY2018)
What changed, FY2025 vs FY2024
read the 10-K →Most of a 10-K is boilerplate carried over verbatim; the signal is in what's new. These lines appear this year and weren't there last, figure updates filtered out, so only the language shift remains.
- “Operating cash inflows consisted of an increase in deferred revenue of $2.5 billion resulting from an increase in product deferred revenue related to customer contracts with acceptance terms and increased customer PCS contracts, and a $379.9 million increase in accounts payable and other liabilities…”
- “Our network-as-a-service approach now empowers customers of all sizes to seamlessly leverage their data through offerings spanning three key categories: Core (AI, Cloud, and Data Center Networking), Cognitive Adjacencies (Campus and Routing), and Cognitive Networks (Software and Services).”
- “Our material cash requirements include the following contractual and other obligations: Purchase Obligations We outsource most of our manufacturing and supply chain management operations to third-party contract manufacturers, who procure components and assemble products on our behalf.”
- “This strategy and differentiation have also allowed us to deliver our comprehensive suite of products, services, and technologies to a global customer base segmented into three primary categories: Cloud and AI Titans, AI and Specialty Providers, and Enterprise.”
- “New and changing laws, regulations, executive orders, directives, and enforcement priorities can adversely affect the Company's business by increasing the Company's costs, limiting the Company's ability to continuously navigate global supply chain options in lieu of optimizing tariff outcomes, offer…”
Classic text analysis over the filing itself, no model wrote a word of this, and every quote is the company's own.
Peers, Computer hardware
The same industry, side by side on owner economics, compare, don't rank by a single number.● marks best in the group.
| Company | Revenue | Gross margin | Op. margin | ROIC | Owner earn. margin |
|---|---|---|---|---|---|
| HPQHP Inc. | $55.3B | 21% | 5.7% | 53% | 5% |
| HPEHewlett Packard Enterprise Company | $34.3B | 59% | -1.3% | -1% | 2% |
| SMCISuper Micro Computer, Inc. | $22.0B | 11% | 5.7% | 77% | 7% |
| WDCWestern Digital Corporation | $9.5B | 39% | 24.5% | 14% | 13% |
| PANWPalo Alto Networks, Inc | $9.2B | 73% | 13.5% | 12% | 38% |
| STXSeagate Technology Holdings PLC | $9.1B | 35% | 20.8% | 50% | 9% |
| ANETArista Networks, Inc. | $9.0B | 64% | 42.8% | 30% | 47% |
| FTNTFortinet, Inc. | $6.8B | 80% | 30.7% | — | 33% |