Cisco’s AI Orders Boom While Gross Margins Move the Other Way

Cisco’s AI order book is expanding rapidly, but the company’s gross margins are moving in the opposite direction. Management now expects fiscal 2026 hyperscaler AI orders of $9 billion and hyperscaler AI revenue of $4 billion, up from prior targets of $5 billion and $3 billion, respectively, following a strong intake through May 13, 2026. At the same time, total non-GAAP gross margin declined to 66.0% in Q3 FY26 from 68.6% a year earlier, and non-GAAP product gross margin fell to 64.3% from 67.6%.

Fact: Cisco reported $5.3 billion of AI infrastructure orders taken year-to-date from hyperscaler customers and raised its full-year AI outlook on May 13, 2026, citing sustained demand for AI infrastructure (Q3 FY26 slides). Fact: margins compressed year over year in Q3 FY26, both total and product, as disclosed in the press release (press release). Management has previously tied product margin pressure to product mix and higher memory costs, noting “mix and higher memory costs” as the primary drivers (prepared remarks).

Why it matters now: Cisco is delivering record revenue and operating leverage even as gross margins slip. In Q3 FY26, revenue reached $15.8 billion, non-GAAP EPS was $1.06, and non-GAAP operating margin hit a record 34.2%. The company returned $2.9 billion to shareholders in the quarter (Q3 FY26 slides). The mixed picture challenges a simple narrative that AI demand automatically expands margins and suggests a sequencing effect between order capture, hardware fulfillment, and later-phase software and services monetization.

Hyperscaler AI pipeline just got bigger

What changed is the size and certainty of Cisco’s hyperscaler AI pipeline for FY26. After reporting $5.3 billion of AI infrastructure orders taken year-to-date from hyperscaler customers, the company lifted its full-year expectation for hyperscaler AI orders to $9 billion from $5 billion and increased expected hyperscaler AI revenue to $4 billion from $3 billion (Q3 FY26 slides). That step-up implies stronger visibility into delivery schedules and customer build plans.

At the same time, the margin profile is signaling near-term cost and mix realities. Total non-GAAP gross margin in Q3 FY26 was 66.0%, down roughly 260 basis points versus Q3 FY25, while non-GAAP product gross margin declined about 330 basis points to 64.3% over the same period (press release). Management has attributed the product-side pressure to a heavier hardware mix tied to AI infrastructure and to higher memory costs (prepared remarks).

The strongest tell: quarterly order cadence and guidance

The quarter-by-quarter intake shows how AI demand has built across the fiscal year. Cisco disclosed the following hyperscaler AI infrastructure orders progression:

These figures, drawn from Cisco’s quarterly disclosures (Q2 release) and reiterated in the Q3 materials (Q3 FY26 slides), indicate a strong multi-quarter intake, with Q2 above Q1 and Q3 remaining elevated. The full-year uplift to $9 billion in expected orders and $4 billion in expected revenue further underlines the scale of deployment activity anticipated for FY26 (Q3 FY26 slides).

Margin math: why more AI orders can mean lower near-term margins

Fact: margins compressed year over year in Q3 FY26. Cisco’s explanation is straightforward. A heavier product mix oriented to AI infrastructure and higher memory costs have weighed on product gross margin (prepared remarks). Hardware-intensive phases of AI rollouts typically precede the fuller capture of software, subscription, and services revenue. That sequencing creates a gap in which revenue scales while margin percent compresses.

Inference: as AI build-outs move from procurement to operation, the mix could rebalance toward higher-margin elements. The company’s updated AI revenue expectation of $4 billion for FY26, against a larger $9 billion orders outlook, implies a delivery and revenue recognition schedule that could push more revenue into later periods (Q3 FY26 slides). If subsequent phases feature more software and services attach, margin pressure may ease. That remains a scenario, not a guarantee.

Complication: the Splunk transition to cloud subscriptions is creating a near-term drag on revenue growth, according to Cisco’s prepared remarks (prepared remarks). Subscription transitions typically defer revenue recognition versus upfront licenses, which can distort the timing of both reported growth and margin mix. Even with improving ARR trends, the near-term optics can look softer than underlying demand.

Operating leverage and cash returns still speaking loudly

Fact: despite margin headwinds, Cisco delivered record revenue of $15.8 billion in Q3 FY26, non-GAAP EPS of $1.06, and a record 34.2% non-GAAP operating margin. The company returned $2.9 billion to shareholders in the quarter (Q3 FY26 slides). Those outcomes indicate that operating discipline and scale are offsetting the headline gross margin compression at the consolidated level.

Market narrative: the AI cycle is rewarding vendors that can ship high volumes of infrastructure while still expanding operating leverage. Cisco’s ability to post record operating margin alongside lower gross margin suggests mix headwinds are not overwhelming the broader model. However, the quality of earnings will matter. If the mix remains hardware-heavy for longer than expected, aggregate margins could stay under pressure even if operating income grows.

The strongest counterargument: margins can recover as deployments mature

Counterargument: margin erosion tied to AI is not structural. As deployments mature, software, subscriptions, and services could represent a larger share of the revenue mix, improving product and total gross margins. Memory cost inflation can also moderate, removing a non-structural headwind that management cited earlier (prepared remarks). Under this view, today’s compression is a byproduct of timing and cost cycles rather than an enduring feature of Cisco’s AI exposure.

Downside scenario: if hyperscaler demand remains concentrated in hardware-intensive configurations and memory prices stay elevated, margin pressure could persist longer. The Splunk cloud shift, while strategically consistent with subscriptions, could continue to blunt reported growth and margin mix in the near term (prepared remarks). That would leave Cisco leaning on operating leverage and volume to defend earnings while waiting for a richer mix to show up in reported margins.

Signals to confirm or challenge the thesis

These are the concrete indicators most likely to validate or weaken the view that AI orders are booming while margins compress in the near term:

• Non-GAAP product gross margin and total non-GAAP gross margin trends in upcoming quarterly results, relative to Q3 FY26 levels of 64.3% and 66.0% (press release).
• Updates to FY26 hyperscaler AI orders and revenue expectations versus $9 billion and $4 billion, and any color on delivery timing and mix (Q3 FY26 slides).
• Quarterly hyperscaler AI orders cadence relative to Q1 FY26 at $1.3 billion, Q2 at $2.1 billion, and Q3 at $1.9 billion, looking for evidence of sustained intake or normalization (Q2 release).
• Management commentary on memory costs and supply dynamics, given prior attribution of margin pressure to memory pricing and mix (prepared remarks).
• Progress metrics on Splunk’s cloud transition, including any noted impact on reported revenue growth and margin mix during the shift to subscriptions (prepared remarks).
• Operating leverage sustainability, given Q3 FY26’s record 34.2% non-GAAP operating margin and $2.9 billion returned to shareholders (Q3 FY26 slides).

Editorial view: the evidence points to a sequencing effect rather than a broken margin model. Cisco’s upgraded AI outlook and record operating performance coexist with lower gross margins because the company is shipping hardware into hyperscaler build-outs while cost inputs like memory remain elevated. The burden of proof now shifts to mix. If software and services attach improve as deployments move from build to run, margin pressure should fade. Until then, investors should expect strong AI orders to show up first in volumes and operating leverage before they flow through to gross margin percent.

Disclaimer: This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.