Every Trigger We Set in May Fired. The Stock Fell 8.4% Anyway.

Results vs. Consensus

Year-over-Year Comparison

Sequential Comparison

Fiscal Year 2026 vs. Fiscal Year 2025

Revenue

Revenue of $17,252M is the largest quarter in Cisco's history and the growth rate of 17.6% is the fastest since the post-pandemic backlog release. It beat the guide's high end by $352M and consensus by $416M, which is a wider beat than either of the prior two quarters. The composition is the interesting part. Product revenue grew 23.6% and services grew 0.2%, so the entire acceleration is hardware, and within hardware the entire acceleration is networking, which grew 28.3% and now represents 72.7% of product revenue and 56.8% of the company.

Three separate engines are running inside that networking number and management sized all three. Hyperscaler AI infrastructure produced approximately $4B of revenue across fiscal 2026, or roughly 6% of the company against less than 2% in fiscal 2025. Campus product orders grew 20%. And roughly 5 points of the quarter's revenue growth is list-price increase rather than incremental units, on the CFO's own arithmetic. Strip the price effect and the underlying volume growth is closer to 12% or 13%, which is still the best quarter in a decade and is a materially different number from 18%.

Margins

Non-GAAP gross margin of 66.3% landed 30 basis points above the guide midpoint and 30 basis points above the prior quarter, which is the first sequential improvement of the down-cycle. It is also 210 basis points below the year-ago quarter, and product gross margin of 64.8% is 270 basis points below. The attribution the CFO gave was "higher hardware mix and memory costs, partially offset by productivity improvements and price increases," which is the same two-factor explanation as the prior two quarters with the offsets now visible in the number rather than promised.

The geographic split has changed shape in a way that matters. In Q3 the entire company-level decline sat in the Americas, where the hyperscaler revenue is booked, while EMEA was flat. This quarter the Americas fell 350 basis points and EMEA fell 160, but APJC rose 310. Cisco is no longer running a single-region margin problem, and the regions carrying the AI mix are the ones absorbing the compression, which is exactly what a mix explanation predicts and a component-cost explanation does not.

The line that should govern the reading is operating margin. At 35.9% it expanded 160 basis points year over year on a gross margin that contracted 210, because non-GAAP operating expenses fell to 30.4% of revenue from 34.1%. Management has been explicit that this is now the metric it wants to be judged on, and the fiscal 2027 guide takes the full-year figure to roughly 35%, which would be a company record. The bear reading is that a company redirecting attention from gross margin to operating margin is telling you the gross margin is not coming back. That reading is correct. It is also compatible with the stock being cheap.

EPS

Non-GAAP EPS of $1.22 beat the guide's high end by four cents and grew 23.2%, comfortably outpacing the 17.6% revenue growth for the second consecutive quarter. The quality is good: the non-GAAP tax rate of 18.8% was 70 basis points higher than a year ago rather than lower, the diluted share count fell only 0.2%, and net income grew 23.2%, identical to EPS. There is no below-the-line contribution and no buyback contribution of consequence. This is an operating result.

Full-year non-GAAP EPS of $4.33 exceeded the guide set ninety days earlier by four cents and exceeded the initial fiscal 2026 guide, issued in August 2025 at $4.00 to $4.06, by 6.7% at the high end. That pattern is the single most useful piece of context for the fiscal 2027 guide, and we return to it in the valuation section.

Segment Performance

Networking

Networking revenue of $9,791M grew 28.3%, accelerating from 25% in Q3, 21% in Q2 and 15% in Q1, and closed fiscal 2026 at $34,668M, up 22% on a base that had shrunk 3% the prior year. Product orders grew 40%, marking the eighth consecutive quarter of double-digit order growth, driven by triple-digit growth in service provider routing and Acacia optics and double-digit growth in data centre switching, compute, campus switching, wireless, enterprise routing and industrial IoT.

"Networking product orders grew 40% in Q4, driven by triple-digit growth in service provider routing, and Acacia optics. And double digit growth in data center switching, compute, campus switching, wireless, enterprise routing, and industrial IoT products. Q4 marked the eighth consecutive quarter of double digit growth for our networking portfolio overall, supporting our view that we are in a multiyear multibillion dollar networking super cycle."
— Chuck Robbins, Chair and CEO

The order rate decelerated from Q3's "more than 50%," which is the arithmetic of a larger base rather than a demand change: the absolute order dollars grew. Data centre networking orders grew more than 35% and campus orders grew 20%. The one disclosure worth flagging is that more than half of Cisco's customers now buy both campus and data centre networking, which is the platform argument stated as a purchasing fact rather than a strategy slide.

Assessment: This is the same judgment we made in May, now with a year of evidence behind it. A segment of this size compounding above 20% across three independent demand pools is a different business from the one that shrank 3% in fiscal 2025, and it is revenue rather than backlog. The margin consequence is unchanged and is the price of the growth: this is the lowest-gross-margin hardware in the portfolio and it is now 57% of the company.

Hyperscaler AI Infrastructure

Hyperscaler AI infrastructure orders of $4.0B in Q4 compare with $1.9B in Q3 and more than $0.8B in the year-ago quarter, closing fiscal 2026 at $9.3B against a forecast of approximately $9B set ninety days earlier and approximately $5B set ninety days before that. The full-year figure is roughly 4.5 times the fiscal 2025 total. Revenue delivered was approximately $4B, or roughly 6% of the company, against less than 2% in fiscal 2025. Fiscal 2027 AI infrastructure revenue is guided to approximately $7.5B, an 88% increase and roughly 10% of the guided revenue base.

Three new hyperscaler design wins landed in the quarter. One was a P200-powered system for scale-across, taking the total to three scale-across wins across three separate hyperscalers, and management disclosed that all three placed orders during Q4. The second was a G200-powered scale-out win. The third was a managed optical fibre network win that positions Cisco as an alternative to an incumbent supplier. Four of the top hyperscalers each grew AI infrastructure orders in triple digits.

"We took $4 billion in orders in Q4, bringing the total for FY 26 to $9.3 billion, approximately 4.5x our fiscal year 25 total. The mix of these orders in both Q4 and FY 26 was approximately 60% Silicon One based systems and 40% optics."
— Chuck Robbins, Chair and CEO

The disclosure regime changes in fiscal 2027. Cisco will publish an annual AI revenue target rather than an annual order target, with orders described only as "meaningfully higher" than fiscal 2026 and updated qualitatively through the year on design wins and orders taken.

Assessment: The single most important thing this line did was validate the forecasting cadence. In May the $9B forecast required a 95% sequential jump in Q4 that management defended with an assertion about nonlinearity rather than a backlog figure, and we wrote that the scale-across wins contributed nothing to it. The jump happened, it overshot, and the scale-across wins are now producing orders. That converts management's forward AI numbers from claims into a track record, which is exactly what the $7.5B fiscal 2027 revenue guide needs in order to be underwritten. The offsetting development is the disclosure change, which removes the metric the market had been marking to at the moment it became most useful.

Acacia and Optics

Acacia booked more than $1B of orders for a second consecutive quarter. Cumulative coherent pluggable shipments now exceed 850,000 units at 400G and 75,000 units at 800G, against more than 750,000 and 40,000 respectively a quarter ago, so 800G shipments nearly doubled in three months. Optics represented approximately 40% of the AI order mix in both Q4 and the full year. Management added a first design win for a managed optical fibre network using Cisco line-system technology, which allows Cisco digital coherent optics to be deployed inside third-party equipment.

Assessment: The 800G shipment curve is the number to watch here, because scale-across is an 800G problem and management sizes the traffic opportunity at roughly 14 times traditional data-centre interconnect. Optics remains the piece of the AI story with the shortest competitive half-life, and the managed-optical-network win is the first evidence that Cisco can sell the component into someone else's system rather than only inside its own. That is a wider addressable market and a thinner moat at the same time.

Campus Networking and the Mythos Pipeline

Campus networking product orders grew 20%, decelerating from more than 25% in Q3 while remaining the second-largest growth engine in the company. Wi-Fi 7 represented more than half of total wireless orders. Industrial IoT orders have now grown double digits for nine consecutive quarters and accelerated again. Management named a leading frontier AI company as a campus customer taking Wi-Fi 7 access points, smart switches and end-to-end segmentation.

The new element is a demand driver that did not exist a quarter ago. Management repeatedly attributed a building pipeline to what it called the Mythos effect: the arrival of AI-driven vulnerability discovery has made equipment past its last day of support, which no longer receives patches, unacceptably risky to operate. Between 8,000 and 9,000 customers are now running Cisco IQ to assess which of their installed infrastructure is patchable and which is not.

"We have got customers that are running Cisco IQ that are assessing their infrastructure, understanding what is last day of support, what can be patched, what cannot be patched, we have seen the pipeline increase meaningfully as a result of Mythos, which is really showing up as a network refresh, but is probably underneath viewed as a cybersecurity spend."
— Chuck Robbins, Chair and CEO

Management stated twice, unprompted, that Mythos and last-day-of-support refresh are not yet material in the reported numbers. The competitive angle is that a competitor's end-of-support installed base is as addressable as Cisco's own, and management gave a specific customer example of exactly that displacement.

Assessment: This is the most consequential new item in the print and the market spent no time on it. A refresh cycle driven by security obsolescence rather than by capacity is budget-inelastic in a way that a capacity refresh is not, because the alternative to spending is operating known-unpatchable infrastructure. It converts the campus story from a discretionary modernisation argument into a compliance-adjacent one, which is the framing that carried enterprise security budgets for the past five years. Critically for the rating, none of it is in the fiscal 2027 guide by management's own statement, so it is upside optionality rather than an assumption we are being asked to pay for. It is also, for the same reason, unfalsifiable this quarter.

Security

Security revenue of $2,226M grew 14.0%, the strongest quarter since the Splunk acquisition anniversaried and a sharp reversal from flat in Q3, 4% down in Q2 and 2% down in Q1. The full year finished up 2% at $8,232M. Orders grew double digits across the entire portfolio including Splunk, with firewall orders above 30% for a second consecutive quarter. Approximately 1,500 customers bought the new products in Q4, taking the cumulative total above 6,400 since launch. Splunk added more than 280 new logos and exceeded its 1,000 new-logo target for the year.

The CFO qualified the 14% himself, without being pushed to.

"But that 14% did have, and you kind of alluded to this, a bit of the fact, Did have a little bit of anomaly in there. You know, we saw Splunk during the quarter, and it always just depends on customer preference. And during the quarter, we had some sizable on prem deals. That also had a little bit longer duration. So, that significantly helped the, the overall security number, in terms of revenue."
— Mark Patterson, CFO

On-premises Splunk deals recognise revenue upfront rather than ratably, so a mix shift toward on-premises inflates a single quarter. Management expects the long-run mix to return to roughly two-thirds cloud and one-third on-premises, guides Q1 security to mid-to-high single digits, and guides fiscal 2027 security and observability together to high single digits from low single digits in fiscal 2026.

Assessment: The commitment carried into this quarter was that the organic Cisco security portfolio would exit fiscal 2026 approaching double-digit revenue growth. The CEO stated it landed at high single digits and claimed that as delivery. That is a fair claim on the organic portfolio and it is the first of the security commitments to be met in four quarters. The reported 14% is not the run rate and management said so first, which is the right behaviour and also means the segment enters fiscal 2027 with a harder comparison than the headline implies. We treat high single digits as the honest exit rate.

Services and the Recurring Franchise

Services revenue of $3,793M was essentially unchanged year over year, ending a run of three consecutive quarters of 1% declines, and the full year finished at $15,030M against $15,046M. Management guided services to turn positive in fiscal 2027 with gradual improvement through the year to low single digits. Services gross margin of 71.6% expanded 80 basis points and was the only margin line that improved.

The recurring metrics beneath services reversed direction for the first time in four quarters. Total ARR ended at $32.1B, up 3% against 2% in Q3. Product ARR grew 5% against 4%. Total RPO of $46,734M grew 7% against 4%, with product RPO of $23,436M up 9% against 6% and services RPO up 6% against 2%. Total software revenue of $6.2B grew 11% against 0.7% in Q3. Subscription revenue represented 48% of total revenue, down from 49.2% in Q3 and 54% a year ago, which is denominator arithmetic in a quarter when product revenue grew 24%.

Assessment: This is the change that should move the rating and it received almost no attention on the call or in the coverage. Every line in this metric set had decelerated for three consecutive quarters, and we wrote in May that it was "the clearest single argument for the downgrade." All of them turned. Software revenue going from 0.7% to 11% is the most striking of them, and it is partly the same Splunk on-premises duration effect that flattered security, so we discount it. Total RPO accelerating three points to 7% with product RPO at 9% is harder to explain away, because RPO is contracted future revenue and it moved in the same direction as orders. Services ARR was not disclosed this quarter, having been negative for three straight, which is the one place the disclosure went backwards.

Key KPIs

Key Topics & Management Commentary

Overall Management Tone: The most forward-leaning posture of the six quarters we have covered, and the first in which the company argued a multi-year thesis rather than defending a quarter. Management volunteered the decomposition of its own guide, named the price contribution without being asked twice, and pre-emptively qualified the security number that flattered the print. The one place the tone hardened rather than opened was disclosure: the annual AI order target, which management created two years ago to demonstrate the size of the opportunity, was retired on the same call that the opportunity got larger.

1. The Fiscal 2027 Guide Is the Print

Cisco guided fiscal 2027 revenue to $72.2B to $73.4B and non-GAAP EPS to $5.05 to $5.11. Against a pre-print consensus of $68.69B and $4.64, that is 5% to 7% above on revenue and 9% to 10% above on earnings. The midpoints imply 15.0% revenue growth and 17.3% EPS growth, both accelerating from fiscal 2026's 11.8% and 13.6%. For a company whose stated long-term model is 4% to 6% revenue growth, this is a different corporate identity, and the CFO framed it that way.

"So I think as you look at the full year, you know, we ended FY 2026 with double digit top line and bottom line growth, 12%, 14%. We see acceleration in both of those. So accelerating to 15% on the top line, 17% on the bottom line. Again, you know, driving that, operating leverage with bottom line growth faster than the top line. With the $7.5 billion in hyperscale revenue, if you kind of back that out, you get to the core business growing at about 10%, which is obviously significantly faster than that investor day that we have done awfully long ago at this point."
— Mark Patterson, CFO

Underneath the total, management guided security and observability from low single digits in fiscal 2026 to high single digits in fiscal 2027, and services from flat to low single digits positive. Both are the first positive segment-level forward commitments Cisco has made on those lines in two years.

Assessment: An initial annual guide that exceeds consensus EPS by 9% to 10% is rare, and it is rarer for a company with Cisco's history of guiding conservatively into a year and raising through it. The relevant benchmark is Cisco's own: the initial fiscal 2026 guide, issued twelve months ago, was $4.00 to $4.06, and the year printed $4.33, 6.7% above the high end. If fiscal 2027 behaves the same way, $5.11 is a floor rather than a ceiling. We do not model the full historical overshoot, for reasons set out in the model section, but we do model above the range.

2. Where the Fifteen Percent Actually Comes From

Three components produce the guided growth and management supplied the size of two of them. Hyperscaler AI revenue goes from approximately $4B to approximately $7.5B, contributing 5.5 points of company growth. Price increases taken to recover memory cost contributed approximately 5 points of revenue growth in Q4 and are planned at 4 to 5 points across fiscal 2027, weighted to the first half as the increases lap. The residual, which is unit volume across everything else, is therefore roughly 5 points.

"In terms of the price increases, you know, as you look at the impact on the financials, it was about 5 points in terms of top-line revenue growth that we saw in Q4. As we look to FY 2027, we are going to begin to lap those price increases, that we really started to put in place in the second half there. So you have more of an impact in the first half than the second half. And, overall, we are we are sort of planning for that kind of 4 to 5 points of impact this year as well."
— Mark Patterson, CFO

Management also volunteered that the price increases are single-digit, apply to hardware rather than software, and are targeted at products with high memory content. Roughly 15% to 20% of Cisco's bill of materials is memory-related, against roughly two-thirds for server vendors, which is why Cisco's pass-through is measured in single digits where server pricing has moved far more.

Assessment: This is the single most important qualification on the guide and management stated it plainly, which is to their credit. Price-driven revenue arrives with no incremental gross profit and reverses when the underlying cost normalises, so 4 to 5 points of the 15% is borrowed rather than earned. Strip it out and the underlying growth is roughly 10% to 11%, of which AI is 5.5 points and the rest of Cisco is about 5, which is exactly the long-term model. The correct reading is not that the guide is inflated. It is that Cisco is a 5%-volume-growth company with a very large AI business bolted on and a temporary pricing tailwind, and the durable acceleration is the AI half.

3. Gross Margin Is Guided Down, and Management Reframed the Metric

Q1 fiscal 2027 non-GAAP gross margin is guided to 65% to 66%, a midpoint 80 basis points below the 66.3% just delivered. No full-year gross margin range was given. The CFO described a "slight gross margin headwind as we move through FY 2027" and redirected to operating margin.

"You know, first off, as Chuck said, we are in this networking super cycle, which, we are very early in the cycle, and we are shipping high volumes of hardware, which is increasing the revenue mix to hardware considerably. So you should expect a slight gross margin headwind as we move through FY 2027 as we address these very high growth opportunities. And operating margin is really a better indicator of our profitability as our guide implies operating margin of about 35% for the full year FY 27, which would be a high watermark for us as a company."
— Mark Patterson, CFO

The CEO extended the argument into a capital-allocation claim: the incremental expense required to capture hyperscaler growth is minimal relative to the revenue, so business that carries a lower gross margin can be more profitable at the operating line than the company average.

"This business, because of the magnitude of it and the and the growth rates, it allows us to take even in some cases, lower margin business from a gross margin perspective that actually turns out to be highly profitable because we do not have to add incremental expenses to go gather that business. So these are strategic decisions we are making about business to pursue, and we feel really good about the profitability impact of those businesses."
— Chuck Robbins, Chair and CEO

Assessment: This is a coherent argument and it is supported by the quarter's arithmetic: 210 basis points of gross-margin decline produced 160 basis points of operating-margin expansion because operating expenses fell 370 basis points as a share of revenue. It is also a permanent structural downgrade to the gross-margin line dressed as a metric preference, and investors are entitled to notice that the metric changed at the same time as the answer. The test is whether the operating expense ratio can keep falling. It fell 370 basis points this quarter against a hardware mix that will be heavier next year, and Q1 is guided to fall another 90.

4. Operating Leverage Answers the Question We Opened in May

In May we added a bear point on the grounds that the $1B restructuring was explicitly "not a savings-driven restructure," so fiscal 2027 operating expenses would grow with the business and Q3's leverage would not repeat. Q4 answered it: non-GAAP operating expenses grew 4.8% in dollars while revenue grew 17.6%, and the ratio fell to 30.4% from 34.1%. For the full year the ratio fell to 32.1% from 34.3%, producing 220 basis points of leverage and an operating margin that expanded 40 basis points despite gross margin falling 180.

The forward guide extends it. Q1 fiscal 2027 implies an operating expense ratio of roughly 29.5%, another 90 basis points of leverage, and the full-year operating margin of about 35% would be a company record. The CFO gave a compact version of the mechanism when asked about the margin trade.

"So I think Q4 is a great example of that. Gross margins on a year over year basis were down 2.1%. But OpEx was also down 3.7% as a percentage of revenue which then, allowed us to move from operating margins or op inc as a percentage of revenue from 34.3% to 35.9%. So really showing, incredible scale and productivity."
— Mark Patterson, CFO

Assessment: The bear point we raised ninety days ago is answered for one year, not permanently. What Q4 demonstrated is that hyperscaler revenue genuinely carries a near-zero incremental cost to serve, which is the structural reason the leverage works and the reason it should persist while the AI mix is rising. What remains unresolved is the steady state: once the AI mix stops rising, the gross-margin drag stops falling away with it, and operating margin becomes hostage to the gross-margin level rather than to its rate of change. That is a fiscal 2029 question, not a fiscal 2027 one.

5. AI Disclosure Moves From Orders to Revenue

Asked how fiscal 2027 AI would be reported, the CFO said Cisco is retiring the annual order target that it had published and twice raised through fiscal 2026, replacing it with the $7.5B revenue target plus qualitative updates on design wins and orders through the year. Orders are expected to be "meaningfully higher" in fiscal 2027 without a figure attached.

"As we get into FY 2027, we will move to a more traditional revenue target, which we gave you a $7.5 billion. You are gonna get regular updates from us. On forward looking metrics, be it design wins, orders taken, etcetera, as we move through the year. So we will continue to update you on that front as well. Just will not put out a, you know, an annual target. But instead, we will have an annual revenue target. You should, though, expect orders to be meaningfully higher, I would say, in FY 2027 than they were in FY 2026."
— Mark Patterson, CFO

The stated rationale is that the order target existed to demonstrate the size of the opportunity while Cisco was establishing itself in the category, and that job is done.

Assessment: The rationale is defensible and the timing is unfortunate. The order number was the leading indicator that let the market underwrite the revenue number, and it was the disclosure that made the two upward revisions in fiscal 2026 legible as they happened. Removing it converts a quarterly test into an annual one and shifts the burden of proof back onto management's credibility, which is at its highest point of the cycle precisely because the order disclosure existed. We read this as a mild negative on transparency rather than as a signal about demand, because a company expecting a weak order year would not volunteer "meaningfully higher."

6. The $7.5 Billion Revenue Guide Against a $9.3 Billion Order Book

The most direct challenge on the call put it plainly: with roughly $3B of the fiscal 2026 order book recognised as revenue in fiscal 2026 by the questioner's arithmetic and $9.3B booked, a $7.5B fiscal 2027 revenue target looks conservative. The CFO corrected the arithmetic and defended the number on shipment timing rather than on demand.

"You know, we did an excess of $9--we did $9.3 billion in orders last quarter. We have now given you a revenue number of $7.5 billion. We that is that is revenue and not orders, just to be clear, in FY 2027."
— Chuck Robbins, Chair and CEO

Management added that the fiscal 2027 AI revenue should be modelled as a gradual quarterly increase from Q1 to Q4 rather than as a step function.

Assessment: "Prudent" is the operative word and it was chosen deliberately. Cisco enters fiscal 2027 with roughly $5.3B of fiscal 2026 orders undelivered on the CFO's own figures, before any fiscal 2027 orders, against a $7.5B revenue target. That does not make $7.5B wrong, because hyperscaler orders are placed far ahead of delivery and some of that backlog is fiscal 2028 revenue. It does mean the guide is set where a beat is more likely than a miss, which is the same posture that produced two upward revisions last year. This is the largest single source of upside to the fiscal 2027 number.

7. Scale-Across Converts From Design Win to Order

The P200 scale-across programme, which in May consisted of two design wins contributing nothing to the fiscal 2026 order target, now has three wins across three separate hyperscalers and all three placed orders during Q4. Management sizes AI-driven scale-across traffic at roughly 14 times traditional data-centre interconnect and has line of sight to multiple further design wins across G300, G200, P200 and optics over the next six months.

"And we are uniquely positioned because in that space, we have got our P200, which we talked about having 3 wins already in that space across 3 different hyperscalers. By the way, we started taking orders from all 3 of them in Q4, so that is positive."
— Chuck Robbins, Chair and CEO

Separately, management disclosed a plan to roll Silicon One across the entire high-performance networking systems portfolio by fiscal 2029, which is a vertical-integration commitment rather than a product announcement.

Assessment: Scale-across is the part of the AI thesis with the longest duration and the clearest architectural logic: power and physical limits inside a single data centre force distribution across facilities, and distribution is a networking problem by definition. Three wins with three separate buyers, all now ordering, is the evidence that this is a market rather than a customer. The Silicon One-everywhere commitment by fiscal 2029 is the more interesting disclosure for gross margin, because owning the silicon removes a layer of merchant margin from the cost stack, and management said as much when discussing TSMC.

8. Enterprise Budgets and the Shift From Discretionary to Mandatory

Two separate questions probed whether enterprise IT budgets can sustain the spending Cisco is seeing across the stack. The CEO's answer described a reallocation rather than an expansion, and named the mechanism that makes it durable.

"Now as it relates to budgets, what I think is going on, we are hearing from our customers, is they are they are currently reprioritizing within their existing budgets. But I would also say that you are beginning to see a trend where they are looking at AI readiness, Mythos readiness, quantum readiness, in a similar vein to how they have looked at cybersecurity spend over the last 3 to 4 to 5 years. it is just not optional."
— Chuck Robbins, Chair and CEO

Enterprise product orders accelerated to 21% and public sector to 30%, with US Federal named as the driver of the Americas acceleration. Telco orders grew more than 30% against 9% in Q3, which management attributed to carriers building bandwidth ahead of scale-across demand from the cloud providers.

Assessment: The reallocation framing is honest and it is the right thing to worry about, because a budget that is reprioritised rather than expanded eventually runs out of things to deprioritise. The countervailing argument is the one management makes: spending forced by an unpatchable installed base behaves like compliance spending rather than capital spending, and compliance budgets are found rather than allocated. The telco acceleration from 9% to 30% is the datapoint that supports the supercycle framing most directly, because telcos are the slowest-moving buyer in the set and they moved.

9. Supply Chain and the Direct TSMC Relationship

Two questions covered component availability. Management reported no significant lead-time issues and no customer escalations, and located the advantage in disintermediation: Cisco buys wafers, substrates, assembly and test directly from TSMC rather than through a merchant silicon supplier, so it is not subject to allocation and de-commitment from a third party.

"I just think that it is important from the standpoint of sort of managing our own destiny and controlling that supply, but also, our own innovation, pipeline and being able to have that much more control over that. I think those are those are the 2 really big things early on. I think in out-years, you are going to see, obviously, I think some avoidance of stacking margins with merchant silicon as well, but that is kind of down the road just a little bit."
— Mark Patterson, CFO

On memory specifically, management named a strategic agreement with Nanya, described more than 30 internal programmes targeting memory efficiency, and gave one worked example: Wi-Fi 7 memory utilisation was reduced by 50% inside roughly 90 days. Management stated it has adequate supply to meet the fiscal 2027 guide and upside to it.

Assessment: The "avoidance of stacking margins with merchant silicon" line is the gross-margin answer that management did not give when asked about gross margin directly. Vertical integration into silicon is a multi-year cost-structure argument that lands after the current mix headwind, which is consistent with the fiscal 2029 Silicon One rollout timeline. It does not help the fiscal 2027 gross margin and management did not claim it would.

10. The Working Capital Bill Keeps Arriving

The cost of the supply-chain position sits on the balance sheet. Inventories closed at $5,694M, up 80% year over year and 21% sequentially, with a $992M outflow in the quarter. Financing receivables rose to $8,332M from $6,527M, with a $1,801M outflow in Q4 alone as Cisco extended more customer financing against the AI ramp. Total debt rose $1,440M year over year to $29,533M while cash and investments fell $192M to $15,918M.

Against that, operating cash flow recovered sharply: $5,386M in Q4, up 27%, the strongest quarter of the year and enough to bring full-year operating cash flow to $14,177M, flat against fiscal 2025. Free cash flow of $12,767M funded $12,659M of capital return, or 99%, comprised of $6,553M of dividends and $6,106M of buybacks. Cisco raised its dividend for a fifteenth consecutive year and repurchased 13 million shares in Q4 at an average of $111.53.

Assessment: This is materially better than the nine-month picture we described in May, when capital return was running at 122% of free cash flow and total debt was up $3,210M since the year end. Q4's cash generation closed most of that gap. What has not improved is the structure: inventory and financing receivables together absorbed $2,793M of cash in a single quarter, capital return is at 99% of free cash flow rather than below it, and the buyback is being funded at the margin by debt rather than by surplus. The fiscal 2026 10-K, which will carry the inventory purchase commitment figure that stood at $16,033M a quarter ago, is the disclosure that will settle how large this position really is. It had not been filed as of this writing.

11. Collaboration Quietly Had Its Best Quarter in Seven Years

Collaboration revenue of $1,167M grew 12% and the segment posted double-digit order growth and its best quarterly performance in seven years, with video devices growing 40%. The full year finished up 4% at $4,300M. This is a segment that has been in structural decline since the hybrid-work pull-forward reversed, and it received one sentence in prepared remarks and no questions.

Assessment: A 12% quarter in Collaboration is not a thesis, but it is the third of Cisco's four product groups to inflect positive within twelve months, after Networking and now Security. The read-through is to the device refresh cycle rather than to Webex software, given video devices at 40%. It is worth noting mainly because the bear case on Cisco has for years rested on the idea that only one segment could grow at a time.

Guidance & Outlook

The guide is unusual in three respects and each is worth separating. First, it is an initial full-year guide that exceeds consensus earnings by 9% to 10%, which for a company with Cisco's guide-low-and-raise history is not a posture it has adopted before. Second, it embeds an explicit segment-level revenue target for hyperscaler AI at a moment when the corresponding order target was withdrawn, so the visible forward metric and the leading indicator have swapped places. Third, it guides gross margin down and operating margin up simultaneously, which is only internally consistent if the operating-expense ratio keeps falling, and the Q1 band says it will.

Implied quarter-over-quarter ramp: the Q1 midpoint of $18.1B is 24.9% of the full-year midpoint of $72.8B, which the CFO said matches Cisco's three-year average seasonality. That leaves $54.7B across Q2 to Q4 against $48,442M in the same quarters of fiscal 2026, or 12.9% growth. The CFO put the implied figure at "more like 13%" and attributed the step down from Q1's 21% to comparisons: fiscal 2026's first half grew in single digits while its second half grew in the mid-to-high teens.

Street at: pre-print consensus sat at $68.69B of fiscal 2027 revenue and $4.64 of non-GAAP EPS. The guide midpoints of $72.8B and $5.08 are 6.0% and 9.5% above those. On the quarter, consensus was $16.8B and $1.16 for Q1 fiscal 2027 against a guide of $18.0B to $18.2B and $1.32 to $1.34. This is the widest gap between an initial Cisco annual guide and the sell-side model in the six quarters we have covered, and it is why the stock traded higher on the release before the call.

Guidance style: aggressive on revenue and conservative on the two lines that matter most to the bear case. The AI revenue target of $7.5B is set against $5.3B of undelivered fiscal 2026 orders plus a fiscal 2027 order book management expects to be meaningfully larger, and the CFO called it "prudent" without being asked. Gross margin was not guided for the year at all, only characterised. That combination, an aggressive top line and an unquantified margin, is the mirror image of May, when every growth figure moved up and the margin band was held. Management has become willing to promise revenue and unwilling to promise cost, which is a consistent posture across two quarters and should be read as information rather than as evasion.

Analyst Q&A Highlights

Nine questioners were taken across a compact session. The dominant line of questioning was not whether the quarter was good but whether the fiscal 2027 guide is real, and it arrived from four different directions: the durability of the ex-AI growth, the conservatism of the AI revenue target, the shape of the implied second-half deceleration, and whether the gross-margin guide is prudence or a structural admission. Management answered all four with arithmetic. It answered none of them with a gross-margin range.

The Durability of Growth That Is Not AI

The opening question did the decomposition before management could: at the guided midpoint, backing out hyperscaler AI still leaves roughly 10% growth against a stated long-term model of 4% to 6%, and the question was what produces that and whether it is a peak. The answer enumerated four separate demand pools rather than defending a single one, and it introduced the security-obsolescence driver that recurred through the rest of the call.

Q: "Chuck, your fiscal 27 guide at the midpoint is in 15% revenue growth. Even if I take the AI revenues out of the equation, it is implying double-digit, 10% growth x AI. Versus your long term model that is at 4% to 6%. Can you just touch on what is driving the XAI growth? And how do you think about the durability of this growth, given your comments around the start of a networking super cycle?"
— Amit Daryanani, Evercore ISI

A: "You have the enterprises that we have talked about for over a year now where they are modernized their networks to be ready for AI. Now we have this impending Mythos effect that they are all looking forward to. Downstream, we have quantum that they need to prepare for. And that those are all drivers of what we have been talking about relative to just an overarching campus refresh where, for the first time, ever, we have our campus networking, our well, switching, routing, as well as wireless, and everything all of those are going through a refresh at the same time."
— Chuck Robbins, Chair and CEO

Assessment: The answer is a list, and a list is the right answer to a peak-cycle question because a single-driver cycle peaks and a four-driver one staggers. What the answer conspicuously omitted is the fifth driver, which management itself quantified twenty minutes later: 4 to 5 points of the growth is price. The ex-AI 10% is closer to 5% on units, and no questioner connected the two disclosures on the call.

Whether the Full-Year Guide Understates the Order Rate

A second line of questioning observed that networking orders have run at 50%, then 40%, while the guide implies networking revenue growing well below that, and asked whether the conservatism is deliberate. The answer did not claim conservatism. It gave two mechanical reasons for the implied deceleration and left the question of whether the guide is beatable unanswered.

Q: "And then I wanted to also ask just kind of another approach here. To the revenue guidance for the year. What is your networking orders are 20%, and plus 50, plus 40 the last couple of quarters. Obviously, that would mean revenue in networking is going to grow slower than what we have seen in orders. In your guidance. Is this a conservative approach that you have Why networking revenue kind of meet the order patterns in the middle? As we go throughout the year."
— Ben Reitzes, Melius Research

A: "I think though, as you get into sort of Q2 through Q4, the implied growth is more like 13%. And it is really the comps. As you as you look at last year, the first half, being in, you know, top line growth of kind of single digit growth and as you got into the second half being sort of mid to high teens growth, the comps are gonna get much tougher there. Also, the linearity, if you just look at Q1 revenue guide versus the full year, is it about 25% of the year, and our 3 year average is Q1 to be about 25%."
— Mark Patterson, CFO

Assessment: A comps-and-seasonality answer is the least committal available and it is also, in this case, the correct one: 13% growth against a second half that grew mid-to-high teens is not conservatism, it is arithmetic. The disclosure worth having was the seasonality anchor, because 25% of the year in Q1 against a three-year average of 25% means the full-year guide is not back-end loaded. A guide that does not require a second-half acceleration is a materially safer guide than one that does, and this is the datapoint that should have reassured the tape.

Whether Security's Fourteen Percent Is Repeatable

Security revenue growth jumped from flat to 14% in one quarter and the question asked directly whether that was the Splunk business-model headwind anniversarying or something more durable. Management volunteered the deflating half of the answer before being pressed, then separately claimed the exit-rate commitment made three quarters ago as delivered.

Q: "Nice to see the pickup in the security business, 14% year-over-year, a pretty meaningful improvement in growth. Just wanted to get a sense there. Is that anniversarying some of the headwinds from a business model perspective on Splunk? Is that a lot of technical debt being replaced?"
— Meta Marshall, Morgan Stanley

A: "I think we talked a lot about our belief that the organic Cisco solutions would begin to improve as we got through last year and then into this year. And what I would say is I think our comment was that they would be approaching double digit growth as we exited the year, and they were high single digits, so they did exactly that."
— Chuck Robbins, Chair and CEO

Assessment: The exit-rate promise has been carried in this thesis for three quarters and this is the first time it has been graded, by the company, as met. High single digits against "approaching double digits" is a fair claim on the organic portfolio, and the firewall order growth above 30% for a second consecutive quarter is the leading indicator behind it. The pre-emptive qualification of the reported 14% is the more revealing behaviour: management gave away the flattering half of its own headline unprompted, which is the opposite of how this segment was discussed a year ago.

The Shape of the AI Revenue Guide and the Inventory Build

A question paired the implied second-half deceleration with the balance sheet, asking both how the $7.5B AI revenue target distributes across the year and what Cisco has done on purchase commitments to support it. The revenue-shape answer is the more useful one, because it removes a back-end-loaded interpretation that would have made the annual number harder to underwrite.

Q: "So I was wondering if you could help us bridge the $7.5 billion of revenue that you expect from AI, the linearity of what you are assuming through the course of fiscal 27 And then as a as a quick follow-up, just curious on the supply side. I see the inventory continues to grow. I guess, given the given the dynamics we are seeing in the component availability environment, what has Cisco done in terms of just continually incrementally adding purchase obligations or expanding the supply chain ecosystem to support the demand?"
— Aaron Rakers, Wells Fargo

A: "On the $7.5 billion revenue for AI hyperscale, you should sort of expect a good, you know, trend line in terms of gradual, increases as we move from Q1 through Q4. And then the last piece on the components, I think the, you know, the big thing is we are really leaning into our financial strength. We feel like we have got a world class supply chain team, and they have been voted as such actually for a number of years. And we really do not have any significant lead time issues."
— Mark Patterson, CFO

Assessment: The linearity answer is a genuine de-risking of the guide and it went uncommented. A gradual quarterly ramp to $7.5B means no single quarter has to carry an implausible number, which is the exact criticism we levelled at the fiscal 2026 order target in May when it required a 95% sequential jump. The inventory half of the question was answered with a description of supply-chain capability rather than with a figure, and the specific number asked for, incremental purchase obligations, was not supplied.

Pricing Mechanics and Customer Receptivity

The pricing question asked what the increases contributed to non-AI networking and how customers are responding. The quantified half of the answer became the most consequential disclosure of the call. The half quoted below is the qualitative half, and it is the part that bears on whether the pricing is durable or defensive.

Q: "On campus, obviously, you have implemented a series of price increases going back a ways now. Can you talk to what was the impact of price increases on the non AI part of the business within networking And what are the conversations sound like from receptivity perspective from customers where they sit in the, I guess, market today? Are they accepting the price increases? Are they just ordering ahead?"
— David Vogt, UBS

A: "Yet, I would just tell you that price increases are certainly a last resort for us. I mean, we are doing everything we can to secure the right supply at the right prices and build up inventory where needed and advance purchase commitments and we are also, you know, doing a lot internally. G2 and the teams have got over 30 different programs that are all around the most efficient utilization of memory in our products. So that certainly helps us as well. Wi-Fi 7 is a big example there. Where the improvements that the team have done within about a 90-day period, actually reduced the memory utilization in Wi-Fi 7 by 50%."
— Mark Patterson, CFO

Assessment: Calling price increases a last resort is management telling you the pricing is defensive and temporary, which is exactly how it should be modelled. The 50% memory reduction in Wi-Fi 7 inside 90 days is the more important disclosure, because engineering the memory content out of the product is the only route back to the old gross-margin structure that does not depend on component prices falling. If those 30 programmes work, the price increases roll back and revenue growth takes a 4-to-5-point hit in fiscal 2028 while gross margin recovers. That trade is not in anyone's model, including ours.

Whether the AI Revenue Target Is Set Too Low

The most direct challenge of the call worked the gap between a $9.3B order book and a $7.5B revenue target and called the target "very, very conservative." The response corrected the questioner's arithmetic on the prior year and defended the number on order-timing physics rather than on demand, using a word management chose deliberately.

Q: "I know you do not give backlog numbers, but I guess what I am trying to understand is understand you took the numbers from $6 billion to $7.5 billion for AI revenue. But, again, that just strikes me as very, very conservative. Is there something about the timing of the order deliveries? Are orders being shipped out north of 12 months? Or is it supply chain related?"
— David Vogt, UBS

A: "So, I think on the AI orders, just for clarity, we did about $4 billion in revenue and $9 billion in orders for FY '26. So, obviously, you are entering with some good backlog as we go into 2027, but these are, you know, these are nonlinear orders that are mass in scale and are usually placed well ahead of time. So I think the $7.5 billion is a good prudent guide for the year."
— Mark Patterson, CFO

Assessment: Management did not dispute the characterisation. "Prudent" is an acknowledgement, not a defence, and it sits alongside a fiscal 2026 order forecast that was raised from approximately $5B to approximately $9B and then delivered $9.3B. The honest reading is that Cisco has chosen to guide the AI line where a beat is the base case, which is the same choice it made a year ago and which produced two upward revisions. This is the largest identifiable source of upside in the fiscal 2027 number, and it is the reason we model above the guided range.

The Hyperscaler Order Deceleration Nobody Else Raised

One questioner did the arithmetic on the customer-market disclosure and found that service provider and cloud order growth slowed to 95% from 105%, even as telco within that group accelerated from 9% to more than 30%, which implies the hyperscaler component decelerated more than the headline shows. The answer attributed it to comparisons and to the changed relative size of the two businesses.

Q: "But if I look at the combined segment growth last quarter, telcos and server sorry, telcos and cloud was up 105%. And this quarter was up 95%, which means that there was some slowdown of the hyper scaler orders this quarter The question is, is there anything notable that you are seeing in terms of the demand environment Or is it just about nonlinearity of orders?"
— Thomas Zilberman (for Tal Liani), BofA Securities

A: "I think going from 105 to 95, I think that is probably some sort of comparison issue from a year ago would be my guess because we had we had 4, I think, the 4 big hyperscalers all grew triple digits for us, I think, during the quarter. So we did not see and, again, at the beginning of the quarter, we said we would do over 4 billion and we saw we did over 4 billion which was obviously the largest quarter of the year for us."
— Chuck Robbins, Chair and CEO

Assessment: The best question of the call and the weakest answer, hedged three times inside two sentences. The substantive rebuttal is the one management gave last: $4B of orders in a single quarter is the answer to whether hyperscaler demand decelerated, and it is a better answer than the growth-rate arithmetic. The detail that went unremarked is that the count of top hyperscalers growing in triple digits went from five last quarter to four this quarter. Nobody asked, and with the annual order target now withdrawn, the customer-count disclosure is one of the few concentration signals left.

Where the Budget for This Actually Comes From

The final exchange returned to the question underneath the whole guide: enterprises are spending across the entire IT stack simultaneously, in an inflationary component environment, and the sustainability of that depends on whether budgets expand or merely reallocate. The answer was an anecdote, and it was more informative than a framework would have been.

Q: "Chuck, I am interested in your thoughts on you know, this incremental demand that you are seeing urgency around some of the newer variables such as Mythos, now where do you see the customer budget coming from? What are they doing to fund this stuff? You know, how are you capturing more of that opportunity?"
— Ben Bohlen, Cleveland Research

A: "And they said, well, I just I want to make sure that we are getting this equipment in here. We gotta get this done, and you know, I think the whole discussion of where the budget comes from is it is not even an issue. I am sure they will figure it out behind the scenes, but I think they start from the premise of we do not have a choice. Then they figure out how to fund it. And I think that is just a much different place than we have been."
— Chuck Robbins, Chair and CEO

Assessment: "We do not have a choice" is the entire investment case for the refresh leg of this thesis, and it is a claim about buyer psychology rather than about budget capacity, which makes it both powerful and unverifiable. The falsifiable part is the pipeline, and management stated twice that none of this is material in the numbers yet. That is the correct disclosure and it is also the reason we treat the refresh as optionality on top of the guide rather than as support for it.

Lead Times and the Direct Silicon Relationship

A follow-up asked where lead times sit for Silicon One products versus the rest of the portfolio and what the direct TSMC relationship does to them. The lead-time half was answered generically. The structural half was the disclosure.

Q: "The other question would be, if I look at you said really no issues on lead times. Could you discuss where lead times are for your non Silicon One product or versus Silicon One product, and then just spend a little bit more talking about that direct line of comp communication with TSMC and how that influences lead times."
— Ben Bohlen, Cleveland Research

A: "On the TSMC, specific question, I think that, you know, the biggest thing is we are we are not waiting on any sort of allocation. And commits and decommits, if you will, from any merchant silicon providers. We are we are dealing directly with TSMC know, G2 and Chuck have been directly engaged and I know our spending time with them."
— Mark Patterson, CFO

Assessment: The question asked for a comparison between two product families and got a company-level answer, so the specific disclosure requested was not provided. What was provided matters more: a supplier position that removes third-party allocation risk from the constraint that is limiting Cisco's competitors. In a quarter when the entire sector is discussing memory and substrate scarcity, "no significant lead time issues" and "adequate supply to meet upside to the guide" is a competitive statement, not a housekeeping one.

What They're NOT Saying

1. Any fiscal 2027 gross margin range: Cisco guided Q1 to 65% to 66% and gave the full year only the word "slight," while guiding revenue, operating margin, EPS, tax rate, segment growth and AI revenue for the full year. Gross margin is the one line in the model that got a quarter of visibility instead of four, and it is the line the stock fell on.
2. What happens when the price increases lap: management disclosed that 4 to 5 points of fiscal 2027 revenue growth is price, weighted to the first half, and that the increases are "a last resort" tied to memory cost. Nobody asked what fiscal 2028 revenue growth looks like when that reverses, and management did not volunteer it. The same disclosure that supports the fiscal 2027 guide is a headwind to the year after.
3. The fiscal 2027 AI order number: replaced with "meaningfully higher." Cisco published an annual order target for two consecutive years, raised it twice inside fiscal 2026, and retired it in the quarter it was exceeded. The stated reason is that the metric had done its job of demonstrating the opportunity. The effect is that the leading indicator for the largest growth driver is now qualitative.
4. Services ARR: disclosed every quarter through fiscal 2026 while it was negative, at minus 1% for three consecutive quarters, and absent from this call. Total ARR and product ARR were both given. The one line that had been going backwards is the one that stopped being reported.
5. Splunk ARR and any Splunk revenue figure, for a third consecutive quarter: management nominated Splunk ARR three quarters ago as the honest measure of that business beneath the accounting transition. It has not appeared since. The substitute disclosure is again a new-logo count, now framed as having exceeded the 1,000 target.
6. The dollar or percentage concentration of the $9.3B order book: four of the top hyperscalers each grew in triple digits, down from five named last quarter, and no split was offered for the third consecutive quarter. With the annual order target withdrawn, this count is now one of the only concentration disclosures remaining, and it moved in the wrong direction without comment.
7. The $869M of investment gains: other income of $822M carried GAAP EPS to $0.97 and GAAP net income growth to 51%, against non-GAAP growth of 23%. The reconciliation removes $0.22 per share of investment gains this quarter against $0.03 a year ago. It was not mentioned in prepared remarks and drew no question, in a quarter where the GAAP headline was widely reported.
8. The restructuring headcount, for a second consecutive quarter: $511M of charges landed in Q4 and $693M across fiscal 2026 under a plan announced in May at up to $1B, with $0.11 per share of further charges guided into fiscal 2027. The company has still never stated how many roles are involved in a plan it describes as a realignment rather than a cost reduction.
9. Inventory purchase commitments: the figure stood at $16,033M a quarter ago, up 153% year over year, and inventories have since risen another 21% to $5,694M. The number is a 10-K disclosure and the fiscal 2026 10-K has not been filed. No commitment figure was given on the call in answer to a question that specifically asked about purchase obligations.
10. Why financing receivables rose $1,801M in a single quarter: Cisco extended more customer financing in Q4 than in any prior quarter of the year, taking the total to $8,332M from $6,527M. Vendor financing against a hyperscaler ramp is a legitimate use of the balance sheet and it is also a form of revenue support. It appeared in the cash flow statement, in no remark and in no question.
11. The sovereign and neocloud pipeline figure: sized at approximately $3B last quarter and not updated. Orders were disclosed at more than $400M in Q4 and more than $1B for the year, which understates the sum of the quarterly figures previously given. The pipeline, which was the forward-looking half of this disclosure, was dropped.
12. Any free cash flow figure: management stated capital return was 99% of free cash flow without giving the denominator, the same construction used in each of the prior three quarters. The statements support $12,767M for fiscal 2026, down 3.9% year over year, in a year revenue grew 11.8%.

Market Reaction

• Pre-print setup: CSCO closed at $123.88 on August 12, up 60.8% year to date against 13.2% for the S&P 500, up 73.6% over the trailing twelve months and up 3.9% over the trailing thirty days. The 52-week closing range entering the print was $66.20 to $130.00, so the stock went in 4.7% below its 52-week closing high.
• Options positioning: the market priced an implied post-earnings move of approximately 8.26%.
• After-hours move: shares rose on the release and reversed during and after the call. Reported extended-hours moves range from roughly 3% higher shortly after the print to nearly 8% higher at the peak, before turning to a decline of more than 6% later in the session. The dispersion reflects thin extended-hours quoting across a reversal rather than a disputed fact.
• Next-day session: the stock gapped down 8.0% to open at $113.93, traded a range of $111.48 to $115.95, and closed at $113.47, down 8.4% or $10.41.
• Volume: 61.6 million shares against a 21.1 million 30-day average, or 2.9 times normal.
• Index comparison: the S&P 500 rose 0.7% on the session.

The sequence is the story. The press release was received as a beat-and-raise and the stock traded higher on it. The reversal began on the call, and it maps to a specific disclosure: the Q1 gross-margin band of 65% to 66%, delivered in prepared remarks, followed within one question by the characterisation of a "slight gross margin headwind as we move through FY 2027" and the redirection to operating margin. Everything else on the call was better than expected. The margin path was worse, and it was the only full-year line management declined to quantify.

Positioning turned a disappointment into a rout. The stock entered at 60.8% year to date against 13.2% for the index, having compounded 73.6% over twelve months, which is a setup that requires the guide to clear the buy side's expectation rather than the published consensus. It cleared the published consensus comfortably, by 6.0% on revenue and 9.5% on earnings. It evidently did not clear the number that had been paid for. The realised move of 8.4% against an options-implied 8.26% says the market got the magnitude right and the direction wrong, which is what happens when a distribution is symmetric around a number nobody outside the company can observe.

The re-rating arithmetic is what makes this actionable. Ninety days ago the stock closed at $115.53 against our fiscal 2027 estimate of $4.70, or 24.6 times. It now closes at $113.47 against a fiscal 2027 estimate of $5.15, or 22.0 times. The price is 1.8% lower and the forward estimate is 9.6% higher, so two and a half turns of multiple came out of the same forward year in a single session. That is the entire content of the upgrade, and it happened without a single operating datapoint deteriorating.

The sell-side response was the tell. In the sample of firms publishing on the reaction day, every one raised its price target, into a range spanning $123 to $165. A universal target increase alongside an 8.4% decline is the signature of a positioning unwind rather than a change in the fundamental view, and positioning unwinds are the setups this desk is paid to buy.

Street Perspective

Debate: Is the Gross-Margin Decline the Cost of Growth or a Permanent Impairment?

Bull view: the bull case being made on the Street is that gross margin is the wrong denominator for a business whose incremental revenue requires almost no incremental operating expense. Q4 proved it arithmetically: 210 basis points of gross-margin decline produced 160 basis points of operating-margin expansion, and fiscal 2027 is guided to a record operating margin on a lower gross margin. If the company earns more operating profit per revenue dollar than it ever has, the mix of where that profit is created is an accounting detail.

Bear view: the bear camp contends that operating leverage is a one-way ratchet that runs out. Cisco is converting a software-adjacent margin structure into a hardware one: product gross margin has fallen 270 basis points in a year, the company will not guide the full-year gross margin at all, and the offset is an expense ratio that cannot fall forever. When the AI mix stops rising, the mix headwind stops shrinking, and what remains is a business earning mid-60s gross margins where it earned high-60s, with the expense lever already pulled.

Our take: the bears are right about the destination and wrong about the timing, and the timing is what the next twelve months pay for. The gross-margin level is permanently reset and management has stopped pretending otherwise, which is why the metric of record changed. But the expense lever has visible runway: the ratio fell 370 basis points this quarter and is guided to fall another 90 in Q1, and the reason is structural rather than discretionary, because hyperscaler revenue genuinely arrives without a sales and support cost to match. We would start worrying about this in fiscal 2029, when the Silicon One rollout either delivers the cost-structure argument management hinted at or does not.

Debate: Is the $7.5 Billion AI Revenue Target Conservative, or Is the Order Book Front-Loaded?

Bull view: the optimistic camp points out that Cisco closed fiscal 2026 with $9.3 billion of hyperscaler orders against approximately $4 billion of revenue, that the difference is roughly $5.3 billion of undelivered backlog before a single fiscal 2027 order, and that management expects fiscal 2027 orders to be meaningfully higher still. Against that, a $7.5 billion revenue target which the CFO himself called "prudent" and declined to defend against a charge of conservatism looks like the same posture that produced two upward revisions inside fiscal 2026.

Bear view: the skeptics answer that a large order-to-revenue gap is what a lumpy, capacity-constrained, multi-year deployment looks like from the inside, and that it says nothing about the pace of conversion. Hyperscaler orders are placed far ahead of delivery and some of that $5.3 billion is fiscal 2028 revenue by design. The customer set narrowed from five hyperscalers growing in triple digits to four, service provider and cloud order growth decelerated from 105% to 95%, and the company has just removed the annual order disclosure that would let anyone verify the trend independently.

Our take: we side with the bulls on the number and with the bears on the ability to check it. The evidence for the forecasting cadence is now a track record rather than a promise: the $9 billion order forecast set in May required a 95% sequential jump that we flagged as the largest disconfirmable claim in that quarter, and it overshot. The scale-across wins we said contributed nothing to it are now producing orders. What genuinely worsened is observability of the line, and that is a real cost. It means the fiscal 2027 AI number gets tested annually rather than quarterly, which raises the variance of the stock around each print without changing the expected value.

Debate: How Much of the Fiscal 2027 Growth Survives the Price Lap?

Bull view: a growing consensus view is that the pricing question is a distraction. The price increases are single-digit, apply only to hardware, are concentrated in high-memory products, and exist to defend gross profit dollars against a component-cost shock that is hitting every hardware vendor far harder than it is hitting Cisco, whose bill of materials is 15% to 20% memory against roughly two-thirds for server vendors. Cisco is taking share and shipping record volume; the composition of the revenue dollar between price and units is a modelling detail that nets out at the gross profit line.

Bear view: the bear camp contends that 4 to 5 points of a 15% growth guide is a third of the acceleration and that it is borrowed from fiscal 2028. Strip price and hyperscaler AI from the fiscal 2027 guide and the rest of Cisco grows about 5%, which is the middle of the same 4% to 6% long-term model the company has had for a decade. The supercycle, on this reading, is one very large customer cohort plus an inflation pass-through, and the market is paying a growth multiple for a mix effect.

Our take: the bears have the decomposition exactly right and draw the wrong conclusion from it. Roughly 5 points of volume growth in the non-AI business is not a disappointment; it is the long-term model delivering while a $7.5 billion AI business compounds on top of it. The important consequence of the price analysis is not for fiscal 2027 but for fiscal 2028, and it is benign: price-driven revenue arrives with no incremental gross profit, so when it reverses it takes revenue growth away without taking earnings with it. That is the single most misunderstood line in this guide. It means fiscal 2028 will look like a sharp revenue deceleration and a much milder earnings one, and investors who sell the revenue print will be selling the wrong number.

Model Update Needed

The fiscal 2027 EPS bridge. Revenue of $74.0B at a 35.0% non-GAAP operating margin gives operating income of $25,900M. Non-GAAP interest and other income, net, was negative $757M in fiscal 2026 and we carry negative $800M on a slightly larger debt balance, giving pre-tax income of $25,100M. At the guided 18.5% non-GAAP tax rate that is $20,457M of net income, and on 3,965M diluted shares it is $5.16, which we round to $5.15. We model above the guided range for one reason: the initial fiscal 2026 guide issued twelve months ago was $4.00 to $4.06 and the year printed $4.33, 6.7% above the high end. Our $5.15 is 0.8% above the fiscal 2027 high end, so we are taking roughly an eighth of the historical overshoot. We take that little because the AI line is now an explicit revenue guide rather than an emergent surprise, and because the gross-margin path is a genuine headwind rather than an unknown.

Valuation impact. At the $113.47 close CSCO trades at 26.2 times trailing fiscal 2026 non-GAAP EPS, 22.3 times the fiscal 2027 guide midpoint, and 22.0 times our $5.15 estimate. In May we said we would apply 22 to 25 times fiscal 2027 earnings, and we set that range explicitly against management's then-only forward framework, which produced roughly 8% revenue growth and 9% EPS growth. The company has now guided 15% and 17%. We raise the range one turn, to 23 to 26 times, and only one turn, because roughly a third of the revenue acceleration is price rather than units and because the gross-margin line is guided lower with no annual range attached. Applied to $5.15, that gives a 12-month fair value of $118 to $134 (prior: $105 to $118), a midpoint of $126, which is 11.0% above the close. Adding the 1.5% yield on the $0.42 quarterly dividend gives an expected total return of roughly 12.5%. At 20 times, which is the multiple the stock carried before this cycle began, fair value is $103, or 9.2% below the close. We are being paid roughly 12 for 9.

Thesis Scorecard Post-Earnings

Overall: the thesis strengthened and the price weakened, which is the inverse of every quarter we have written on this name. All four bull pillars are confirmed, and Bull #2 gained a demand driver that did not exist ninety days ago and is explicitly excluded from the guide. Three of the seven carried bear points were challenged: valuation resolved on an 8.4% decline against a 9.6% estimate increase, the recurring franchise reversed on every published line, and the fiscal 2027 operating-leverage question we raised in May was answered by a record margin guide. One deteriorated: gross margin was contained in the quarter and re-opened by a forward path management would not quantify. Three are unchanged, with concentration now materially harder to monitor. Two new bear points are added, on the borrowed nature of the price-driven revenue and on the steady withdrawal of the metrics that make this thesis testable.

Action: Upgrading to Outperform from Hold. We want to be precise about the basis, because a rating that moves both ways inside two quarters deserves an explanation. In May we downgraded on valuation alone and published two conditions for returning: hyperscaler AI orders landing at or above the roughly $3.7 billion the fiscal 2026 forecast required, or a pullback with the recurring metrics stabilising. The first condition was met at $4.0 billion. The second was met on the metrics and half met on the price. We also published an Underperform trigger, gross margin at the low end of the guide with total ARR growth at or below 2%, and neither leg fired: margin came in above the midpoint and ARR growth accelerated to 3%. The fundamental case is a better business than we underwrote in May by every measure except the forward gross-margin path, and the stock is 1.8% cheaper against an estimate that is 9.6% higher. Fair value moves to $118 to $134 from $105 to $118, a midpoint of $126 against a $113.47 close, for an expected total return of roughly 12.5% including the dividend. We would move back to Hold if fiscal 2027 gross margin tracks below 65% for two consecutive quarters, or if the price recovers above $130 without a raise to the $7.5 billion AI revenue target. We would move to Underperform if hyperscaler AI revenue tracks below the $7.5 billion guide at the halfway mark, because the entire re-rating case rests on that line converting from orders into revenue on schedule.