AI Orders, Campus Refresh, and a $1B Full-Year Raise Inflect the Story — Even With DRAM Pressure on Product Margins

Results vs. Consensus

Quality of Beat/Miss

• Revenue: Beat is fully organic and concentrated in Networking, which over-delivered relative to Street by ~$300M on its own. Networking product revenue growth was up 15%, with service provider routing and AI infrastructure cited as primary drivers. Security and Collaboration both missed but for explainable reasons — see segment commentary on the Splunk on-prem-to-cloud mix shift.
• Margins: Headline non-GAAP gross margin was in line with the guide midpoint but down 120 bps YoY, with the entire compression in product. Non-GAAP product gross margin was 67.2%, down 170 basis points, driven by negative impacts from mix and pricing, partially offset by productivity improvements. Non-GAAP services gross margin was 70.7%, up 40 basis points. CFO Patterson cited tighter memory/PCB/optics supply and meaningful DRAM price increases — these are real, but baked into the updated guide.
• EPS: Non-GAAP EPS grew faster than revenue (+10% vs +8%) on operating-margin expansion. Tax rate of 19% non-GAAP was unchanged; share count modestly lower on the $2B Q1 buyback. EPS quality is high — operational, not financial-engineering.
• Cash flow: The –12% operating cash flow decline is the one notable yellow flag, but the explanation is constructive: operating cash flow decreased by 12% due to investments to meet growing customer demand for AI infrastructure. Inventory plus advance purchase commitments are up ~$3B (+38%) YoY, supporting the doubling-of-AI-orders commitment.

Segment Performance

Networking

Networking is doing all the heavy lifting and then some. Within product revenue, all technologies within campus networking (switching, routing, wireless and IoT) saw accelerated order growth in Q1, and the segment beat Street by $300M against an already-elevated bar. The order book is the more interesting data point: networking product orders grew at a high-teens pace, the fifth straight quarter of double-digit order growth, with hyperscaler AI infrastructure, enterprise routing, campus switching, and industrial IoT all contributing.

"AI infrastructure orders taken from hyperscalers in Q1 total $1.3 billion, balanced between Silicon One systems and optics, marking a significant acceleration in growth… We expect to recognize roughly $3 billion in AI infrastructure revenue from hyperscalers in fiscal year 2026." — Chuck Robbins, Chair & CEO

Two structural points stood out. First, AI demand visibility is broadening beyond hyperscalers to sovereign, neocloud, and enterprise customers, with a pipeline in excess of $2.0 billion for high-performance networking — and Robbins explicitly confirmed this is incremental to the 2x hyperscaler orders target. Second, the pluggable-optics business has reached saturation in the hyperscaler buyer base: every major hyperscaler is now a customer. That makes Acacia a real franchise rather than a niche.

Assessment: Networking is the entire investment case right now, and the order book says the growth is durable for at least the next several quarters. The 51.2 Tb/s P200-based 8223 router targets scale-across data-center interconnect — a category Cisco hasn't historically owned — and gives the segment a third leg next to hyperscaler scale-out and campus refresh.

Security

The –2% revenue print and ~$180M miss against Street look ugly on the headline, but the call did a credible job of decomposing it. The story has two parts. First, the security segment was down 2% due to declines in prior generation products and a mix shift toward cloud subscriptions in the Splunk business, partially offset by growth in firewalls, Duo, and SASE. Second — and this is the more important point — the Splunk mix shift is an accounting timing issue, not a demand issue.

"What you are seeing in, I think, a better measure of the health of the business of Splunk is really to look at ARR and RPO. Both of those grew in the double digits for the quarter. Again, while we're disappointed with the timing a little bit in the quarter, overall, it's actually a really good thing for us." — Mark Patterson, CFO

Robbins also re-anchored the mid-teens long-term security revenue guide — a number that, if it holds, implies meaningful acceleration over the next four quarters as the Splunk mix shift normalizes and prior-gen drag continues to shrink. He notably added that security does not need to materially improve to hit the Q2 or FY guide, meaning the bar for the segment in the back half is low.

Assessment: Security is in a transition quarter with messy optics but constructive underlying trends. Nearly 3,000 customers have purchased a new product since launch, mid-teens demand growth on next-gen firewalls, and double-digit Splunk ARR/RPO are the data points that matter. The Q2 setup is helped by both an easier comparable and a small absolute contribution to the FY guide.

Collaboration

Down 3%, missed Street by a small amount, and barely got airtime on the call. Devices and Webex remain the soft spots. At ~7% of revenue, the segment is getting smaller as a share of the mix and is increasingly a "manage in cash mode" line item.

Assessment: Not a thesis driver in either direction. Watch for any signs that Webex AI features start to differentiate against Zoom and Teams, but don't underwrite it.

Key KPIs

Key Topics & Management Commentary

Overall Management Tone: Management was the most forward-leaning it has sounded in several quarters — the prepared remarks and Q&A consistently framed AI, campus refresh, and sovereign/neocloud as concurrent rather than sequential opportunities, and the guide raise was delivered without the usual conservatism hedges. Pushback was narrow and concentrated on margin (DRAM/PCB pricing) and on whether ex-hyperscaler growth is good enough; on both, the response was framework-level rather than defensive.

AI Infrastructure: From Narrative to Number

For three quarters Cisco has been telling the AI story; this quarter put a hard dollar figure on it. AI Infrastructure orders taken from hyperscaler customers totaled $1.3 billion, reflecting a significant acceleration in growth, and management committed to recognizing roughly $3B of AI infrastructure revenue from hyperscalers in FY26. Robbins was unusually specific on the order trajectory.

"What we expect from an orders perspective this year is that we are expecting at least two times the orders that we received in fiscal year 2025 from that same set of customers." — Chuck Robbins, Chair & CEO

Two further details mattered. First, the enterprise pipeline for Neocloud and sovereign cloud exceeds $2 billion for the next three quarters, with $200 million booked in Q1 — and this is incremental to the 2x hyperscaler number. Second, the product mix is broadening: Silicon One systems, pluggable optics (now sold to every major hyperscaler), and the new 51.2 Tb/s P200-based 8223 router for scale-across DCI. Four major hyperscalers grew triple-digit in the quarter, and each contributed at least one "meaningful use-case win" — i.e., it isn't one customer carrying the segment.

Assessment: The AI exposure is real, accelerating, and broadening across products and customers. Importantly, sovereign/neocloud upside is explicitly carved out of the FY guide, which de-risks the back half.

Campus Refresh: The Quietly Underwritten Second Leg

The under-appreciated piece of the print is the campus refresh framing. Robbins linked the strength to multiple drivers: Cat 4K/6K end-of-support, AI-driven enterprise network modernization, and integrated security-in-network as a competitive moat against rivals partnering with third-party security vendors.

"When we launched the Catalyst 9K in 2017… that transition just kept going for five, six, seven years. That is just typically what we see when these things move. The fact that they are ramping faster than what we have seen in the past would indicate there is a lot of interest in these portfolios." — Chuck Robbins, Chair & CEO

The 9K-vs-pre-9K install-base point is the most important nuance: management argues the size of the still-untouched pre-Cat 9K install base is larger than consensus assumes, and that the 2023 backlog flush did not actually clear the install-base overhang. If true, this is a multi-year tailwind, not a 2026 phenomenon.

Assessment: Campus is the bridge from the "AI hyperscaler" tailwind (concentrated, cyclical-feeling) to a durable enterprise spend cycle. If next-gen products keep ramping faster than prior cycles for another two quarters, this becomes a real second leg of the bull case.

Splunk: Cloud Mix Shift Is a Feature, Not a Bug

The mechanics: in the prior quarter the on-prem-vs-cloud Splunk mix was roughly 50/50; in Q1 FY26 it dropped to roughly one-third on-prem. Revenue for cloud subscriptions is recognized ratably, whereas product revenue for on-prem deals is recognized on delivery, which created the visible revenue drag.

"While we're disappointed with the timing a little bit in the quarter, overall, it's actually a really good thing for us… When we look at the cloud offers, they're stickier than the on-prem offers. Customers are actually able to adopt the technology faster, adopt features faster." — Mark Patterson, CFO

Double-digit ARR and double-digit product RPO growth at Splunk corroborate the "demand is fine, recognition is just deferred" framing.

Assessment: The risk is whether the mix shift normalizes at the assumed pace; management estimates roughly four quarters for year-over-year comparisons to become apples-to-apples. If the cloud mix keeps climbing past ~67%, the in-period revenue drag could persist longer than guided. Watch ARR vs. revenue divergence in Q2 and Q3 closely.

Input-Cost Pressure: DRAM, PCB, and Optics

The clearest negative on the call. Management was direct that supply is tightening and that prices in DRAM, PCB, and optics are moving up materially. The mitigant is that this is already in the updated guide.

"I would just say across memory as well as PCB and optics, we've noticed a bit of a tightening of supply. On the memory side, we've seen what you all have all seen as well, and that's pretty significant price increases as well. Both of those, in terms of the supply as well as the pricing, though, are both included and considered in our updated guide for the Q2 as well as the year." — Mark Patterson, CFO

Inventory plus advance purchase commitments are up ~$3B YoY (+38%), with $1B of that added in Q1 alone — management is pre-buying capacity to support the doubling of hyperscaler orders. The Q2 non-GAAP gross-margin guide of 67.5–68.5% implicitly bakes in a continued ~120 bps YoY product-margin drag.

Assessment: The margin compression is real but ring-fenced and being addressed via supply commitments and pricing. The right way to read it is that revenue growth is being paid for with margin in the short term — an acceptable trade if the order acceleration sustains.

Capital Return and Free Cash Flow Mechanics

The company returned $3.6B in Q1 (125% of free cash flow) — $1.6B dividend + $2.0B buybacks at an average price of $68.28. The pay-out-above-100%-of-FCF posture combined with the –12% operating cash flow makes the cash dynamics worth watching: ending cash and investments dropped from $16.1B at FYE25 to $15.7B at Q1-end, and the share repurchase authorization remaining is $12.2B.

Assessment: Comfortable for now, but not infinitely sustainable if operating cash flow continues to compress while buybacks stay at the 125%-of-FCF pace. Watch Q2 FCF disclosure carefully — if it normalizes (working capital reverses out), the picture is fine; if not, expect the buyback pace to moderate in H2.

Guidance & Outlook

The guide raise is meaningful and accomplishes three things. First, it sets a Q2 revenue bar ~$400M above the LSEG midpoint, which the order book makes look achievable rather than aspirational. Second, the full-year guide is raised on both revenue (by ~$1B at the midpoint) and non-GAAP EPS (by ~$0.06). Third, the guide explicitly excludes sovereign and neocloud build-out from material FY26 contribution.

"As you all know, Samik, there's export licenses that have to be attained in many of these cases. We're still working through that. We expect some of that stuff will really start flowing in probably the second half of our fiscal year. To Mark's point, it's not a material part of the guide this year." — Chuck Robbins, Chair & CEO

Implied H2 ramp: Q1 came in at $14.88B; Q2 guide midpoint is $15.1B. FY26 midpoint is $60.6B. That implies H2 revenue of ~$30.6B, or ~$15.3B average per quarter. Achievable given the order book — H1 will exit at a $30.0B run rate, so H2 only needs to step up by ~1% on average — but the comps get harder in Q3/Q4 (Cisco printed +11% / +8% revenue growth in Q3/Q4 FY25).

Street at: FY26 revenue consensus pre-print sat at $59.7B and EPS at $4.04; the new guide brackets the Street about $400M and $0.06 above on the midpoints. Q2 revenue Street was $14.6B — a clear and material upward revision.

Guidance style: Less conservative than Cisco's historical pattern. The prepared remarks called Q1 results "above the high end of our guidance" twice; the new guide does not appear to embed the same level of cushion. That increases beat-and-raise risk-of-disappointment in Q2 if the AI order pace slows or DRAM costs accelerate further than modeled.

Analyst Q&A Highlights

AI Hyperscaler Orders & the "$3B" / "2x" Framework

• Aaron Rakers, Wells Fargo: Asked Robbins to clarify the $3B AI infrastructure number and how scale-across opportunities are evolving in the hyperscaler vertical. Robbins clarified $3B is FY26 hyperscaler AI revenue, while $1.3B was Q1 new orders, and reaffirmed the 2x-FY25-orders trajectory for the full year. Confirmed pluggable optics are now sold to every major hyperscaler.
  Assessment: The most important exchange of the call. Disambiguates a number that had been confusing on the buy side ($3B revenue ≠ $3B orders) and reaffirms the order ramp at 2x.
• Meta Marshall, Morgan Stanley: Asked whether AI strength is coming from scale-across or deepening existing engagements, and probed DRAM pricing impact on gross margins. Robbins said most of Q1 was deepening of existing use cases — scale-across is emerging but not yet meaningful in the order book. Patterson confirmed DRAM/PCB/optics tightening and price increases, already baked into the updated guide.
  Assessment: Important — Q1 AI strength is from existing customer wallet share, not new use cases. Scale-across is upside that hasn't yet shown up. Margin is the real watch item.
• Samik Chatterjee, JPMorgan: Asked about optical opportunity inside vs. outside the data center, and clarified whether sovereign customers are in the 2x hyperscaler order number. Robbins confirmed Cisco is participating in both intra-DC and DCI optical; explicitly carved sovereign/neocloud/enterprise out of the 2x number — that $2B+ pipeline is incremental upside.
  Assessment: Reinforces the guide-conservatism point. Sovereign upside is real but timing depends on export licenses — second-half FY26 phenomenon.

Ex-AI Growth & the "3.6% Core" Pushback

• Tal Liani, Bank of America: Pressed on the math that, ex-$1B of FY25 AI revenue and ex-$3B FY26 AI, the rest of the business is only growing ~3.6%. Robbins offered a new disclosure: ex-hyperscaler orders, the rest of the business grew 9% in Q1. Patterson conceded the math is right but flagged tougher H2 comparable quarters.
  Assessment: The most important "tough" question of the call. The 9% ex-hyperscaler order disclosure is constructive — it argues the core enterprise business is healthier than the headline 8% revenue suggests. Watch this metric next quarter.
• Karl Ackerman, BNP Paribas: Asked how much of the ~$1B FY26 guide raise is AI vs. campus refresh, and whether Cisco has secured enough supply to fulfill 2x hyperscaler orders. Robbins said it's a mix of both. Patterson disclosed that inventory plus advance purchase commitments are up nearly $1B QoQ and ~$3B (+38%) YoY.
  Assessment: The $3B advance-commitment number is the most concrete piece of evidence that the AI order book is real — Cisco is putting capital at risk to fulfill it.

Campus & Public Sector

• David Vogt, UBS: Asked why next-gen campus products are ramping faster than prior cycles — competitive disruption or end-of-life motivation — and how government shutdown affects the federal piece. Robbins cited Cat 4K/6K end-of-sale, AI preparation, security-in-network integration, and noted competitor Wi-Fi confusion as positives. Federal grew high-single-digits in orders despite the shutdown.
  Assessment: The competitive-disruption answer was hedged ("I can't give you any specifics"), but the structural drivers (end-of-life base, AI prep, integrated security) are credible.
• Simon Leopold, Raymond James: Asked whether the campus install base was actually flushed in 2023 and got an unusually direct answer: management argued the pre-Cat 9K install base flushed in 2023 was "probably overestimated" and that "billions and billions and billions of dollars" of pre-9K install base remains. Public sector strength is broader than US federal; EMEA and APJC grew mid-to-upper teens.
  Assessment: This is the most important campus-related answer of the call. If the 2023 backlog flush did not actually clear the install base, the multi-year refresh framing is much more credible.

Security & Splunk Mix

• James Fish, Piper Sandler: Asked about Silicon One's penetration and the Splunk cloud mix shift's impact on security revenue. Robbins targeted "fully rolled across the entire portfolio" by end of FY29. On Splunk, on-prem mix dropped from ~50% prior quarter to ~33% in Q1, explaining the revenue drag — but ARR and RPO both grew double-digit.
  Assessment: Useful framing. Silicon One is still a multi-year roll-out story (a third leg of margin expansion). Splunk Q&A confirms it's a revenue-recognition timing issue rather than a demand problem.
• Amit Daryanani, Evercore ISI: Asked about AI revenue margins and what normalized security growth looks like once the mix stabilizes. Patterson punted on AI margins ("part of the mix and part of the guide"). Robbins reaffirmed mid-teens long-term security guide and committed to 4-quarter normalization timeline. Critically added: "We do not need security to materially improve from here to hit the guide for Q2 or the year that we have given you."
  Assessment: The "we don't need security to inflect to hit guide" line is the most important risk-reduction comment of the call. It implies security upside is unmodeled.

Adjacencies & Architecture

• Mike Ng, Goldman Sachs: Asked about the G42 / AMD partnership and the channel-partner program changes. Robbins committed to GPU/XPU ecosystem partnerships broadly (not just AMD) and described the partner-program changes as recognition of growth opportunities. Patterson cited simplification, depth-of-expertise rewards, and focus on campus refresh / AI / security / premium services.
  Assessment: AMD answer is constructive — Cisco's connectivity layer is positioned as GPU-agnostic, which is the right strategic position.
• Antoine Chkaiban, New Street Research: Asked about Cisco Unified Edge size opportunity and scale-up participation. Robbins described Unified Edge as a retail/restaurant/healthcare opportunity and confirmed Cisco intends to participate in the scale-up Ethernet market — "you should expect to see something from us over time."
  Assessment: The scale-up confirmation is new and important — it expands the addressable AI infrastructure opportunity beyond scale-out and scale-across. Timeline is unspecified.
• Ben Reitzes, Melius Research: Asked Robbins to elaborate on the "multi-year" cycle framing. Robbins cited end-of-support Cat 4K/6K base, Wi-Fi 7 ramp, and security-in-network as drivers, and explicitly compared to the 2017 Cat 9K launch ("that transition just kept going for five, six, seven years").
  Assessment: Management is now formally framing campus as a 5+ year cycle, which is a notable upgrade in confidence vs. prior quarters.
• Ben Bollin, Cleveland Research: Asked Robbins to compare the AI build-out to the late-1990s internet build-out. Robbins acknowledged the parallel, highlighted faster pace this time, and noted the cohort of buyers is materially stronger ("massive, strong balance sheet, strong cash flow, profitable companies") — a clear pushback on the bubble-comp narrative.
  Assessment: A defensive question handled directly. The "stronger buyer cohort" point is the right one — but the implicit acknowledgment that the question is being asked tells you the bubble debate is starting to register in the boardroom.

What They're NOT Saying

1. Hyperscaler customer concentration: Management cited "four major hyperscalers grew triple-digits" but did not disclose what percentage of the $1.3B Q1 order figure those four represent. If three or four customers drive 80%+ of hyperscaler orders, the risk profile is materially different than if it's spread across eight or ten.
2. Exact composition of the $3B AI infrastructure FY26 revenue: Robbins disclosed it is "balanced" between Silicon One systems and optics but declined to give a hard split. Optics and systems carry materially different margin profiles, so the mix matters for FY26 EPS quality.
3. Specific DRAM pricing assumption in the guide: CFO Patterson acknowledged "pretty significant price increases" and said impacts are "included and considered" in the guide, but did not quantify what DRAM/PCB/optics inflation rate is being modeled. If actual prices accelerate past the modeled assumption, Q2/Q3 gross margin downside is unhedged.
4. FY26 free cash flow guide: The full-year guide raises revenue and non-GAAP EPS but does not provide a FCF target. With Q1 operating cash flow down 12% and capital returns at 125% of FCF, the absence of a FCF guide leaves the buyback sustainability question open.
5. Sovereign export-license timing: Robbins flagged "we're still working through" export licenses for sovereign deals and described H2 FY26 as the expected ramp window — but the question of which jurisdictions, which products, and which timeline was deflected.
6. Security mid-teens growth path: Robbins reaffirmed the mid-teens long-term security target but described the normalization path as "probably going to take us four quarters." That's a long runway for a segment that's currently declining 2%. No bridge from –2% to mid-teens was offered.
7. Insider selling pattern: Not asked on the call, but worth flagging — multiple senior executives (President & CPO, CFO, EVP Sales, EVP Operations, Chief Legal Officer) have been net sellers over the past 90 days. While these are often planned-sale programs, the cluster signals the management team sees the stock as fully valued at current levels.
8. Collaboration strategy: The segment got effectively zero airtime. With –3% revenue and ongoing competitive pressure in video/devices, the absence of any forward commentary is itself a signal that management views it as non-strategic.

Market Reaction

• Pre-print setup: Cisco shares are up 25% this year as of Wednesday's close, topping the 21% gain for the Nasdaq. Stock had recently broken out to multi-decade highs, with sentiment elevated heading into the print and one boutique upgrade in the days prior (Erste Group Hold→Buy on Nov 10).
• After-hours move: The stock rose more than 7% in extended trading immediately following the print on Nov 12. The combination of the headline beat, the AI orders disclosure, and the FY26 guide raise drove the initial spike.
• Next-day open: Shares opened higher and traded up ~4.5% by mid-morning Nov 13, fading from the after-hours highs as the market processed the gross-margin compression detail. The day-after print is consistent with a "well-received but not euphoric" reception — gains held, but the AH peak was not sustained.

The fade from the +7% AH peak to a ~+4–5% intraday gain reflects two competing reactions to the print. The bull case — clean beat, $1B guide raise, $1.3B AI hyperscaler orders disclosure, multi-year campus refresh framing — was strong enough to drive the initial spike. The bear pushback — 170 bps of product gross-margin compression, –12% operating cash flow, security segment miss, and a stock already up 25% YTD into the print — was credible enough to cap the move. The post-print level still puts CSCO firmly at multi-decade highs, which is the right reading of the print: a clean fundamental inflection, partially priced in.

Street Perspective

Debate: Is the AI Orders Number Sustainable?

Bull view: The "2x FY25 orders" commitment from the same customer cohort, combined with the $2B+ incremental sovereign/neocloud pipeline and a ~$3B advance-purchase-commitment build, says management has line-of-sight to a multi-quarter ramp. Pluggable-optics penetration is now 100% of hyperscalers, and the P200-based 8223 router opens scale-across as a new product category.

Bear view: $1.3B of Q1 hyperscaler orders is one data point. Hyperscaler capex cycles have historically been lumpy, and the 2x commentary leans on triple-digit-growth comps from four customers — extremely sensitive to capex deferral by any single name. The "deepening of existing use cases" admission suggests Q1 was wallet-share, not market-share — and wallet share is finite.

Our take: Bulls have the better argument over the next 2–3 quarters, given the order book and the advance-purchase commitments. Beyond that, the scale-across product category (P200) and the sovereign pipeline have to start materializing in real orders to extend the story past FY26. Watch the absolute hyperscaler order number quarterly — if Q2 prints below $1.3B, the durability thesis cracks.

Debate: Is the Margin Compression Structural?

Bull view: The 170 bps product GM compression is component-pricing-driven (DRAM, PCB, optics), already in the guide, and offset by services GM expansion. Operating margin expanded 30 bps YoY — the leverage story is intact, and as Silicon One reaches greater penetration (target: fully rolled by FY29), it should structurally lift product GM.

Bear view: AI infrastructure carries lower gross margin than core enterprise networking (pluggable optics in particular). As the AI mix grows, structural product GM faces a long-term headwind that no level of cost productivity will offset. The Q2 GM guide implies the pressure continues — and DRAM pricing has more room to rise.

Our take: The bear view is partially right on AI mix, but Silicon One penetration and price increases on next-gen campus products should largely offset over the 12–24 month window. The right way to underwrite this is to model gross profit dollar growth (which is accelerating) rather than gross margin rate.

Debate: Is the Stock Already Priced?

Bull view: At ~$76–78 post-print, CSCO trades at ~19x forward non-GAAP EPS midpoint ($4.11) — not demanding for a company guiding 8%+ revenue growth with operating leverage, $42.9B RPO, and a multi-year campus refresh just starting. Compared to other AI infrastructure beneficiaries trading at 30x+ on lower revenue visibility, the relative valuation is attractive.

Bear view: Stock is up ~25% YTD heading into the print and is at multi-decade highs. The 2x-AI-orders commentary is now consensus expectation, and any quarter that doesn't beat the new bar will get sold. Insider selling clustering (~$80M over the trailing 90 days across multiple C-suite officers) signals the team sees the price as fully valued.

Our take: Valuation is no longer cheap, but the order book and guide raise create a buffer. The risk/reward is asymmetric in the bull's favor only if you believe campus refresh and sovereign/neocloud become real contributors in H2 — which is exactly what management is staking the FY26 strongest-year-ever framing on. We are willing to underwrite that.

Debate: Does the Bubble-Comp Question Matter?

Bull view: Robbins's response — that the AI buyer cohort is concentrated in cash-generative, balance-sheet-strong hyperscalers, not VC-funded dot-coms — is structurally correct. Cisco's exposure is to capex from companies that have the cash to absorb a 2-quarter pause without canceling orders.

Bear view: The question being asked of management is itself the signal. When the topic of conversation shifts from "growth durability" to "Are we in 1999?", multiple compression is the historical outcome regardless of whether earnings actually fall.

Our take: The fundamentals support the bull case, but the bear case on multiple compression is plausible. Position size accordingly — Outperform is appropriate, but maximum-weight is not.

Model Update Needed

Valuation impact: Applying ~19–20x to the new $4.11 FY26 non-GAAP EPS midpoint and giving partial credit for sovereign/neocloud upside (not in the guide), we see fair value in the $80–$88 range over the next 12 months, vs. ~$76 post-print. That implies ~5–15% upside, which combined with the dividend yield (~2.1%) supports the Outperform rating but argues for moderate, not aggressive, position sizing.

Thesis Scorecard Post-Earnings

Overall: Thesis strengthened. Four of four bull points confirmed; two of five bear points confirmed, one challenged, one neutral, one partially confirmed. The fundamental quality of the print is better than the modest stock reaction implies.

Action: Upgrade to Outperform from Hold. The order book, guide raise, and multi-quarter campus refresh framing materially improve the forward setup vs. the prior quarter's view, and the explicit exclusion of sovereign/neocloud from the FY26 guide creates a credible upside scenario. Margin compression and full-ish valuation argue against maximum position weight, but the trade-off is favorable.