Cisco Systems, Inc. CSCO

Key data

The business

Cisco sells the plumbing of computer networks: the switches and routers that move data inside corporate buildings and data centers, plus the security, observability and networking software that sits on top. The customers are enterprises, telecom carriers and, increasingly, the hyperscale cloud operators building AI clusters, and almost all of it moves through a two-tier channel of distributors and resellers rather than direct. The money is made twice, once on the box when it ships, and again at higher and steadier margin on the software subscriptions and support contracts attached to it, a base that now carries $46.7 billion of committed future revenue. The moat is switching costs: once a network runs on Cisco's operating software and the staff is trained on it, ripping it out is expensive and risky, which is why gross margin has sat near 65% for years. The Splunk acquisition, folded in fully last year, bolted a large security and data-analytics software business onto that base.

The numbers

The five most recent quarters show a business reaccelerating, not drifting.

Revenue rose every quarter and then jumped in the fourth to $17.3 billion, the largest three months in Cisco's history, up 17.6% on the year and 9% on the quarter, with net income up 51%. Non-GAAP earnings of $1.22 beat the $1.17 the Street looked for. The watch from the last look, whether this year's AI-infrastructure orders were a one-quarter stock-up, held: product demand kept building into the fourth quarter rather than fading. That is the inflection, and it runs well ahead of the low-single-digit growth Cisco posted through most of the prior three years.

Step back to fiscal years and the breakout stands out against a slow base.

From fiscal 2022 to 2026 revenue compounded at about 5% a year and earnings per share at about 4%, the profile of a mature, cash-rich incumbent. Fiscal 2026 broke that pattern: revenue up 11.8% and EPS up 30.6%, lifted by a full year of Splunk and by the AI orders. The Street now models $73.1 billion for fiscal 2027, up 15%, and $5.12 of non-GAAP earnings. What this memo doubts the tape has settled is whether that growth is the new rate or a one-year bulge from Splunk timing and an AI order surge. The print that decides it is product-order growth and gross margin over the next two quarters.

For a hardware maker the tell is not the headline but the inventory and the margin underneath it.

Inventories nearly doubled to $5.7 billion, about 92 days of sales, while gross margin slipped to 64.5% from 64.9%. The 10-K flags excess inventory as a direct risk to gross margin, and the two-tier channel makes it hard to see whether that stock is pre-positioning for real AI demand or building ahead of orders that may not come. Committed future revenue rose 7% to $46.7 billion, which argues the demand is real, for now.

Management

Chuck Robbins has run Cisco since 2015. Over the past year insiders sold about $17.4 million of stock across 63 transactions and bought nothing, the largest a $1.65 million sale by Robbins in August; the filings do not disclose whether those sales were under preset plans, so read them as routine rather than a signal either way. Capital return is the real story: Cisco paid $6.55 billion in dividends and repurchased $6.1 billion of stock in fiscal 2026, together about 99% of free cash flow, with $8.1 billion of buyback authorization left. That discipline sits against $29.5 billion of total debt, up from roughly $8 billion before the Splunk deal, held against $15.9 billion of cash and investments.

How it fails or surprises you

Inventory turns to writedowns. Inventories doubled to $5.7 billion, 92 days of sales, the one number this read explains least well. If the AI orders behind that build slow, Cisco is left discounting stock into a two-tier channel and gross margin, already at 64.5%, gives way. Watch inventory days rising while gross margin falls over the next two prints; a second such quarter would confirm it.

AI networking is a multi-year build (right tail). The Street models 15% revenue growth for fiscal 2027 and prices the stock at 21 times that. If hyperscalers and enterprises keep buying Ethernet fabric for AI clusters at this year's pace, growth holds in the mid-teens for years and the multiple re-rates rather than compresses. The market is paying for one strong year, not a sustained cycle. The tell is product-order growth staying double digit into fiscal 2027.

The multiple reverts to the compounder. At about 33 times trailing earnings the stock sits above any level in twelve years, priced as a grower. If fiscal 2026 proves a Splunk-and-AI bulge and growth falls back toward the 5% it averaged over the prior four years, the multiple compresses toward the high-teens it long carried, and the earnings gain is swamped by the de-rating. The pair to watch is revenue growth against the 15% now in estimates.

Closing thoughts

What matters is whether the AI networking wave that drove this year's 17.6% fourth-quarter growth keeps coming or fades back to the low-single-digit drift Cisco showed for years, and the data that tells you which is product-order growth and gross margin, quarter by quarter through fiscal 2027. The survival question is settled, Cisco is entrenched and cash-rich, so the only uncertainty is the growth rate, and two or three clean quarterly prints turn that uncertainty into fact. If orders decelerate while the stock still carries 33 times trailing earnings, you've paid a grower's multiple for a compounder and the downside is the re-rating; if they hold double-digit, the market was too cautious and the upside is the multiple expansion. The fatter tail is modestly to the downside from valuation: the business is excellent and the balance sheet sound, so what is at risk is the multiple, not solvency, while the upside if the AI cycle is real is a genuine re-rating.

The bet is still that companies keep building bigger networks and keep buying Cisco to run them, with AI data centers as the swing factor. What breaks it is the inventory build revealing itself as orders pulled forward, and the pair of numbers that tells you first is product-order growth and gross margin, quarter by quarter through fiscal 2027. If product growth slides back toward low single digits while the stock still carries a grower's multiple, the read was wrong.