SYNTHOS RESEARCH

Cadence Design Systems CDNS

Technology · Software - Application · Synthos Deep Dive · 2026-08-04

$340.39
Watch

The Overview

There are essentially two companies in the world whose software you need to design a computer chip. Cadence is one of them. If you are building a processor, you use their tools, or you do not build the processor.

That is an extraordinarily strong position, and the numbers show it. Revenue has grown about 14% a year for three years and 24% in the most recent quarter. The company has beaten analysts' profit forecasts five quarters in a row. It has more cash than debt. It spends a third of its revenue on research, which is what keeps the position.

The shares have nonetheless fallen 18% from their high, and 8% below their own average price of the last fifty days. By the standard technical measures, this is the most beaten-down stock in this batch of twelve.

So why are we saying Watch rather than Buy?

Three reasons.

The price is still high. At $340.49 the shares cost about 43 times last year's profits and 36 times what analysts expect for 2027. That is not cheap even for a very good business, and Cadence pays no dividend — its buyback last year was less than 1% of the company. If the shares do not rise, you earn nothing.

A large slice of the profits goes to employees rather than owners. Cadence pays $455 million a year in stock — about 8.6% of all its revenue — and shareholders approved five million more shares this May.

And the story we are being told about the future comes mostly from someone who benefits from telling it. The bull case for Cadence today is that artificial-intelligence "agents" will design vastly more chips, so demand for design software explodes. In our research library, nine of the nineteen recorded claims about Cadence come from one person: the chief executive of a large chip-hardware company that partners with Cadence, sells to Cadence and buys from Cadence. Six more come from a single podcast episode recorded in 2023. That is not fifteen independent opinions; it is two sources, one of whom is selling.

There is also a specific, dated legal problem. In July 2025, Cadence settled with two US government agencies over export violations between 2015 and 2021, pleading guilty to one count of conspiracy to commit export-control violations. It is now under ongoing compliance supervision. Separately, US regulators briefly banned exports of chip-design software to China in mid-2025, and a broader restriction is currently suspended only until 9 November 2026. China is about 13% of Cadence's sales.

Our estimate of fair value is $381, about 12% above the price. Analysts on average say $404.56. We would buy this at $300 — which is, coincidentally, exactly what the most pessimistic analyst thinks it is worth.


Putting a number on it: our fair-value estimate is $381 against a current price of $340.39 — real upside if our numbers are right.

Our summary metrics

Downside Risk (lower = safer)6/10High

"Rated 6 — an outstanding business carrying valuation, regulatory and concentration risk, at a genuinely oversold moment. The supports: half of a mission-critical duopoly, in the knowledge base's own words a business where 'you literally cannot design a chip without their tools'; revenue growth of 14.8%, 13.5% and 14.1% in the last three fiscal years accelerating to 24.2% in the June quarter; five consecutive earnings beats; NET CASH, with `netDebt` of MINUS $521.2 million; and research and development of $1,768.8 million, 33.4% of revenue, which is the moat being maintained. Against that, five risks. FIRST, valuation: 42.6x trailing adjusted earnings and 35.7x FY2027 consensus, with a shareholder yield near 1% — no dividend and a $925.0 million fiscal-2025 buyback against a $93.8 billion market capitalisation. SECOND, export controls, and this is worse than a standard risk factor: the 10-Q discloses that in July 2025 Cadence settled with BIS and the DOJ over export violations between 2015 and 2021 and 'agreed to plead GUILTY to one count of conspiracy to commit export controls violations', with ongoing audit and compliance obligations; BIS separately imposed licence requirements on EDA software to China from May to July 2025 which 'impacted our ability to deliver software offerings to our customers in China'; and a BIS interim final rule extending Entity List restrictions to entities 50% or more owned by listed parties is SUSPENDED ONLY UNTIL 2026-11-09. China was $679,971,000 of fiscal-2025 revenue — 12.8%. THIRD, stock-based compensation of $455.2 million is 8.6% of revenue, and the 2026 annual meeting approved 5,000,000 additional shares under the equity plan and removed its fixed term. FOURTH, the knowledge-base lane's bull case comes overwhelmingly from a commercially interested party. FIFTH, the technical position — 8.1% below the 50-day average with MACD at −9.72 — is a washout, which cuts both ways."

Growth Quality8/10Very High

"Rated 8 — high, durable, and currently accelerating. Revenue: $3,561.7M (FY2022), $4,090.0M (FY2023, +14.8%), $4,641.3M (FY2024, +13.5%), $5,296.8M (FY2025, +14.1%) — a business that has compounded in the mid teens through a semiconductor cycle. The June 2026 quarter was $1,584.5M against $1,275.4M, UP 24.2%, the fastest in the file, and the four most recent quarters run $1,338.8M, $1,440.0M, $1,474.2M, $1,584.5M. Adjusted EPS from the `earn_cal` actuals is $1.65, $1.93, $1.99, $1.96 and $2.11 across five quarters — a trailing $7.99 — with beats of 5.8%, 7.8%, 4.2%, 2.6% and 2.9%. FIVE CONSECUTIVE BEATS. Consensus wants revenue of $6,319.4M in FY2026 (17 analysts, +19.3%), $7,176.8M in FY2027 (20, +13.6%) and $8,017.4M in FY2028 (14), with adjusted EPS of $8.141 (10 analysts), $9.526 (15) and $10.891 (7) — 17.0% then 14.3%. Geographic detail from `seg_geo`, using only the top-level lines because the block also carries components: Americas $2,479.3M (46.8%), Asia $1,685.2M (31.8%, of which CHINA $680.0M, 12.8%), EMEA $790.6M (14.9%), Japan $341.7M (6.5%). What holds this at 8 rather than 9: part of the FY2026 acceleration is a soft comparison — the 10-Q states the increase in China revenue was 'partially attributable to export license requirements that were temporarily imposed by BIS on EDA Software and Technology from May to July of fiscal 2025' — so the base was artificially depressed."

Exponential Potential7/10High

"Rated 7 — a genuine structural argument, materially compromised by who is making it. The bull case is that agentic artificial intelligence multiplies the number of chip designs and design iterations, and because electronic-design-automation tools are licensed per seat and per compute-hour, the addressable pool expands non-linearly rather than being displaced. The knowledge base states it explicitly: 'Agentic AI lifts the number of ASIC designers ~1000x, so EDA tool-use and license consumption will skyrocket — Cadence's TAM explodes as agents orchestrate its design tools' (2026-05-01, conviction 78), and separately that software-licensing firms 'will grow much larger as they add specialized token rental to their business models' (2026-03-18, conviction 80). Both come from the same commercially interested source. The independent version of the argument is older and weaker: `business_breakdowns` (2023-05-20) that eight of the top ten technology firms now design their own silicon, expanding the research-and-development pool Cadence is paid from; and `no_priors` (2024-11-07, conviction 82) that AI chip designers are already essential and rentable design-space-exploration agents are coming. The underlying moat is not in dispute and is exceptional: a two-firm duopoly, tools that cannot be substituted mid-project, and a cost of design failure the knowledge base puts above $500 million per chip against $10 million masks, which is why price sensitivity is low. A 7: a real and possibly very large TAM expansion, whose principal advocate in our store sells the hardware it would run on."

Fair value$381 $249–$509
What does “fair value” mean?

Fair value is Synthos’s estimate of what one share is worth today, based on our model of the company’s future cash generation and the risks to it. It is not a price target or a prediction of where the stock trades next quarter — it’s the price where we think risk and reward are balanced. Above it, you’re paying for outcomes better than our base case; below it, the market is offering a margin of safety.

The model’s inputs and weightings are proprietary — they’re the product. What we publish is the output, plus an independent, public-math cross-check further down the page (“Check our number”) so you can judge it for yourself.

What do the 5 tiers mean? (Core · Tactical · Watch · Hold · Avoid)
Buy — CoreOwn it as a foundation — start or add now, size it for years, let dips be gifts.
Buy — TacticalGood price + confirmed trend + a defined exit — buy the setup, not a marriage.
WatchWe want the business, just not at this price/setup — act only when the listed trigger hits.
HoldFine to keep if you own it — no reason to buy more; new money does better elsewhere.
AvoidDon't own it — the problem is the business or the expectations, so a cheaper price won't fix it.

The Road Ahead

What we expect to matter in each window, and the evidence that would prove us wrong.

Short term 0-6 months

Neutral
Driver
"The most oversold technical position in this batch, in a business whose revenue just grew 24%. Cadence closed 2026-08-04 at $340.49, up 1.29%, which is 18.2% below its 52-week high of $416.39, 28.2% above its low of $265.66, 8.1% BELOW a 50-day average of $370.67 and 3.9% above a 200-day average of $327.76. RSI is 37.1 — the lowest in this batch and the only reading in oversold territory — and MACD is MINUS 9.72, the most negative in the batch. Three-month return is MINUS 2.6% against SPY's +7.6%; twelve-month MINUS 4.6% against +24.3%; but six-month is PLUS 17.7% against +11.1%, so the weakness is recent and sharp rather than chronic. The fundamental news is good and eight days old: June-quarter revenue up 24.2% and a fifth consecutive earnings beat. What holds the stance at neutral rather than a tailwind is the price: 42.6x trailing and 35.7x FY2027 consensus is a full multiple for a business whose shareholder yield is roughly 1%, and a stock 8% below its 50-day average is falling for a reason the file does not fully explain."
What we’re watching
"The 2026-10-26 print against consensus adjusted EPS of $2.04 and revenue of $1,612M — note that consensus expects a 3.3% SEQUENTIAL EPS DECLINE from the June quarter's $2.11, which is worth understanding before the print rather than after it. Within it: whether the 24.2% revenue growth rate holds, and how much of it is the soft China comparison the 10-Q identifies (BIS licence requirements from May to July 2025 'impacted our ability to deliver software offerings to our customers in China'); the China revenue line, $679,971,000 or 12.8% of fiscal-2025 revenue; and stock-based compensation, running at $455.2M or 8.6% of revenue against a newly enlarged equity plan. Separately and with a hard date: the BIS interim final rule extending Entity List restrictions to entities 50% or more owned by listed parties is under a one-year suspension that EXPIRES 2026-11-09 — thirteen days after the earnings print — 'absent a future extension.' That is a specific, dated regulatory cliff on 12.8% of revenue and it is the single most watchable item in the next six months."
Confidence
Low

Medium term 6-24 months

Tailwind
Driver
"The medium-term case is a mid-to-high-teens compounder in a duopoly, bought after an 18% drawdown. Consensus has adjusted EPS at $8.141 in FY2026, $9.526 in FY2027 and $10.891 in FY2028 — 17.0% then 14.3% — on revenue rising from $5,296.8M actual to $6,319.4M, $7,176.8M and $8,017.4M, growth of 19.3%, 13.6% and 11.7%. The structural drivers are the ones the knowledge base describes and they do not all depend on the interested party: eight of the top ten technology firms now design their own silicon, which expands the customer base rather than consolidating it; the cost of a failed chip design exceeds $500 million against $10 million masks, so tools are bought on capability rather than price; and Cadence's subscription transition, completed years ago, converted a cyclical licence business into recurring revenue. Cadence holds net cash of $521.2 million, generated $1,586.9 million of free cash flow in fiscal 2025 and spent $925.0 million on buyback. The offsets are the price and the dilution: at 35.7x FY2027 the multiple assumes the growth continues, and stock-based compensation at 8.6% of revenue plus 5,000,000 newly authorised shares means a meaningful share of that growth accrues to employees rather than to shareholders."
What we’re watching
"Whether the agentic-AI TAM expansion is real, and whether evidence for it arrives from a disinterested source. The specific claim — that agent proliferation multiplies EDA seat and token consumption — is testable in the revenue line, and Cadence's own disclosure of what drives growth is the place it would appear. Whether the China position stabilises: 12.8% of revenue, a guilty plea to conspiracy to commit export-control violations with ongoing compliance obligations, and a suspended BIS rule expiring 2026-11-09. Whether the duopoly holds — the knowledge base's structural case rests on two firms and a fragmented intellectual-property market with 'no GitHub of IP', and both conditions are contingent rather than permanent. Whether stock-based compensation moderates from 8.6% of revenue, and how quickly the 5,000,000 newly authorised shares are used. Whether acquisitions continue at the recent pace — $429.5M in fiscal 2025, $737.6M in fiscal 2024, $613.8M in fiscal 2022 — and whether they earn their price, given that goodwill and intangibles of $3,467.4M are now 63% of the $5,474.2M equity base. And whether revenue growth holds above 15% once the depressed 2025 China comparison is lapped."
Confidence
Medium

Long term 2+ years

Tailwind
Driver
"Long-run, Cadence occupies one of the most defensible positions in technology and the knowledge base describes it well, from the one source in the lane that is independent of the hardware ecosystem: a mission-critical duopoly where 'you literally cannot design a chip without their tools'; a customer base whose research budgets, not its unit volumes, determine the addressable market — roughly 15% of a $550 billion semiconductor industry's sales; a feared wave of customer consolidation between 2014 and 2017 that did not shrink demand because 'nobody fires engineers post-merger'; and pricing power derived from the cost of failure rather than from switching cost alone. That combination has produced fifteen-percent compound revenue growth through a full semiconductor cycle and 33.4%-of-revenue research spending that widens the moat each year. Layered on top is the agentic-AI argument: if the number of chip designs and design iterations rises by orders of magnitude, a per-seat and per-compute-hour licensing model captures it directly. That argument may well be right. It is, in our store, made almost entirely by a party that sells the accelerators the tools would run on, and we say so rather than counting it as consensus. Anirudh Devgan is chief executive; approximately 13,800 employees."
What we’re watching
"Whether the duopoly survives artificial intelligence rather than being expanded by it — the same agent technology that multiplies tool consumption could in principle compress the design work those tools perform, and no claim in this lane addresses that possibility. Whether an open or model-generated alternative to licensed intellectual-property blocks emerges; the knowledge base's own observation that there is 'no GitHub of IP' is a description of today, not a law. Whether the United States and China settle into a stable export-control equilibrium or continue to oscillate; Cadence has already pleaded guilty once, has ongoing compliance obligations, and has 12.8% of revenue exposed. Whether the semiconductor research budget — the actual pool Cadence is paid from — keeps growing at the industry's historic ~15%-of-sales rate as capital shifts toward manufacturing capacity. Whether acquisition-led expansion into system design, simulation and intellectual property continues to earn its cost of capital, with goodwill and intangibles now 63% of book equity. And whether the share count is genuinely reduced or merely offset: fiscal-2025 buyback of $925.0M against stock-based compensation of $455.2M leaves a net reduction far smaller than the headline."
Confidence
Medium

Exponential Potential

Exponential Potential7/10High

"Rated 7 — a genuine structural argument, materially compromised by who is making it. The bull case is that agentic artificial intelligence multiplies the number of chip designs and design iterations, and because electronic-design-automation tools are licensed per seat and per compute-hour, the addressable pool expands non-linearly rather than being displaced. The knowledge base states it explicitly: 'Agentic AI lifts the number of ASIC designers ~1000x, so EDA tool-use and license consumption will skyrocket — Cadence's TAM explodes as agents orchestrate its design tools' (2026-05-01, conviction 78), and separately that software-licensing firms 'will grow much larger as they add specialized token rental to their business models' (2026-03-18, conviction 80). Both come from the same commercially interested source. The independent version of the argument is older and weaker: `business_breakdowns` (2023-05-20) that eight of the top ten technology firms now design their own silicon, expanding the research-and-development pool Cadence is paid from; and `no_priors` (2024-11-07, conviction 82) that AI chip designers are already essential and rentable design-space-exploration agents are coming. The underlying moat is not in dispute and is exceptional: a two-firm duopoly, tools that cannot be substituted mid-project, and a cost of design failure the knowledge base puts above $500 million per chip against $10 million masks, which is why price sensitivity is low. A 7: a real and possibly very large TAM expansion, whose principal advocate in our store sells the hardware it would run on."

“In software buy the expensive names (security, Cadence, Synopsys, CrowdStrike, Palo Alto) not cheap ones — cheap means higher disruption risk; fast-growers lead the rebound.”
Compound And Friendsconviction 60

What could take this further than the base case — and where the market may be underpricing it. Full forward-growth and acceleration math in Deeper analysis, §4 below.

Deeper analysis

Technicals, fundamentals, valuation, and the full expert-claim evidence panel — the detail behind the numbers above.

Check our number Market-implied growth ≈ 33%/yr This isn’t how we calculate fair value — it’s public math you can verify yourself. To justify today’s $340, earnings would have to compound roughly 33% a year for 10 years (9% discount rate). Analysts forecast ~11%/yr, so the market is pricing in MORE than what the Street expects.

Reference table

Street consensus$404.56 (+18.8%) · median $425 (+24.8%) · high $450 (+32.2%) · low $300 (−11.9%) · 0 strong buy / 26 buy / 4 hold / 1 sell across 31 analysts
Valuation42.6x trailing adjusted EPS ($7.99) · 41.8x FY2026E $8.141 · 35.7x FY2027E $9.526 · 31.3x FY2028E $10.891 · 17.1x book · NET CASH of $521.2M
The June quarter (reported 2026-07-27)Revenue $1,584.5M, +24.2% year on year — the fastest in the file · adjusted EPS $2.11 vs $2.05, +2.9% · the fifth consecutive beat
Geography — usable only if you know which lines to sumAmericas $2,479.3M (46.8%) · Asia $1,685.2M (31.8%), of which CHINA $680.0M (12.8%) · EMEA $790.6M (14.9%) · Japan $341.7M (6.5%) · top-level lines sum to $5,296.8M — exactly revenue; summing every line gives 178.6%
The export-control position — filing-disclosed and datedJuly 2025 settlement with BIS and the DOJ over 2015-2021 violations, including a GUILTY PLEA to one count of conspiracy to commit export controls violations, with ongoing audit and compliance obligations · BIS licence requirements on EDA software to China May–July 2025 which "impacted our ability to deliver software offerings to our customers in China" · a BIS rule extending Entity List restrictions to 50%-owned affiliates is suspended only until 2026-11-09
Capital return and dilutionNo dividend · fiscal-2025 buyback $925.0M (0.99% of market capitalisation) · stock-based compensation $455.2M — 8.6% of revenue · the 2026 annual meeting approved 5,000,000 additional shares and removed the plan's fixed term
ConvictionUniformly bullish and the most contaminated lane in this batch — 22 raw KB hits, 5 used after collapsing. NINE of 19 entity claims (47%) from a commercially interested party; six more from a single 2023-05-20 episode
Technicals−18.2% from the 52-week high of $416.39; +28.2% above the low of $265.66; −8.1% BELOW the 50-DMA ($370.67) but +3.9% above the 200-DMA ($327.76); RSI 37.1 — lowest in this batch; MACD −9.72 — most negative in this batch; 12-month −4.6% vs SPY +24.3%

What the experts actually said 11 traceable claims on CDNS · showing the highest-conviction voices

“Cadence and Synopsys are part of the AI thematic portfolio as software-design names; plans to talk about them soon because there's been some developments.”
Jordi Visserbullishconviction 552026-08-09jordi_visser_ai-HFEex-dRjNs:baedd309cf
“Cadence and Synopsys are a mission-critical EDA duopoly; you literally cannot design a chip without their tools, giving durable pricing power.”
Business Breakdownsbullishconviction 902023-05-20business_breakdowns-qZ0t_p4Wktw:06e999f785
“Software-licensing firms like Cadence, Synopsys, Siemens will grow much larger as they add specialized token rental to their business models.”
Jensen Huangbullishconviction 802026-03-18jensen_huang-xv7UVAfyebk:5537104b3a
“In software buy the expensive names (security, Cadence, Synopsys, CrowdStrike, Palo Alto) not cheap ones — cheap means higher disruption risk; fast-growers lead the rebound.”
Compound And Friendsbullishconviction 602026-04-24
“Durable compounders growing 10-20% at 30-40% free-cash-flow margins that dominate a vertical for decades (Adobe, Autodesk, Ansys, Cadence, Synopsys) are underappreciated great businesses.”
Bill Gurleyneutralconviction 552024-02-08bill_gurley-xgvoYJvbNT8:ad44577c87

Every claim reconciles to a real claim_id in the Synthos knowledge base — this is the evidence the verdict is built on, not vibes. Management (the company itself) is shown but half-weighted; one cautionary voice is included on purpose.

Price & moving averages 12 months · 50 & 200-day averages · 52-week range

254297341385428Aug '25Nov '25Jan '26Apr '26Jun '26Aug '2652w hi $41650-DMA 348Price 340200-DMA 32752w lo $266

Solid line = price · dashed line = 50-day average · dotted line = 200-day average · the two thin horizontal lines mark the 52-week high and low. Price above both averages is an uptrend.

Data summary: last close $340.39, 2% below the 50-day average ($348), 4% above the 200-day average ($327) — a mixed trend. 18% below the 52-week high of $416, 28% above the 52-week low of $266.

Bollinger Bands 20-day average ± 2 standard deviations

240289339389438Aug '25Nov '25Jan '26Apr '26Jun '26Aug '26Price 34020-day avg 329

The shaded band widens when the stock gets more volatile. Riding the upper edge = strong momentum (sometimes stretched); the lower edge = weak / potentially oversold.

Data summary: price $340.39 is currently inside the band (band $309–$349).

RSI (14) momentum gauge · 0–100

705030Aug '25Nov '25Jan '26Apr '26Jun '26Aug '26RSI 54.0

Above 70 (overbought zone, shaded) = overbought, below 30 (oversold zone, shaded) = oversold. Currently 54.

MACD 12 / 26 / 9 · trend & momentum

0Aug '25Nov '25Jan '26Apr '26Jun '26Aug '26MACD -3.3signal -7.1

The MACD line crossing above the signal line (bars flip to the up color) = momentum turning up; crossing below (bars flip to the down color) = turning down. Bar height = the size of that gap.

Data summary: MACD is currently above its signal line by 3.76, positive momentum.

Relative performance vs S&P 500 & its sector (XLK (sector)), set to 100 a year ago

6990112133155Aug '25Nov '25Jan '26Apr '26Jun '26Aug '26XLK (sector) 139S&P 500 119CDNS 96

Solid = CDNS · dashed = S&P 500 · dotted = XLK (sector). A rising line means it is beating that benchmark — the sector line shows whether it is a leader or laggard within its own group.

Forward revenue & earnings actual → estimate · "FY" = fiscal year, "E" = estimate

035810$4BFY23EPS $5$5BFY24EPS $6$5BFY25EPS $7$6BFY26EEPS $8$7BFY27EEPS $10$8BFY28EEPS $11$9BFY29EEPS $11$9BFY30EEPS $11

Darker bars = actual results, brighter = analyst estimates. Taller bars to the right = expected growth.

Key stats an RIA wants

Price$340.39
Market cap$94B
P/E trailing67×
P/E FY26E / FY27E42× / 36×
EV / Sales16.2×
EV / EBITDA43.4×
Gross margin88.5%
Net margin23.6%
Dividend yield0.00%
Beta1.142
52-wk range$266 – $416
RSI(14)56
50 / 200-DMA$348 / $327
12-mo return+-2% (SPY +19%)
Street target$405 ($300–$450)
Analyst grades26 Buy · 4 Hold · 1 Sell
FMP ratingB
Next earnings'2026-10-26 (Q3 2026 earnings, 83 days away; vendor consensus adjusted EPS $2.04 and revenue $1,612M, implying a 3.3% sequential EPS decline and 1.7% sequential revenue growth). Second-quarter results were released 2026-07-27, eight days before this dive, and beat consensus adjusted EPS by 2.9% ($2.11 against $2.05) on revenue of $1,584.5M — up 24.2% year on year — against $1,576.9M estimated. A second dated event follows: the BIS suspension of the expanded Entity List rule expires 2026-11-09.'

1. The business and the numbers

Cadence Design Systems is a Delaware-incorporated software company, Nasdaq-listed, with approximately 13,800 employees and Anirudh Devgan as chief executive. It supplies electronic-design-automation software, licensable intellectual-property blocks and hardware emulation systems used to design integrated circuits and electronic systems.

The financial record:

Fiscal yearRevenuegrowthR&DR&D % revOperating incomeOp marginGAAP EPS
FY2021$2,988.2M$1,134.3M38.0%$779.1M26.1%$2.54
FY2022$3,561.7M+19.2%$1,251.5M35.1%$1,073.7M30.1%$3.13
FY2023$4,090.0M+14.8%$1,441.8M35.3%$1,251.2M30.6%$3.86
FY2024$4,641.3M+13.5%$1,549.1M33.4%$1,350.8M29.1%$3.89
FY2025$5,296.8M+14.1%$1,768.8M33.4%$1,649.8M31.1%$4.09

Revenue has compounded 15.4% a year over four years and the operating margin has risen from 26.1% to 31.1% while research spending held above a third of revenue. That is the signature of a business with pricing power reinvesting in its moat.

The quarterly record, and the acceleration is recent:

Report dateQuarterRevenueYoYAdjusted EPSestimatebeat
2025-07-28Q2 2025$1,275.4M$1.65$1.56+5.8%
2025-10-27Q3 2025$1,338.8M$1.93$1.79+7.8%
2026-02-17Q4 2025$1,440.0M$1.99$1.91+4.2%
2026-04-27Q1 2026$1,474.2M$1.96$1.91+2.6%
2026-07-27Q2 2026$1,584.5M+24.2%$2.11$2.05+2.9%
2026-10-26Q3 2026?$2.04

Trailing four quarters of adjusted EPS: $1.93 + $1.99 + $1.96 + $2.11 = $7.99. At $340.49 that is 42.6x trailing.

Five consecutive beats, and the June quarter's 24.2% revenue growth is the fastest in the file. But the September consensus of $2.04 is BELOW the June actual of $2.11 — a 3.3% sequential decline the street already expects, and worth understanding before the print rather than after it.

And part of the acceleration is a soft comparison the 10-Q identifies explicitly:

> "Additionally, for the three and six months ended June 30, 2026, as compared to the three and six months ended June 30, 2025, the increase in revenue in China was partially attributable to export license requirements that were temporarily imposed by BIS on EDA Software and Technology from May to July of fiscal 2025, which impacted our ability to deliver software offerings to our customers in China."

The June 2025 quarter was depressed by a US export ban. The June 2026 quarter is being compared against it. We do not know how much of the 24.2% is this, because the filing quantifies neither — but "partially attributable" is the company's own word and it should temper any extrapolation of the growth rate.

Geography — and the block only works if you know which lines to sum

seg_geo for fiscal 2025 contains EIGHT lines, four of which are components of the other four. Summing all eight gives $9,460.3M against revenue of $5,296.8M — 178.6%.

LineFY2025Relationship
Americas$2,479.3M= United States + Other America
United States$2,311.0Mcomponent
Other America$168.3Mcomponent
Asia$1,685.2M= China + Other Asia
CHINA$680.0Mcomponent — 12.8% of revenue
Other Asia$1,005.2Mcomponent
EMEA$790.6M
JAPAN$341.7M
Top-level sum$5,296.8M= reported revenue EXACTLY
All-eight-line sum$9,460.3M178.6% of revenue

Used correctly the block is exact; used naively it is 79% over. This is the same double-count structure found on CSX in this batch, and the fiscal-2024 entry — which carries only the four top-level lines and sums to $4,641.3M, exactly revenue — shows the payload has been inconsistent between years about which level it reports.

The genuinely useful datum the block yields: China is $679,971,000, or 12.8% of fiscal-2025 revenue. That number is the size of the export-control exposure and it is not available anywhere else in the payload.

The export-control position — a guilty plea and a dated cliff

From the 10-Q filed 2026-07-29:

> "As previously disclosed, in July 2025, Cadence reached a settlement with each of BIS and the U.S. Department of Justice ('DOJ') that resolved matters relating to export violations that took place between 2015 and 2021. As part of the settlements, Cadence entered into a plea agreement with the DOJ pursuant to which Cadence agreed to plead guilty to one count of conspiracy to commit export controls violations. In addition, Cadence entered into an administrative settlement agreement with BIS. The agreements include ongoing audit, compliance and other obligations."

And the forward-looking restriction, with a date:

> "...effective September 29, 2025, BIS issued an interim final rule that extended the export restrictions imposed on entities identified on the Entity List or the Military End-User List and other certain sanctioned parties, to entities that are 50% or more owned by one or more such entities. However, on November 11, 2025, BIS published a one-year suspension of the new rule that is currently set to expire on November 9, 2026, absent a future extension. We expect the impact of these current expanded trade control laws and regulations on our business to be limited, but we will continue to monitor future developments."

Three things follow. A corporate guilty plea to a conspiracy count is not an ordinary compliance matter; it carries ongoing supervision and materially raises the consequence of any further violation. BIS has already demonstrated, in May-July 2025, that it will suspend EDA software delivery to China outright — and the effect on revenue was large enough for the company to cite it as a driver of the following year's growth. And there is a specific date — 2026-11-09, thirteen days after the next earnings print — on which a suspended rule either lapses or is extended. Management's assessment is that the impact will be "limited"; under the standing house rule, management framing is weighted at half.

2. Valuation — priced in or room?

At $340.49 (market capitalisation $93.77B, implied 275.4M shares):

TrailingFY2026EFY2027EFY2028E
Adjusted EPS$7.99$8.141 (10)$9.526 (15)$10.891 (7)
Estimate range$7.960–$8.352 (4.9%)$8.898–$10.185 (14.5%)$8.796–$14.147 (60.8%)
Growth+17.0%+14.3%
P/E42.6x41.8x35.7x31.3x
Revenue$5,296.8M (FY2025)$6,319.4M (17)$7,176.8M (20)$8,017.4M (14)
Revenue growth+14.1%+19.3%+13.6%+11.7%
Net debtNET CASH −$521.2M
Price / book17.1x
Free cash flow (FY2025)$1,586.9M
Shareholder yield~1.0%

Estimate coverage is very good on the near years. Ten analysts on FY2026 adjusted EPS, fifteen on FY2027 and 20 on FY2027 revenue — the broadest revenue coverage in this batch. The FY2027 EPS range of $8.898 to $10.185 is 14.5% wide; the FY2028 range of $8.796 to $14.147 is 60.8% wide, which is where the agentic-AI disagreement shows up. The FY2029 row rests on 2 analysts and the FY2030 row shows $10.93 — LOWER than FY2029's $11.259, a non-monotonic pair. Both are excluded from every conclusion.

est.ebitdaAvg and est.ebitAvg carry the fixed-ratio fabrication signature and are optimistic in level — REJECTED. ebitdaAvg is exactly 40.697% of revenueAvg and ebitAvg exactly 36.093% in FY2026, FY2027, FY2028, FY2029 and FY2030 alike. The implied 40.7% EBITDA margin compares with the 35.3% Cadence actually achieved in fiscal 2025 ($1,872.4M on $5,296.8M). All forward valuation uses epsAvg.

The balance sheet is clean and lightly levered. Cash of $3,001.3M against total debt of $2,480.2M gives net cash of $521.2M. capitalLeaseObligations reads $0 in fiscal 2025 against $108.9M in fiscal 2024 — a disclosure change rather than an economic one, and it means total debt is not comparable across years. Noncontrolling interests are zero. Goodwill of $2,749.1M plus intangibles of $718.2M is $3,467.4M against equity of $5,474.2M — 63% of book value is acquisition accounting, following $429.5M, $737.6M and $613.8M of acquisitions in fiscal 2025, 2024 and 2022. Tangible equity is approximately $2,007M, giving a price-to-tangible-book near 47x, which is why no book-value anchor is used in the fair-value work.

Peer context is not available in usable form. The vendor peer set is not reproduced in this dive because the relevant comparison — the other half of the EDA duopoly — is the only one that matters, and this file contains no forward-EPS data for it. The knowledge base identifies the comparison explicitly and repeatedly: "Cadence and Synopsys are a mission-critical EDA duopoly." No cross-multiple is drawn.

2a. What today's price assumes (the inversion)

At $340.49 — 41.8x FY2026 consensus, 35.7x FY2027 — the price embeds:

2b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: adjusted EPS growth (+17.0%, from FY2026E $8.141 to FY2027E $9.526) + multiple drift (roughly HELD, 41.8x on the forward year moving to about 40x on the then-forward year) + shareholder yield (+1.0%, buyback only)+12% to +14%.

A caution the addendum requires: the rolldown from 41.8x FY2026 to 35.7x FY2027 at a constant price is mechanical and is not counted as compression. Our base applies 40x to the FY2027 number, which is essentially the multiple the market pays on FY2026 today — so the base assumes the multiple HOLDS. We are neither underwriting a re-rating toward the higher end of Cadence's historic band nor forecasting one toward the lower.

Note the composition: essentially all of the return is earnings growth, and the shareholder yield is roughly 1% before stock-based compensation of $455.2 million — 8.6% of revenue — is netted against the $925.0 million buyback. On that arithmetic the net return of capital to existing shareholders is closer to 0.5% than 1%, and a shareholder here owns growth almost exclusively.

If the multiple fell to 28x FY2027E the price is $267 (−21.6%). If it expanded to 45x, $429 (+26.0%).

2c. Variant perception (where we differ, what would surprise)

Synthos fair values

All three anchors are multiples of the FY2027 consensus adjusted EPS distribution (mean $9.526, low $8.898, high $10.185, 15 analysts). No book-value cross-check is applied, because 63% of book equity is acquisition accounting and price-to-tangible-book is near 47x.

Base is 11.9% above spot; asymmetry roughly 1.84:1 (26.9% down, 49.5% up), with a shareholder yield near 1% before stock-based compensation. A base under 12% with a payoff ratio under 2:1 and effectively no yield does not clear the Buy bar in this batch, where the three Tactical names carry base returns of 12.0%, 13.1% and 15.5% with payoff ratios of 2.0:1, 2.4:1 and 2.17:1. Watch, with the trigger named in Section 4.

3. Knowledge base — nineteen bullish claims, two sources, one of them selling

Raw hits: 22. Entity matches: 19. Free-text hits: 3. Used after collapsing: 5. Discarded or collapsed: 17. Distinct independent voices: 4.

The sweep ran case-sensitive entity tokens CDNS, Cadence and Cadence Design Systems, plus case-sensitive free-text patterns for the same, across all 51,928 distilled claims. Every one of the nineteen entity claims is bullish or neutral. Not one is bearish. That uniformity is itself the warning, and the reason is in the sourcing.

DEFECT ONE — COMMERCIAL INTEREST, at 47% of the lane. Nine of the nineteen entity claims come from the channel jensen_huang, the chief executive of a large accelerated-computing hardware company that is simultaneously Cadence's co-development partner, a customer, and the supplier of the acceleration technology Cadence embeds in its tools. His claims, dated between 2024-11-14 and 2026-05-01, are the entire forward bull case:

> 2026-05-01 · bullish · conviction 78 · entities: CDNS · channel: jensen_huang · speaker: Jensen Huang

> "Agentic AI lifts the number of ASIC designers ~1000x, so EDA tool-use and license consumption will skyrocket — Cadence's TAM explodes as agents orchestrate its design tools."

> 2026-03-18 · bullish · conviction 80 · entities: Cadence, CDNS, Synopsys, SNPS, Siemens · channel: jensen_huang

> "Software-licensing firms like Cadence, Synopsys, Siemens will grow much larger as they add specialized token rental to their business models."

> 2026-02-04 · bullish · conviction 85 · channel: jensen_huang

> "The thesis that AI will replace software tool companies is 'the most illogical thing in the world'; AGI will use tools like these, not reinvent them."

Six further claims from the same channel make variants of the same argument across 2024-11-14, 2026-01-05, 2026-02-08 (two), 2026-02-26 and 2026-04-15, including that his company "infuses its tech into EDA/design partners" and that EDA leaders are "integrating" his platforms. One additional claim, dated 2025-06-11, concerns European sovereign artificial-intelligence infrastructure and names Cadence only in a list; it is DISCARDED as not a view on this company, and it also contains an AI-laboratory name which is redacted here on house rules.

This is precisely the defect the programme catalogued when a chip-maker's chief executive supplied 29% and 27% of two software lanes while being a co-development partner. Here it is 47%, and it is the same person. We do NOT exclude the claims — the mechanism described is specific, quantified and testable, and excluding the most informed voice in an industry because he is also in it would be its own error. We discount them heavily, we run the concentration test with and without them, and we do not treat "nine bullish claims" as nine votes.

DEFECT TWO — SINGLE-EPISODE INFLATION. Six further entity claims come from business_breakdowns and ALL SIX ARE DATED 2023-05-20 — one podcast episode distilled into six records:

> 2023-05-20 · bullish · conviction 90 · entities: Cadence Design Systems, CDNS, Synopsys · channel: business_breakdowns

> "Cadence and Synopsys are a mission-critical EDA duopoly; you literally cannot design a chip without their tools, giving durable pricing power."

The remaining five from the same episode cover: Cadence holding "roughly a third of a ~$10B EDA market sitting atop a ~$550B semiconductor industry that spends ~15% of sales on R&D"; that feared 2014-17 customer consolidation "didn't shrink EDA demand; nobody fires engineers post-merger"; the subscription turnaround; that "high cost of design failure (>$500M/chip, $10M masks)" supports pricing power; and that fragmented intellectual property with "no 'GitHub of IP'" drives licensing. Collapsed to ONE analysis. This is the highest-quality independent content in the lane and it is three years old.

FIFTEEN OF NINETEEN ENTITY CLAIMS ARE THEREFORE TWO SOURCES, ONE OF THEM COMMERCIALLY INTERESTED.

What survives as genuinely independent — three claims:

> 2024-02-08 · neutral · conviction 55 · entities: ANSS, CDNS, SNPS, ADBE, ADSK · channel: bill_gurley · speaker: aaron_levie · speaker_role: independent

> "Durable compounders growing 10-20% at 30-40% free-cash-flow margins that dominate a vertical for decades (Adobe, Autodesk, Ansys, Cadence, Synopsys) are underappreciated great businesses."

> 2026-04-24 · bullish · conviction 60 · entities: CDNS, SNPS, CRWD, PANW · channel: compound_and_friends · speaker: Adam Parker

> "In software buy the expensive names (security, Cadence, Synopsys, CrowdStrike, Palo Alto) not cheap ones — cheap means higher disruption risk; fast-growers lead the rebound."

> 2024-11-07 · bullish · conviction 82 · entities: Nvidia, Synopsys, Cadence · channel: no_priors

> "AI chip designers are already essential (couldn't build Hopper without them); expect rentable Synopsys/Cadence AI engineer agents exploring design space combinatorially."

Note that even the no_priors claim sits inside the same hardware ecosystem and references the same vendor's product generation. The genuinely arm's-length content in this lane is two claims: bill_gurley and compound_and_friends, both of which are category observations rather than analyses of Cadence.

THREE FREE-TEXT DISCARDS, and one is a confirmed store defect:

Concentration test — RUN TWICE, and it fails both ways. WITH the interested party: one channel supplies 47% of the entity lane, far past the one-third threshold. WITHOUT him: the lane is one collapsed episode plus three claims, of which one is itself ecosystem-adjacent — breadth 4, and the forward TAM argument disappears entirely. Neither configuration supports treating this lane as multi-sourced corroboration.

Attribution note. Two claims carry named speakers (Jensen Huang on his own channel; Adam Parker on compound_and_friends; aaron_levie on bill_gurley); the rest are channel-attributed with speaker: null. No Cadence management voice appears — the interested party is a counterparty, not this company's management, which is a different and in some respects worse problem: management framing is at least labelled.

Conclusion. Breadth 4, claim count 5 after collapsing, net conviction positive-moderate-discounted. The knowledge base is uniformly bullish on Cadence and it is not a consensus. It is one excellent three-year-old independent analysis of the duopoly, two category observations, and a large block of forward-looking claims from a company that sells into and buys from the business being praised. That is a genuine finding about our own store, and it is a material input to the Watch verdict.

4. Data integrity, insiders, and the verdict

Data integrity — six findings

1. seg_geo double-counts to 178.6% of revenue if summed naively — usable only at the top level. The fiscal-2025 entry contains Americas AND its United States and Other America components, and Asia AND its China and Other Asia components. Top-level lines alone (Americas, Asia, EMEA, Japan) sum to $5,296.8M — exactly reported revenue. All eight lines sum to $9,460.3M. The fiscal-2024 entry carries only the four top-level lines and reconciles exactly, so the payload is inconsistent between years about which level it reports. This is the same structure found on CSX in this batch. The block is used only at the top level, and it is the sole source for the China figure of $679,971,000 (12.8% of revenue).

2. est.ebitdaAvg and est.ebitAvg carry the fixed-ratio fabrication signature and are optimistic in level — REJECTED. ebitdaAvg is exactly 40.697% of revenueAvg and ebitAvg exactly 36.093% in FY2026, FY2027, FY2028, FY2029 and FY2030 alike, against an actual fiscal-2025 EBITDA margin of 35.3% ($1,872.4M on $5,296.8M). All forward valuation uses epsAvg.

3. The est.epsAvg array is non-monotonic at the far end and thin. FY2029 shows $11.259 on 2 analysts and FY2030 shows $10.930 on 5 — a 2.9% decline. Both are excluded from every conclusion. The FY2028 range of $8.796 to $14.147 — a 60.8% spread on 7 analysts — is reported because it is the honest measure of disagreement about the agentic-AI thesis, and it is used for cross-check only.

4. capitalLeaseObligations reads $0 in fiscal 2025 against $108.9M in fiscal 2024 and $115.4M in fiscal 2023. Finance leases do not vanish; the line was reclassified. Total debt is therefore not comparable across years ($2,480.2M in FY2025 against $2,585.1M in FY2024), and netDebt of −$521.2M is stated on the fiscal-2025 basis. The direction is conservative — if the leases still exist, net cash is smaller — and net cash is not load-bearing in this dive.

5. Book value is 63% acquisition accounting, so no book-based anchor is usable. Goodwill of $2,749.1M plus intangibles of $718.2M is $3,467.4M against total equity of $5,474.2M. Tangible equity is approximately $2,007M, giving price-to-tangible-book near 47x against the vendor's priceToBookRatioTTM of 17.1x. No book-value cross-check is applied to any fair value in this dive.

6. Stock-based compensation of $455.2M is 8.6% of revenue and is not reflected in the adjusted-EPS basis the estimates use. The fiscal-2025 buyback was $925.0M; stock compensation was $455.2M. The net return of capital to existing shareholders is therefore roughly $470M — 0.5% of market capitalisation — not the 0.99% the buyback alone implies. The 2026 annual meeting approved an amendment adding 5,000,000 shares to the Omnibus Equity Incentive Plan (1.8% of the share count) and eliminating the plan's fixed term, which was to have expired 2030-04-30.

Tests that PASSED. inc_q passed the MO/APP two-basis test — nine quarters on one basis, with the four most recent matching earn_cal.revenueActual exactly. earn_cal passed the SPGI mixed-basis test — all six actual-versus-estimate pairs on the adjusted basis. seg_geo reconciles exactly at the top level in both fiscal 2025 and fiscal 2024. Enterprise value carries no noncontrolling-interest omission, because NCI is zero. quote.yearHigh/yearLow ($416.69 / $262.75) against tech.hi52/lo52 ($416.39 / $265.66) — a 0.07% and 1.1% discrepancy; tech is used throughout. The implied share count (275.4M) is 0.8% above inc_a's weighted-average diluted 273.3M, consistent with a rising count net of buyback. returnOnEquityTTM is present in km_ttm and absent from ratios_ttm; it is not used, because the equity base is 63% intangibles.

Non-equity tripwire — checked and passed. CDNS is common stock, Nasdaq-listed. Price of $340.49 is not par-like; beta is 1.148; there is no dividend (lastDividend 0); volume was 1.86M shares (~$634M of turnover); the 52-week band of $265.66 to $416.39 is a 57% range. This is common equity.

Insiders — one officer, one day, six sales

DatePersonRoleTypeSharesPriceHeld after
2026-07-22Chin-Chi TengSenior Vice PresidentM-Exempt (option exercise)1,000$202.94146,281
2026-07-22Chin-Chi TengSenior Vice PresidentS-Sale720$335.695145,561
2026-07-22Chin-Chi TengSenior Vice PresidentS-Sale252$336.711145,309
2026-07-22Chin-Chi TengSenior Vice PresidentS-Sale200$337.782145,109
2026-07-22Chin-Chi TengSenior Vice PresidentS-Sale1,968$339.054143,141
2026-07-22Chin-Chi TengSenior Vice PresidentS-Sale200$340.704142,941
2026-07-22Chin-Chi TengSenior Vice PresidentS-Sale440$342.150142,501
2026-07-30Young SohnDirectorG-Gift9$015,243

Seven of the eight transactions are one officer on one day. Chin-Chi Teng, a Senior Vice President, exercised 1,000 options at $202.94 and sold 3,780 shares across six tranches at $335.70 to $342.15 — that is 2,780 more shares sold than were exercised, reducing the holding from 146,281 to 142,501, a 2.6% reduction. The sale was executed on 2026-07-22, five days before the second-quarter results, at prices essentially identical to today's. The eighth transaction is a nine-share director gift.

The reading is limited and we state it as limited. The pattern is a scheduled ladder — six tranches at rising prices on a single day is one order executing — and a 2.6% reduction by one Senior Vice President is not a signal. What matters more is what the file does NOT contain: no transaction of any kind by Anirudh Devgan, the chief executive, or by the chief financial officer, and not one open-market purchase by anyone. On a stock 18% below its high with the lowest RSI in this batch, the absence of any insider buying is mildly notable and is recorded as such.

Verdict

Watch.

The case for interest is strong. Cadence is half of a duopoly the knowledge base describes as one where "you literally cannot design a chip without their tools"; it grew revenue 24.2% in the June quarter and 14-15% a year for four years; it has beaten consensus five quarters running; it spends 33.4% of revenue on research; it holds net cash; and it is 18.2% below its 52-week high with an RSI of 37.1 and a MACD of −9.72 — the most oversold configuration in this batch. Base fair value $381 is 11.9% above spot with 1.84:1 asymmetry.

The case for waiting. A 11.9% base with a 1.84:1 payoff and effectively no yield does not clear the bar in a batch where the three Tactical names carry 12.0%, 13.1% and 15.5% base returns with payoff ratios of 2.0:1, 2.4:1 and 2.17:1. The price is 42.6x trailing and 35.7x FY2027, and roughly half the buyback is consumed by stock-based compensation at 8.6% of revenue. Part of the current growth acceleration is a comparison against a quarter the US government depressed — the 10-Q says so. There is a specific dated regulatory cliff on 2026-11-09, on a company that has already pleaded guilty to one count of conspiracy to commit export-control violations and operates under ongoing DOJ and BIS compliance obligations, with 12.8% of revenue in China. And our own knowledge base, which reads as nineteen bullish claims, is fifteen claims from two sources, one of whom sells into and buys from the business he is praising.

The named trigger: $300. At $300 the same FY2027 consensus of $9.526 is 31.5x, the base case becomes +27%, asymmetry improves past 2.5:1, and the entry sits 12.9% above the 52-week low. That is also, precisely, the street's lowest published target. At $300 this becomes Buy — Tactical on arithmetic alone, absent an adverse development below.

Pre-registered conditions that would move CDNS to Buy — Tactical:

Pre-registered conditions that would move CDNS to Avoid:

Where CDNS fits in the Synthos Framework Portfolio. The technology sleeve — on the watchlist at zero weight today, with a $300 limit and a review scheduled after both the 2026-10-26 print and the 2026-11-09 BIS date. Sizing note: this is a business we would very much like to own and a price we would not pay. The two dated events in the next hundred days — an earnings print the street expects to show a sequential decline, and a regulatory suspension that expires — are exactly the kind of catalyst that produces the entry, and the discipline is to have the number ready rather than to anticipate it. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $340.49, with the fair-value anchors, the $300 trigger and both condition sets gradeable.

Single biggest risk: China export controls, on a company that has already pleaded guilty once. 12.8% of fiscal-2025 revenue — $679,971,000 — came from China. In July 2025 Cadence settled with the Bureau of Industry and Security and the Department of Justice over violations between 2015 and 2021 and pleaded guilty to one count of conspiracy to commit export-control violations, accepting "ongoing audit, compliance and other obligations." BIS has already demonstrated it will stop EDA software delivery to China outright: it did so from May to July 2025, and the effect was large enough that the company cites the depressed base as a driver of this year's growth. A BIS rule extending Entity List restrictions to entities 50% or more owned by listed parties is suspended only until 2026-11-09, "absent a future extension." Management expects the impact to be "limited" and that framing is weighted at half under house rules. A dated binary on an eighth of revenue, at a company under active compliance supervision, at 35.7x forward earnings, is the reason this is a Watch with a named price rather than a position.


Provenance & disclosures