Cadence Design Systems, Inc. (CDNS) — Coverage File

Data as of: Q2 FY2026 (quarter ended June 30, 2026); FY2025 10-K filed Feb 19, 2026; latest source Q2 FY2026 call, Jul 27, 2026 File last updated: October 10, 2026 (initial build)

---

0. 30-second reopen

• Business: One of two full-flow chip-design software vendors, plus emulation hardware, interface IP and system simulation; ~79% recurring revenue, 44.6% non-GAAP operating margin, ~30% free-cash-flow margin.
• Core debate: Is the AI-driven acceleration to ~19% growth a durable step-up in design activity, or a design-cycle peak amplified by up-front hardware, IP, an acquisition and a China rebound, with AI agents threatening how the tools are priced?
• Base thesis: Design complexity, custom silicon and labour scarcity keep customer EDA spend compounding at mid-teens; agents add a paid layer on top of Cadence's engines; IP and new foundries widen the franchise; margins compound once Hexagon is fixed.
• Main bear case: Agents reprice per-user licences, export controls cut China (13–15% of revenue), and a rising up-front mix ties results to the AI capex cycle, while M&A moves capital into lower-return simulation.
• What to watch: Core EDA and ex-acquisition recurring growth, backlog vs. revenue, up-front share, IP growth, China share and the 9 Nov 2026 BIS rule date, Hexagon accretion in 2027, GAAP-to-non-GAAP gap.
• Valuation: ~$97B market cap, ~44x guided 2026 non-GAAP EPS, ~70x TTM GAAP EPS, ~1.7% TTM FCF yield (Oct 9, 2026 web price): priced for many years of high-teens earnings growth.

---

1. Business snapshot

Cadence sells the software, hardware and IP that semiconductor and systems companies need to design, verify and tape out chips and electronic systems. It reports three product categories [Q2 FY2026 10-Q]: Core EDA (68% of Q2 FY2026 revenue): digital implementation and sign-off, Virtuoso/Spectre analog and custom design, verification software and Palladium/Protium emulation and prototyping hardware; Semiconductor IP (15%): HBM, UCIe, PCIe, LPDDR and SerDes interface and memory IP plus Tensilica DSPs; and System Design & Analysis (SD&A) (17%): PCB and package design (Allegro), 3D-IC, electromagnetic, thermal and CFD simulation, and structural simulation from BETA CAE (2024) and Hexagon's design and engineering (D&E) business (closed Feb 23, 2026).

FY2025 revenue was $5.30B (+14%), GAAP operating income $1.49B (28.2%) and non-GAAP operating margin 44.6% [FY2025 10-K; Q4 FY2025 call]. About 80% of FY2025 revenue was recurring (time-based software licences, maintenance, royalties, hardware leases), and 20% up-front (hardware sales, individual IP licences, SD&A licences longer than a year) [FY2025 10-K]. No customer is 10% of revenue [Q2 FY2026 10-Q]. Revenue by region in FY2025: U.S. 44%, China 13%, other Asia 19%, EMEA 15%, Japan 6% [FY2025 10-K]. ~13,800 employees, mostly engineers [FY2025 10-K].

2. Core debate

The first question is whether 2026's acceleration is structural. Revenue is guided to grow ~19% to $6.26–6.34B, and bookings in a light renewal year produced a record $8.1B backlog [Q2 FY2026 call; Q2 FY2026 10-Q]. But roughly 3 points of growth is Hexagon (computed: $160M ÷ $5.30B), the fastest-growing lines are booked up-front (record hardware, IP +40% in Q2), and China grew 96% in Q2 against a quarter when U.S. licence rules blocked shipments [Q2 FY2026 10-Q; Q2 FY2026 call]. On the 10-Q's pro forma basis Q2 growth was ~17.5% (computed).

The second is who captures value when AI agents do engineering work. Management's position is a "three-layer cake": compute at the bottom, Cadence's physics-based engines in the middle, agents on top, with agents calling the engines more often and the LLM inside the agent being what commoditizes [Q2 FY2026 call]. The bear reading is that licences priced partly on the number of users shrink if agents replace engineers, and that orchestration value accrues to whoever owns the agent.

3. Base thesis

Chip design is getting harder faster than design teams can grow: hyperscalers are building their own silicon, inference is spawning new memory and networking architectures, and chiplets and 3D-IC push packaging into the design flow [Q2 FY2026 call]. Cadence is one of two vendors with a full flow (Synopsys, now with Ansys, is the other; Siemens EDA third) [FY2025 10-K], sells the emulation capacity customers need to verify the largest designs, and is entering foundry ecosystems where it was historically weak. Its revenue is mostly contracted (backlog ~1.4x TTM revenue, computed), its capex is under 3% of revenue, and management says organic incremental margins are close to 60% [Q1 FY2026 call]. Over 2021–2025 revenue compounded ~15% a year and non-GAAP operating income ~21% (computed from [2026 proxy] reconciliation and 10-Ks).

Assumptions this rests on:

1. Design complexity keeps customer EDA, emulation and IP spend growing at mid-teens after 2026, visible in backlog growing at least as fast as revenue.
2. Agentic AI is additive: agents are sold as a new subscription-plus-consumption layer and increase calls to Cadence engines rather than reducing licensed seats [Q1 FY2026 call].
3. IP and SD&A (each approaching a $1B run rate [Q2 FY2026 call]) keep compounding, and Intel 14A, Samsung and Rapidus relationships turn into revenue.
4. Non-GAAP operating margin returns above 45% in 2027 as Hexagon turns accretive, following the BETA CAE pattern [Q1 FY2026 call].

4. Main bear case

The bear case is not that EDA disappears; it is that Cadence is valued as a durable mid-teens compounder while 2026 is a cyclical peak. The evidence that the mix is getting more cyclical: up-front revenue rose from 17% of FY2024 revenue to 22% in Q2 FY2026, and recurring fell from 83% to 79% on a trailing basis [FY2025 10-K; Q2 FY2026 10-Q]; inventory rose from $304M to $390M in six months [Q2 FY2026 10-Q]; the CFO says not to annualize IP's 40% quarter [Q2 FY2026 call]. China is 13–15% of revenue and exposed to U.S. rules that blocked shipments for three months in 2025 [Q2 FY2026 10-Q].

How it breaks: agents let customers do more design with fewer engineer seats before Cadence's consumption pricing is established, so core EDA growth slips below ~10% even as agent "engagements" multiply; the BIS 50%-affiliates rule takes effect in November 2026 or a compliance breach under the 2025 plea agreement restricts exports; and AI capex pauses, hitting hardware and IP first. In that world the multiple compresses on a business whose cost base is mostly engineers.

5. Business quality

High. The tools are embedded in customers' design flows and co-developed with foundries node by node (TSMC N2/A16, Samsung SF2, Intel 14A), and full-flow adoption keeps spreading: 25 new digital full-flow logos in 2025 [Q4 FY2025 call; Q2 FY2026 call]. Revenue is mostly recognised over time on multi-year contracts; 58% of the $8.1B backlog is due within 12 months [Q2 FY2026 10-Q]. Capital intensity is minimal: capex $142M on $5.30B of revenue in FY2025 [FY2025 10-K]. Hardware adds a moat software rivals struggle to copy: 7 of the top 10 customers in 2025 bought both Palladium and Protium [Q4 FY2025 call].

Weak points: R&D runs at 33% of revenue and stock compensation at 8.6% [FY2025 10-K], so the 44.6% non-GAAP margin overstates owner earnings (non-GAAP operating income less stock compensation was 36% of revenue in FY2025, computed from [2026 proxy]); the duopoly is contested by Synopsys and Siemens in every major account; and the 2025 guilty plea puts a compliance condition on Cadence's ability to export [8-K Jul 2025].

6. Key value drivers

Revenue = recurring software (seats, copies and add-ons on ~3-year contracts, with renewal cycles that vary in size year to year [Q1 FY2026 call]) + up-front hardware and IP + royalties and services. Volume drivers: number and complexity of design starts, new customers (hyperscalers, AI startups, physical-AI and auto), foundry ecosystems supported. Price drivers: value-based renewals, add-ons, and the new agent products priced on subscription plus consumption [Q1 FY2026 call]. Mix: hardware carries lower gross margin than software [Q2 FY2026 10-Q]; Hexagon runs at 5–10% margin in 2026 [Q1 FY2026 call]. By group in 2025: core EDA +13%, IP ~+25%, SD&A +13%; in Q2 FY2026: core EDA +18%, IP >40%, SD&A +37% (with Hexagon) [Q4 FY2025 call; Q2 FY2026 call].

7. Financial profile

Sources: [FY2022 10-K (fiscal 2021)], [FY2025 10-K], [2026 proxy, Appendix B].

FY2025: net income $1.11B, diluted EPS $4.06 (non-GAAP $7.14), operating cash flow $1.73B after $140.6M of BIS/DOJ penalties, capex $142M, free cash flow ~$1.59B (computed), buybacks $925M [FY2025 10-K; Q4 FY2025 call]. Diluted shares were 272.7M in 2023 and 273.3M in 2025: buybacks offset dilution but did not shrink the count [FY2025 10-K].

H1 FY2026: revenue $3.06B (+21%), GAAP operating income $882M (28.8%), net income $703M, operating cash flow $991M, buybacks $400M [Q2 FY2026 10-Q]. Hexagon cost $2.2B cash plus 3.2M shares ($902M); goodwill and acquired intangibles rose from $3.47B to $6.79B, cash fell from $3.0B to $1.44B, and the $2.5B of senior notes (2027, 2029, 2034) is unchanged with the revolver repaid [Q2 FY2026 10-Q]. FY2026 guidance: revenue $6.26–6.34B, non-GAAP operating margin 43.75–44.75%, non-GAAP EPS $8.05–8.15 (unchanged from February despite ~$350M more revenue, because Hexagon is ~$0.28 dilutive), operating cash flow ~$2.0B including ~$180M of pre-close Hexagon tax [Q2 FY2026 call; Q1 FY2026 call].

8. Valuation

Management gives no multi-year targets in sources; the framework is annual guidance plus "Rule of 60" (growth plus non-GAAP margin), which 2026 guidance meets for the first time [Q1 FY2026 call]. The 2025 LTP award for executives pays in full only if market value reaches ~$143.6B within five years of grant [2026 proxy], a useful marker of what management is incentivised to deliver. At ~44x guided 2026 non-GAAP EPS and ~70x TTM GAAP EPS, the shares discount high-teens earnings growth for many years; the TTM free-cash-flow yield (~1.7%) is well below what a mid-teens grower needs to justify on cash alone, so the valuation leans on the durability of growth and on GAAP margins converging toward non-GAAP.

Market data (outside sources/, web, dated) Share price $352.84 at the Oct 9, 2026 close; market cap ~$97.2B; 275.4M shares outstanding; P/E 70.15 (stockanalysis.com, retrieved Oct 10, 2026). Derived by this file from filings: TTM revenue $5.84B, TTM GAAP operating income $1.77B, TTM net income $1.38B, TTM non-GAAP EPS ~$7.99 (FY2025 $7.14 + H1 FY2026 $4.07 − H1 FY2025 $3.22, from calls), TTM free cash flow ~$1.68B (operating cash flow $1.85B less capex $0.18B); net debt ~$1.06B at June 30, 2026 ($2.50B notes less $1.44B cash). Implied: EV ~$98.2B, ~16.8x TTM revenue, ~55x TTM GAAP operating income, ~44x TTM and 2026-guided non-GAAP EPS, FCF yield ~1.7%.

9. Management & capital allocation

Anirudh Devgan has been CEO since 2021 and President since 2017 (joined 2012); he also sits on Lam Research's board. CFO John Wall; the chair, ML Krakauer, is independent, and imec's Luc Van den hove joined the board in January 2026 [2026 proxy; 8-K Nov 2025]. Execution has been strong: revenue ~15% and non-GAAP operating income ~21% compounded over 2021–2025, and the 2026 guide was raised twice [2026 proxy; Q2 FY2026 call].

Capital allocation has two parts. Organic reinvestment is disciplined and asset-light. M&A has stepped up: BETA CAE (2024) and Hexagon D&E (€2.70B enterprise value, ~30% paid in stock) push Cadence into industrial simulation, where an analyst on the Q1 call put its share at about a tenth of the market [8-K Sep 2025; Q2 FY2026 10-Q; Q1 FY2026 call]. Returns: ~50% of free cash flow to buybacks is the stated policy [Q2 FY2026 call]; no dividend.

Pay: CEO total compensation was $56.7M in 2025, mostly a five-year LTP award granted every three years that requires roughly doubling market value, with a relative-TSR gate; the new annual PSUs (50% of CEO annual equity) pay half on one-year operating-income growth (paid at maximum for 2025) and half on 1–3-year relative TSR; the cash bonus uses revenue (45%) and non-GAAP operating margin (55%) [2026 proxy]. Non-GAAP excludes stock compensation, so the bonus does not penalise dilution. Shareholders approved 5M more plan shares (~1.8% of outstanding) in May 2026 [8-K May 2026]. Insider transaction data is not in sources.

Claims to verify over time:

• Agentic AI "does not require some massive step increase in investment" and could lower Cadence's own costs [Q2 FY2026 call].
• Hexagon will be accretive in 2027, as BETA CAE was in 2025 [Q1 FY2026 call].
• The Intel agreement is "100% incremental" with most revenue still to come [Q2 FY2026 call].

10. Risks ranked by damage

1. Agents reprice the tools Damage rank: High · Permanence: High · Fixability: Medium Mechanism: Some licences are priced on the number of licensed users [Q2 FY2026 10-Q]; if agents (Cadence's, customers' own, or an LLM vendor's) let customers design with fewer engineers, seat demand flattens unless consumption pricing replaces it, and orchestration value may sit with whoever owns the agent. Evidence today: management expects agents to raise engine usage and prices them on subscription plus consumption; customers will keep writing their own agents; agent revenue is not disclosed or in guidance [Q1 FY2026 call; Q2 FY2026 call]. What would confirm it: core EDA growth below ~10% while agent engagements rise; flat pricing in the next heavy renewal year. Impact: lower growth and margin on the ~80% recurring base that underpins the multiple.

2. China and export controls Damage rank: High · Permanence: Medium · Fixability: Low Mechanism: China was $680M (13%) of 2025 revenue [FY2025 10-K]; U.S. rules can cut shipments overnight, as in May–July 2025, and the BIS 50%-affiliates rule is suspended only until Nov 9, 2026 [Q2 FY2026 10-Q]. Cadence pleaded guilty in July 2025 to conspiring to violate export controls (sales of $45.3M to a Chinese customer in 2015–21), paid $140.6M and is on three-year probation; BIS compliance is a condition of its export privileges [8-K Jul 2025]. Chinese EDA vendors (Empyrean, X-EPIC, Primarius and others) are named competitors that benefit from any disruption [FY2025 10-K]. Evidence today: management expects the current rules' impact to be limited [Q2 FY2026 10-Q]. What would confirm it: the rule taking effect, China below ~10% of revenue, or any compliance finding. Impact: high-margin software revenue lost almost one-for-one to operating income; share lost to domestic tools is permanent.

3. Up-front mix and the AI capex cycle Damage rank: Medium · Permanence: Low · Fixability: Medium Mechanism: Hardware and IP are recognised at delivery and depend on AI/HPC tape-out budgets; the up-front share rose from 17% (FY2024) to 22% (Q2 FY2026) [FY2025 10-K; Q2 FY2026 10-Q]. Evidence today: record, supply-constrained hardware; IP +40% in Q2; inventory +28% in six months [Q2 FY2026 call; Q2 FY2026 10-Q]. What would confirm it: a down year-on-year hardware or IP quarter; backlog stalling. Impact: opex is 54% of revenue and mostly people, so up-front shortfalls hit operating income directly; cyclical, cushioned by backlog.

4. Capital allocation drift Damage rank: Medium · Permanence: Medium · Fixability: High Mechanism: Acquisitions in lower-margin simulation markets dilute returns on a business whose organic engine needs little capital, and buybacks only offset stock compensation. Evidence today: Hexagon: $3.1B total consideration, $2.15B goodwill, $160M of 2026 revenue at 5–10$0.28 EPS dilution [Q2 FY2026 10-Q; Q1 FY2026 call]; flat diluted share count 2023–25 despite $2.2B of buybacks [FY2025 10-K]. What would confirm it: Hexagon still dilutive in 2027; another large SD&A deal; share count rising. Impact: lower per-share value compounding; fixable by policy.

5. Competitive share loss to Synopsys or Siemens Damage rank: Medium · Permanence: Medium · Fixability: Medium Mechanism: Synopsys now owns Ansys [FY2025 10-K], combining EDA with simulation exactly where Cadence is expanding SD&A; leading-edge digital and verification seats are fought account by account [FY2025 10-K]. Evidence today: management reports share gains, competitive wins in hardware and analog, and an improving position at Intel and Samsung [Q2 FY2026 call]; no market-share data in sources. What would confirm it: core EDA growth trailing customers' R&D growth; loss of a marquee digital flow. Impact: slower growth; durable once a flow is lost at a customer.

11. Metrics to monitor

12. Thesis tripwires

13. Open questions

1. How much of 2026's growth is pull-forward in hardware and IP versus durable recurring growth? Filings give only percentages by product group, not hardware revenue.
2. Why does Hexagon contribute only $160M in 2026 when the 10-Q pro forma implies ~$158M in H1 2025 alone (computed)? Purchase accounting, first-half seasonality and contract conversion are possible; sources do not reconcile it.
3. How are agents priced in practice, and do they sit inside renewals or as separate add-ons? Management describes subscription plus consumption but discloses no revenue [Q1 FY2026 call].
4. Which year is the next heavy renewal year? Management says 2026 is light relative to 2025 [Q1 FY2026 call].
5. What does the BIS 50%-affiliates rule do to China revenue if it takes effect on Nov 9, 2026?

14. Update log

15. Source map