Cadence Raises Beyond the Beat, but the Rebound Leaves the Stock Just Short of an Upgrade

Results vs. Consensus

Q3 2026 Scorecard

Results vs. July Guidance

Year-Over-Year Comparisons

Quarter-Over-Quarter Comparisons

Quality of Beat

Revenue assessment: Growth of 22.0% eased from Q2's 24.2% but remains broad: product and maintenance revenue grew 23.2% and every product group grew at a double-digit rate. Two adjustments keep it honest. First, Hexagon's design and engineering business was not in the year-ago base, and management's comments put organic growth "kind of mid teens" for the year once it is excluded. Second, the year-ago quarter itself carried an unusual 18% China share, which management attributed to China revenue timing, so the 22% is achieved against a comparison that was not soft. The recurring share was 79%, up a point sequentially but below the 82% of Q3 2025, which tells us upfront hardware and IP did more of the year-over-year work than in Q2.

Margin assessment: Adjusted operating margin fell 350bp year over year to 44.1% even as non-GAAP gross margin rose 80bp to 88.8%. The squeeze is below gross profit: GAAP R&D grew 27.1%, marketing and sales 25.9% and G&A 44.8%, with headcount up 15% to 15,753. Part of that is the acquired cost base, part is the charitable-contribution timing in G&A, and part is deliberate investment in the agent portfolio. On our math the year-over-year incremental adjusted margin was about 28%, roughly half of the 54%–55% organic range management cited. The sequential 140bp decline was inside the July guide, so this is the planned integration trough, not a surprise; whether it reverses in 2027 is the central test of our Bull 3 pillar.

EPS assessment: Adjusted net income grew 10.2% and adjusted EPS 9.3%, much slower than revenue. Two acquisition effects explain the gap. Adjusted operating income grew only about 13% on the lower margin, and non-GAAP net nonoperating expense rose to about ($23M) from about ($5M) a year ago on our reconciliation, as cash used for the purchase fell to $1.45B from $2.75B and no longer earns interest against the same debt. Diluted shares rose only 0.5% year over year and fell sequentially to 275.3M, so the $400M buyback is now offsetting issuance. GAAP EPS was flat at $1.05: the $42.2M restructuring charge, higher amortization and a 29.0% effective tax rate absorbed the operating growth. We value the adjusted print but note that the gap between GAAP and adjusted EPS widened to $1.06 from $0.88 a year ago.

Segment Performance

Cadence reports revenue mix by product category in rounded percentages and states growth rates on the call; it does not disclose profitability by category. Approximate dollar values below apply the rounded mix to total revenue and are indicative only.

Product Category Mix — Q3 2026

Product Category Mix Trend

Core EDA — Acceleration in the Largest Category

Core EDA growth stepped up to 20% from 18% in Q1 and Q2, and the drivers were spread across all three of its major parts. Digital implementation and signoff added 16 full-flow wins, which matters because the full flow is where Cadence displaces a competitor's synthesis or signoff tool around its already-dominant place-and-route position. Verification software grew strongly in large digital and memory designs, and Palladium and Protium emulation delivered another record quarter with several new logos and expansions at marquee customers. In analog, SpectreFX won a major competitive displacement at an infrastructure silicon supplier with multiple 2nm projects.

"As agentic flows increasingly require full vertical workflows. There is growing demand for our synthesis and sign off solution."
— Anirudh Devgan, President and CEO

The CEO also said the full-flow trend now extends beyond small and mid-sized customers to the largest accounts, noting that "some very, very large customers are deploying our full digital flow." Hardware remains the swing factor inside the category: it is recognized upfront and its timing is lumpy, but management described a strong Q4 pipeline and reiterated another record hardware year.

Assessment: An acceleration in the 69% category is the strongest evidence for Bull 1 this year, because it is not explained by Hexagon, which reports in system analysis. The full-flow wins add a share-gain argument to the volume argument. We still expect some deceleration in 2027 as hardware laps two record years, which is why our base case grows the business excluding Hexagon by about 14%. That is below the roughly 18% that 2026 shows after removing only the acquired business's ~$160M plan, and close to the mid-teens organic rate management describes once smaller acquisitions are also excluded.

Semiconductor IP — Normalizing From a Contract-Driven Peak

IP grew 24%, down from more than 40% in Q2, with the mix steady at 15% of revenue. Demand drivers are unchanged: AI infrastructure, custom silicon at hyperscalers, advanced-node migration and chiplet architectures pull interface, memory and foundation IP. Tensilica had a record revenue quarter and announced a joint development with Analog Devices for audio, and a space-technology company selected Cadence IP for a custom chip program that management framed as the entry point for a broader services and silicon engagement.

"Our IP business delivered another outstanding quarter. Driven by deepening engagements with leading hyperscalers and semi customers."
— Anirudh Devgan, President and CEO

In Q&A the CEO also disclosed that the design IP business, not only Tensilica, now earns royalties.

Assessment: The slowdown is consistent with the July warning that IP is timing-dependent, and 24% is still above the company's organic rate. The new disclosure is that design IP now carries royalties, which the competitor's royalty framing put on the agenda. Management offered no sizing, so we treat royalties as an option on the IP line, not a forecast input.

System Design and Analysis — Scale Arrives, Profit Still Pending

SD&A grew 30% (37% in Q2) and fell a point to 16% of revenue. Allegro won a competitive displacement at a leading EV maker, the Millennium platform gained adoption with its GPU-accelerated electronic and mechanical suite, and data-center operator Switch selected Millennium and Reality Digital Twin for next-generation AI factory design. Inside the acquired Hexagon business, MSC Adams helped secure a displacement at a power company. The CFO said the acquired design and engineering business is running slightly ahead of the roughly $160M annual revenue plan, but that it is "not really contributing much in the way of operating profit at all this year."

"The MSC business is an integral part of our physical AI strategy. And its integration is progressing well."
— Anirudh Devgan, President and CEO

Assessment: Revenue is ahead of plan and the organic SD&A franchise is winning, but the profit contribution is still near zero by management's own account. The Q3 restructuring charge and the CFO's reference to steps being taken now for next year point to a cost reset in the acquired base. That is the mechanism behind our modestly higher FY2027 margin; it is not yet a result.

Revenue Mix by Geography

China — Exposure Keeps Rising Into Q4

China rose to 16% of revenue, its third consecutive sequential increase, or roughly $261M at the rounded mix. The share remains below the 18% of Q3 2025, so it did not flatter the year-over-year comparison, but the year-to-date share is about 14.7% on our math, above the roughly 13% annual assumption management gave in April. The outlook again assumes current export-control rules remain substantially unchanged.

Assessment: Rising China exposure raises the revenue at stake under Bear 2 without any new restriction appearing on this call. We keep a policy interruption in the downside case and do not treat the strong regional trend as removing the risk.

Recurring and Upfront Revenue

Assessment: Recurring revenue grew 18% year over year, against 11% for 2025 as a whole, so the software base is accelerating independently of hardware. The trailing-twelve-month upfront share has risen to 22% from 20%, however, which means more of the year's growth depends on deliveries that can slip between quarters.

Key KPIs

Key Topics & Management Commentary

Overall Management Tone: Management was the most expansive of the three calls this year, leading with a product launch and claims of share gains in almost every segment, and it answered competitive questions with conviction rather than hedging. It was noticeably less forthcoming on economics: monetization questions about the agent portfolio drew product explanations, the restructuring charge went unaddressed, and the 2027 outlook was deferred to February beyond a general promise of integration benefits.

1. The Third Raise of 2026 Is Aimed Mostly at Q4

The annual revenue midpoint rises to $6,402M from $6,300M, now 20.9% growth, and the adjusted EPS midpoint to $8.40 from $8.10. Q3's outperformance against the July midpoint accounts for only $23.3M of the $102M revenue increase and $0.07 of the $0.30 EPS increase. The rest is Q4: the new Q4 midpoint of $1,710M compares with roughly $1,631M implied by July's annual outlook, and Q4 adjusted EPS of $2.19–$2.25 compares with about $1.99 implied in July.

"Given our business momentum, expanding technology leadership, and compelling AI growth opportunities, we are raising our revenue outlook for the year to 21% growth and our EPS outlook by $0.30 to $8.40."
— Anirudh Devgan, President and CEO

The CFO framed the year as the third raise in a row, and the annual outlook now implies Q4 growth of 18%–20% year over year against a strong Q4 2025 base of $1,440M. Revenue from beginning backlog slips to ~64% of the year from ~65%, meaning slightly more of the higher total depends on in-year bookings.

Assessment: A raise that reaches into the next quarter is a stronger signal than a beat-and-hold, and both the July and October raises added more to the quarters still ahead than to the quarter just reported. It confirms Bull 1 and supports a higher FY2027 starting point. The modest increase in in-year dependence is the cost of the higher target.

2. Maestro: Cadence Builds Its Own Chip-Design Models

The quarter's headline launch is Maestro, a portfolio of domain-specialized models post-trained on Cadence's tools and data, paired with a router that picks between Maestro models and commercial or open-weight models task by task. Management claimed up to 30% better quality and up to 5x better token efficiency than general-purpose foundation models, customer post-training on proprietary data, and a choice of cloud or on-premises deployment. Several deployments are under way.

"Initial results are very compelling. Demonstrating up to 30% higher quality and up to 5x greater token efficiency versus general purpose foundation model."
— Anirudh Devgan, President and CEO

The CEO presented Maestro as the "next generation of agents," built on three elements: specialized models, model routing and reinforcement-learning loops with Cadence's engines. He also said the incremental cost of post-training a frontier-class open-weight model is now manageable, and that Cadence's own internal AI use offsets part of the training expense, so the launch should not change the company's financial model.

Assessment: Owning the model layer strengthens Cadence's claim to be the customer's primary interface and protects against a world where general-purpose models commoditize workflow automation. It also adds a new cost line and a new competitive front. The performance claims are management's own benchmarks, and we give them no revenue value until customers pay separately for them.

3. Agent Deployments Move From Pilots to Named Production Use

Management cited more than 100 active agent engagements, up from the 20-plus ChipStack and 25-plus ViraStack engagements it described in July. Named production results included MediaTek deploying ChipStack in its digital verification flow, Renesas achieving a 34x productivity gain with InnoStack, Samsung Foundry using InnoStack for technology migration, Socionext expecting up to 5x productivity, and Nokia reporting 5x acceleration with AuraStack on Millennium. ViraStack engagements rose fourfold.

"But we have maybe 10 endorsements this quarter, so that tells you the extent of the deployment of all these agents."
— Anirudh Devgan, President and CEO

The CEO's framing was that public endorsements typically come from about one in ten deployments, implying a much larger installed base than the named customers. The back-end and analog agents are also more tightly integrated with Cadence's base tools than the front-end ChipStack product, which matters for how much engine usage each agent drives.

Assessment: This is the clearest advance yet on the Bull 2 signpost we set in April: production deployments across all four agents, with named customers for three of them. It still measures adoption, not revenue. Until Cadence discloses agent bookings, pricing or attach rates, we underwrite the value through recurring growth and engine consumption rather than a separate revenue stream.

4. A Different Route From the Competitor's Frontier-Model Partnership

Two analysts asked how Cadence's approach compares with its main competitor's partnership with a frontier AI lab. Management's answer was that Cadence already works with all three leading frontier labs, but wants to control the customer interface, the agent layer and the engines, using whichever model is best for each task. The CFO added that customers can bring their own model infrastructure and pay for its compute, while licensing Cadence's agents and orchestration.

"The key thing in our solution is that we want to control the 3 layers of the cake and we want to control the interaction with our customers. We want to be the primary interface to our customers, which is what our customers want."
— Anirudh Devgan, President and CEO

Assessment: The two strategies diverge on who owns the customer relationship once agents run design flows. Cadence's model-agnostic stance keeps compute costs off its income statement and keeps it in the customer's workflow, which we prefer to a dependency on one lab. The risk is that a frontier-lab-backed rival sets the standard user experience faster; nothing in this quarter's numbers shows that happening.

5. Recurring Growth of 18% and the Agentic Pull on Base Tools

Recurring revenue grew 18% year over year, down from 20% in Q2 but well above the 11% recurring growth of 2025. Management attributed the slower rate to China revenue timing in the year-ago quarter and credited the acceleration over 2025 to agentic workflows, both through new agent licenses and through more calls to existing engines.

"So that is not hardware. that is x hardware. Because hardware would typically show up in nonrecurring growth."
— Anirudh Devgan, President and CEO

Assessment: This is the strongest indirect evidence for the agentic monetization case, because recurring software growth is what an engine-consumption effect would show up in. It is still a correlation: management did not split the 18% between agents, design starts and price. We credit it to Bull 1 with an upside skew for Bull 2.

6. Bookings and Backlog Strength Despite a Light Renewal Calendar

Backlog reached a record $8.2B, up from $8.1B in Q2 and $7.8B at year-end, while next-twelve-month RPO revenue held at $4.2B. One analyst estimated first-half bookings growth in the mid-20% range. Management stressed that 2026 is a comparatively light renewal year, so the growth reflects add-ons and expansions rather than the contract calendar.

"But strong bookings in a lighter renewal year particularly encouraging because they reflect expanding customer engagements and not simply the renewal calendar."
— John Wall, Senior Vice President and CFO

The CEO also highlighted two wide-ranging agreements: an expansion across core EDA, hardware and SD&A with a marquee systems customer, and a broad expansion with a top telecom company that establishes Cadence as its primary design partner.

Assessment: Backlog growth in an off-cycle renewal year supports a higher FY2027 base. The flat near-term RPO is the caveat: the incremental backlog in Q3 sits further out, so it supports duration more than next year's revenue. We raise FY2027 revenue modestly rather than in proportion to the backlog.

7. Hardware: Record Demand, Rising Inventory, No Supply Constraint

Palladium and Protium delivered another record quarter and management reiterated a record hardware year, with a strong Q4 pipeline and growing demand for cloud-based hardware access. Inventory rose to $448.8M from $390.4M in Q2 and $303.5M at year-end. The CFO said the build is for orders in backlog and that there are no capacity or supply-chain constraints.

"So TSMC showed their process road map and said in the next 5 years, the chip size will go up by 48x."
— Anirudh Devgan, President and CEO

The CEO's argument is that emulation capacity scales with design size, and that 3D integration and chiplets make the systems being verified larger and more numerous, so demand is structural rather than a one-off AI build.

Assessment: The design-size argument is sound, but a 48% year-to-date inventory increase is cash tied up until delivery. The hardware line has now set records for two years, which makes 2027 the harder comparison. We assume hardware growth slows next year and treat continued records as upside.

8. Margins: Hexagon Absorbs the Incremental Profit This Year

The annual adjusted operating margin midpoint rises 45bp to 44.7%, slightly above 2025's 44.6%. Non-GAAP operating income grows about $500M on roughly $1.1B of additional revenue, an incremental margin near 45%. The CFO said excluding the acquired design and engineering business, which contributes little profit this year, Cadence would show its usual mid-teens revenue growth, high-teens EPS growth and 54%–55% incremental margin.

"So the so the extra D&E is kind of impacting the incremental margin for this year. But we would expect that to help us next year because the integration has gone really well."
— John Wall, Senior Vice President and CFO

The implied second-half adjusted margin is now about 44.3% (Q3's 44.1% plus a 44.5% Q4 midpoint), against roughly 43.4% implied by the July outlook. Q4 GAAP margin guidance of 30%–31% assumes no further restructuring.

Assessment: Better H2 margins than July implied, plus a restructuring action, move the integration recovery from a hope toward a plan. Our FY2027 margin rises to 46.0% from 45.5%, a 130bp step-up from the higher 2026 base (July's case used 125bp), and implies a 55% incremental margin, the top of management's organic range. That still requires the acquired business to start earning.

9. Buybacks Doubled and a New $2.5B Authorization

Cadence repurchased $400M of stock in Q3, twice the planned ~$200M, at an average of about $307 per share, and plans ~$400M more in Q4. The board added $2.5B of authorization, and the annual commitment moves to "more than 50%" of free cash flow from about 50%. On our math, ~$1.2B of 2026 buybacks equals about 68% of the ~$1.76B of free cash flow implied by the ~$2.0B operating cash and ~$240M capex outlook.

"Underscoring that commitment, our board has authorized an additional $2.5 billion for our share repurchase program."
— John Wall, Senior Vice President and CFO

Debt principal rose to $2.65B from $2.5B with $150M drawn on the revolver, and $500M of term debt is now current. The annual diluted share midpoint falls to 275M from 276.5M.

Assessment: Buying more stock near $307 than planned, partly with borrowed money, is a signal of management's confidence, and it ends the net share creep that concerned us in July. It also leaves less cash cushion against the acquired tax obligation and the inventory build. We lower our FY2027 share count to 274.5M and keep the leverage point on the Bear 1 watch list.

10. Export Rules Remain the Unpriced Condition

The outlook carries the standard assumption that current export-control regulations stay substantially unchanged for the rest of the year. China contributed 16% of Q3 revenue and has risen each quarter of 2026.

"Before I provide our updated outlook, I would like to highlight that it contains the usual assumption that export control regulations that exist today remain substantially similar for the remainder of the year."
— John Wall, Senior Vice President and CFO

Assessment: The caveat is routine, but the revenue it covers is larger than in April. Bear 2 stays EMERGING: no new restriction, more exposure.

11. System Companies Now 45% of the Business

The CEO said 45% of Cadence's business now comes from system companies rather than traditional semiconductor vendors, and pointed to new entrants: a space company, finance firms designing their own chips, automotive and robotics designers. He argued that agents make chip design another 5x–10x easier, which should expand the number of design teams. New design entrants also tend to buy IP, hardware and design services as they start.

"I have mentioned that now 45% of our business is coming from the system companies, the nontraditional versus the traditional semis."
— Anirudh Devgan, President and CEO

Assessment: A broader buyer base lowers dependence on the semiconductor cycle and on a handful of chip vendors, which supports the durability of Bull 1. New logos start small; the effect on growth is gradual, and we do not model a step change from them.

Guidance & Outlook

FY2026 Outlook: July vs. October

Q4 2026 Outlook

The guide pairs a confident revenue call with a cautious read-through to cash. The CFO's framing was that growth is accelerating and broad-based, while the CEO repeatedly described the fourth quarter as an important one still to be delivered and deferred any 2027 view to February. Operating cash flow guidance did not move with the higher revenue and earnings, which we attribute to the inventory build and working capital rather than weaker collections, though management did not say so.

Implied Q-over-Q ramp: Nine-month revenue of $4,692.0M leaves $1,710M for Q4 at the annual midpoint, identical to the Q4 guidance midpoint and 4.7% above Q3. Nine-month adjusted EPS of $6.18 leaves $2.22 for Q4, again matching the guidance midpoint. Nine-month operating cash flow of $1,414.4M leaves about $586M for Q4, versus $553M in Q4 2025.

Street at: Pre-print consensus had FY2026 at $6.32B of revenue and $8.15 of adjusted EPS, so the new midpoints sit 1.3% and 3.1% above. The larger gap is in Q4: consensus of $1.65–$1.66B and $2.03 compares with midpoints of $1,710M and $2.22, which are 3.0%–3.6% and 9.4% higher. Early FY2027 figures of about $7.26B of revenue and $9.73 of adjusted EPS may already include post-print revisions; our $9.93 sits about 2% above them.

Guidance style: July put about $115.5M of its $125M revenue raise into the second half, and this quarter's raise lands mostly in Q4 rather than banking Q3. That pattern points to a deliberately achievable guide. With nine months delivered, the remaining risk sits in hardware delivery timing and upfront IP. We treat the Q4 range as achievable and the top end as plausible.

Analyst Q&A Highlights

What Maestro Is, and How It Gets Paid For

The opening question asked management to explain Maestro in plain terms and to describe how it will be monetized. The answer covered the product architecture in detail (specialized models, routing and integration with Cadence's engines) but did not address pricing, packaging or revenue.

Q: "So is Cadence developing these domain-specific models? Maybe just elaborate and then, how monetization of this product will work."
— Jason Celino, KeyBanc Capital Markets

A: "So if you combine these 3 things together, you know, domain specialized model, domain specialized family of models, plus model routing, plus deep integration with our tools and data, gives a unique differentiation that nobody else can do, and it also allows us to control all 3 layers of the cake and innovate in all 3 layers of the cake."
— Anirudh Devgan, President and CEO

Assessment: The monetization half of the question went unanswered. Management is confident in the technical differentiation and silent on the commercial model, which keeps Bull 2 a deployment story rather than a revenue story for another quarter.

Why Recurring Growth Slowed From 20% to 18%

A question on agentic monetization was framed through the recurring-revenue growth rate, which eased two points from Q2. Management attributed the slowdown to China revenue timing in Q3 2025 and pointed to the much larger acceleration from 2025's 11% recurring growth.

Q: "I believe last quarter you talked about recurring revenue growth of 20%, this quarter was 18%. Maybe talk about what the difference there in the recurring was?"
— Jim Schneider, Goldman Sachs

A: "And if you step back and compare last year, recurring revenue was 11%. And, you know, it is 18%. For this quarter, and it was very good last quarter."
— Anirudh Devgan, President and CEO

Assessment: The explanation is plausible given the elevated China share a year ago, and the year-over-year acceleration is real. Management again declined to isolate the agentic contribution within recurring revenue.

How Cadence Answers a Rival's Mid-Term Targets and Royalty Model

One analyst laid out the competitor's newly stated targets (mid-teens growth next year, a path to 50% operating margins and a royalty model worth nearly $1B over several years) and asked how Cadence sees its own growth, share and margin trajectory. Management answered with its 2026 rule-of-40 math, more than 65 at the guidance midpoint, and treated royalties as one tool among several.

Q: "Anirudh, we recently heard 3 things from your competitor that they can grow sales at a mid-teens pace next year and then compounded to 30%, that they can achieve 50% of the margins. And then also, interestingly, there is a royalty based model that can help add nearly a billion dollars to their top line over the next several years."
— Vivek Arya, Bank of America Securities

A: "In terms of your specific question on royalty, we are always looking at different kinds of business models. So in IP, we always had royalty for a while in Tensilica. And now we have royalty in our design IP business as well. But to me, that is just 1 vector."
— Anirudh Devgan, President and CEO

Assessment: Management declined to match the rival with multi-year targets and pushed the 2027 conversation to February. The disclosure that design IP now earns royalties is new, but unsized. The absence of a counter-target leaves the medium-term margin debate open, which we address through scenarios rather than a house target.

Whether Hardware Can Keep Growing, and What Drove G&A

An analyst asked about hardware sustainability into 2027 given rising inventory, and about a jump in G&A. The CFO said inventory is being built to fulfill backlog with no supply constraint, and attributed the G&A increase to the timing of annual charitable contributions.

Q: "I see even the inventory again grew this quarter. I am wondering, How did it have a strong quarter? How sustainable hardware growth heading into next year, 2027?"
— Siti Panigrahi, Mizuho

A: "We have a strong pipeline entering Q4, and we continue to expect another record year for hardware. As always, hardware could be lumpy from quarter to quarter to quarter depending on product deliveries and customer installation schedules."
— John Wall, Senior Vice President and CFO

Assessment: The 2026 commitment was firm; the 2027 question received the CEO's structural design-size argument rather than a forecast. The G&A explanation is credible and non-recurring, which modestly improves the quality of the Q3 margin.

Why the Raise Carries a Low Incremental Margin

An analyst noted that the profit flow-through on the revenue raise looked thin against Cadence's targeted incremental margin of over 50%. The CFO answered that the acquired design and engineering business contributes almost no operating profit this year, and that the underlying business is running at its usual 54%–55% incremental margin.

Q: "Because I think when you flow through just kind of the incremental margin on the upside and revenue, it is quite significant relative to maybe your kind of targeted over 50%."
— Joe Quatrochi, Wells Fargo

A: "But, I mean, if you took that out, we would be where you typically expect us to be, which is kind of mid teens for revenue growth, high teens for EPS growth, and, you know, with our. In our usual kind of 54%, 55% incremental margin range."
— John Wall, Senior Vice President and CFO

Assessment: This is the most useful financial disclosure of the call. It frames organic growth as mid-teens, quantifies the margin the acquisition is diluting, and commits to a 2027 benefit from integration. It supports our higher FY2027 margin while reminding us that 2026's headline growth includes acquired revenue.

Owning the Models vs. Partnering With a Frontier Lab

A recurring line of questioning contrasted Cadence's own-model strategy with the competitor's frontier-lab partnership. One question asked whether performance or delivery cost drove Cadence's choice. Management stressed model choice, customer post-training and deployment flexibility, and the CFO said the economics should remain those of a software company.

Q: "But is there something in the performance that we need to better appreciate or even in the delivery cost that really led you to go down this route?"
— Lee Simpson, Morgan Stanley

A: "Our strategy remains centered on delivering high value engineering software rather than building a large general purpose AI infrastructure business. We are focused on improving efficiency and the value of project workflows and maintaining a software economics as customer adoption scales."
— John Wall, Senior Vice President and CFO

Assessment: The answer protects the margin model by refusing to carry compute costs, which we view favorably. It also concedes that "the economics of particular deployment options can vary," a reminder that agent pricing may differ by customer and that the revenue per productivity gain is still being set.

What Is Driving Bookings in an Off-Cycle Year

An analyst estimated first-half bookings growth near the mid-20% range and asked whether the driver was agent-led engine usage or semiconductor strength. The CFO pointed to design activity, expanded engagements and add-ons, while cautioning that bookings and backlog reflect contract structure and recognition timing.

Q: "So my question is, you know, what is driving the strong backlog, the strong bookings?"
— Gary Mobley, StoneX

A: "But it is important to remember that back log and bookings number reflects the timing of revenue recognition, that depends on the structure and delivery of individual arrangements and stuff like that."
— John Wall, Senior Vice President and CFO

Assessment: Management neither confirmed nor corrected the analyst's bookings estimate and added a timing caveat. Bookings strength in a light renewal year is a positive, but the flat near-term RPO and this caveat argue against reading backlog growth straight into 2027 revenue.

Whether EDA Growth Is Mostly Hardware

The final question suggested that most of the EDA acceleration came from hardware and that agentic revenue remained small. Management rebutted by pointing to recurring revenue, which excludes hardware, and described agents as a demand accelerator for the core tools.

Q: "And it sounds like majority of that growth primarily driven by hardware. And agentic AI attribution is still quite small."
— Kelsey Chia, Citi

A: "So like I was mentioning earlier and John mentioned also, our recurring growth has improved from 11% last year to 18%. So that is not hardware."
— Anirudh Devgan, President and CEO

Assessment: The rebuttal is fair on hardware: recurring software growth did accelerate. But it did not dispute the second half of the premise, that direct agentic revenue is still small, and the CFO's description of agents as an accelerator of engine consumption implicitly accepts it.

What They're NOT Saying

1. Any price or revenue figure for agents or Maestro: After three quarters of agent launches and now a model portfolio, Cadence still gives no bookings, pricing, attach rate or paying-customer count. The opening question on monetization drew only an architecture answer.
2. The $42.2M restructuring charge: The quarter's restructuring line took GAAP operating margin and GAAP EPS below the July GAAP ranges, yet neither prepared remarks nor Q&A described what was restructured, the headcount involved or the expected savings. The CFO's remark that steps are being taken now to improve next year is the only link.
3. A refreshed acquisition profit bridge: The design and engineering business is "slightly better" than ~$160M of revenue, but there is no update on April's $0.28 EPS dilution estimate, no numerical 2027 accretion target and no status on the roughly $180M acquired pre-close tax obligation.
4. Why operating cash flow did not rise with the raise: Revenue rose $102M and adjusted EPS $0.30 at the midpoint, yet the ~$2.0B operating cash outlook was unchanged. Inventory up $145M year to date and DSO at 66 days are the likely reasons; management did not reconcile them.
5. Any 2027 framework: With a competitor publishing multi-year targets, Cadence deferred all 2027 commentary to February and offered no view on renewal timing, hardware comparisons or margin progression beyond "help us next year."
6. The full-year China share: China has risen to 16% of revenue, but management did not update April's roughly 13% annual assumption or say whether Q4 continues the trend.

Market Reaction

• Pre-print setup: CDNS closed October 26 at $322.17, up 1.8% in that session ahead of the after-close release, 3.1% year to date (versus 12.1% for the S&P 500), down 6.6% over twelve months and 1.2% over thirty days. The pre-print 52-week closing range was $265.66–$416.39.
• After-hours move: Shares were roughly flat in the first minutes after the release, then traded about 2.0%–2.6% higher in after-hours trading on October 26 as the call progressed, and were indicated 2%–3% higher before the October 27 open.
• Reaction session, October 27: The stock opened at $325.96 (+1.2% gap), traded between $323.68 and $348.80, and closed at $343.08, up $20.91 or 6.5%.
• Volume: About 4.7M shares traded versus a 2.4M thirty-day average, roughly 2.0 times normal.
• Market and peer comparison: The S&P 500 rose 0.6% and Synopsys rose 4.4%, after a 5.6% gain on October 26. Cadence outperformed both on the reaction session.
• Follow-through, October 28: The stock slipped 0.2% to $342.46 after again touching $348.80 intraday, holding nearly all of the reaction gain.

A Low Bar Met by a Forward Raise

Cadence entered the print as a relative laggard: up 3.1% for the year against a 12.1% market gain, below its July reaction close, and trading in the lower half of its 52-week range. That setup, combined with a raise that reached beyond the quarter, explains most of the 6.5% move. The gap at the open was small, and the stock built through the session, which suggests buyers were responding to the call and the Q4 guide rather than to the headline beat alone.

A Sector Bid Around the Print

Part of the move was not Cadence-specific. Synopsys gained 5.6% on the day of the release and another 4.4% on the reaction session, so roughly two points of Cadence's reaction-day gain was excess return over its closest peer. The competitive questions on the call show that investors are reading both companies' AI strategies together; on this print, the market rewarded Cadence's version as well as its rival's.

At $343.08 the stock trades at 40.8 times our FY2026 adjusted EPS and 34.5 times FY2027, essentially where it closed after the Q2 print but on higher estimates. The rally restores the July price rather than re-rating the stock, which is why our higher fair value does not translate into an upgrade.

Street Perspective

Debate: Is Owning the Model Layer a Moat or a New Cost Center?

Bull view: Post-trained, domain-specific models plus routing let Cadence deliver better results at lower token cost, keep customer data in the customer's environment and remain the interface for every agentic design flow. A model-agnostic stance avoids dependence on any one lab and keeps compute off the income statement.

Bear view: Model training is a new expense line in a fast-moving field where frontier labs improve faster than any single vendor can, and a rival with a frontier partnership may set the user experience. Management's own benchmarks are not independent proof, and the commercial model is still undefined.

Our take: The bull argument is stronger on strategy, because Cadence's engines and data are what make the models useful and those do not change hands. The bear argument is stronger on timing of value capture. We credit the strategy in our multiple and wait for disclosure before crediting the revenue.

Debate: Can 2027 Sustain High-Teens Growth After a 21% Year?

Bull view: Record backlog, 18% recurring growth, accelerating core EDA and share gains in digital and analog support another year of high-teens growth, and the acquisition rolls into the base with a profit tailwind. More system companies designing chips widen the customer base.

Bear view: 2026 benefits from acquired revenue, two record hardware years and strong IP contracts, all of which create harder comparisons. Flat near-term RPO and management's own mid-teens organic framing point to a slower 2027, and management will set the February guide from a much higher base.

Our take: Mid-teens is the realistic base. Our FY2027 revenue of about $7.3B implies 14.0% reported growth, with upside if hardware sets a third record. The higher margin matters more than the revenue debate for 2027 EPS.

Debate: Has the Rally Used Up the Upside?

Bull view: Estimates are rising every quarter, the stock is roughly flat since July, and at about 35 times next year's earnings a franchise compounding EPS near 18% with a doubled buyback deserves a higher price. The forward multiple also sits below the range the stock held for most of the past five years, and the pre-print discount to the market has not fully closed.

Bear view: A 6.5% one-day move already captured the raise, and at more than 40 times this year's earnings any wobble in hardware, China or the acquisition recovery leaves meaningful downside. The competitor's targets also raise the bar for what investors will expect in February.

Our take: The bear side has the better of it on price, narrowly. Our $370 fair value implies 7.8% from $343.08, roughly in line with the market hurdle, so Hold remains appropriate; a pullback toward the pre-print level, or a February guide that supports FY2027 EPS above $10, would change that.

Model Update & Valuation Framework

We move FY2026 to the October guidance midpoints. For FY2027, our ~$7.3B revenue estimate assumes approximately 14% growth from a ~$6.24B FY2026 base excluding the acquired business's ~$160M plan, plus about $200M from the acquired business, unchanged from July. The base is $102M higher than in July because of the raise. We lift the FY2027 adjusted margin to 46.0% from 45.5%, a 130bp step-up from the higher 2026 starting margin, close to the 125bp step-up in our July case. The FY2027 operating-income increase over FY2026 implies a 55% incremental margin, the top of management's stated organic range, which requires the acquisition to begin contributing.

Earnings bridge: The $100M increase in FY2027 revenue adds about $46M of operating income at the prior margin, and the 50bp margin increase adds about $37M on the new revenue base. With $5M less nonoperating expense, that is about $0.26 per share after tax at the prior share count; the 2M lower share count adds about $0.07, taking the estimate to $9.93 from $9.59. From FY2026 to FY2027, $898M of additional revenue and a 130bp margin increase add $496M of operating income and lift adjusted EPS 18.2%.

Valuation: We retain the 37.5 times FY2027 adjusted EPS multiple used since initiation. Applied to $9.93 it gives $372, which we round to a twelve-month fair value of $370, up from $360 because of higher earnings alone. At the October 27 close of $343.08, fair value implies +7.8% price and total return, assuming no dividend.

The downside assumes growth slows to 7.0% on a hardware air pocket, a China interruption or weaker IP, with margin falling below this year's level as the acquisition fails to improve. The upside requires 20.3% growth, paid agentic adoption, a third record hardware year and the acquired business reaching corporate margins. Each point of FY2027 margin changes EPS by about $0.22 and fair value by about $8; each turn of multiple changes fair value by about $9.93.

Rating judgment: We retain an 8% assumed twelve-month S&P 500 total-return hurdle. The base return of 7.8% is roughly in line with it, and the scenario skew has improved to about +26% / -22% from +22% / -25% in July. Fundamentals have improved for three straight quarters; the rating has not changed because the price has kept pace with our estimates. Hold is the right call at $343, with an upgrade bias if the stock falls back toward the low $320s or the February outlook supports earnings above our FY2027 case.

Thesis Scorecard Post-Earnings

Q3 strengthens the demand and integration pillars and fulfills July's commitments on revenue, adjusted margin, EPS and buybacks. All six pillar statuses are retained; the improvements stay within their current tags.

Commitments to watch: Q4 revenue of $1,695–$1,725M, adjusted EPS of $2.19–$2.25, adjusted margin of 44%–45% and ~$400M of repurchases; FY2026 operating cash flow near $2.0B and buybacks above 50% of free cash flow; another record hardware year; the February 2027 outlook, including the promised benefit from integrating the acquired business. Carry forward the $0.28 dilution estimate, the ~$180M acquired tax obligation, paid agentic or Maestro economics, the full-year China share and the savings from the Q3 restructuring.

Overall: The thesis strengthened. Demand is broader than in July, the integration recovery has a mechanism and an early margin signal, and capital return is more aggressive. The case still lacks two proofs: paid agent revenue and an earning acquired business.

Action: Maintain Hold; fair value $370. The stock offers a market-like return at $343.08 with balanced risk. We would move to Outperform on a pullback toward the low $320s under unchanged assumptions, or on a February outlook and agent disclosure that support FY2027 EPS above our $9.93 estimate. Weaker hardware, a China interruption or a stalled integration would move the case toward the $268 downside.