Hook: The CEO Who Called His Own Stock Cheap
On a quiet Tuesday, Cadence’s CEO stepped onto a niche crypto-finance podcast and dropped a bomb: the company is "undervalued" amid the AI boom. The market barely blinked. Cadence shares inched up 2% before fading. But the statement was not a PR stunt. It was a structural signal.
I have spent 14 years dissecting narratives in this industry. When a CEO of a $70 billion semiconductor tool company goes public with a valuation gripe, it is rarely about the next quarter. It is about a paradigm shift the market has not yet priced in. The shift here is the transformation of EDA from a cyclical software line item into a permanent, accelerating tax on the AI economy. The market sees a $150 billion EDA market. I see a $300 billion platform-layer that owns the keys to every AI chip ever designed. That gap is the alpha.
Context: The Invisible Leverage of EDA
Electronic Design Automation (EDA) is the software that designs chips. Without it, no GPU, no ASIC, no TPU. Every modern AI accelerator—from NVIDIA’s Blackwell to Google’s TPU v6—is born inside Cadence or Synopsys tools. The economics are extreme: every $1 of EDA revenue supports $200–$300 of semiconductor output and $5,000–$10,000 of end-device value. This is not a software company. It is a lever on the entire physical computing stack.
Cadence and Synopsys form a duopoly, controlling ~60% of the global EDA market. Their moats are not just code—they are decades of process design kits (PDKs) bonded to every major foundry, from TSMC to Samsung to Intel. Switching a full-chip design flow costs millions in retraining and revalidation. Customers do not leave. They pay the annual subscription and move on. This is the closest thing to a software tollbooth in the industrial world.
Yet the market has historically valued EDA firms like cyclical semiconductor plays, with P/E multiples in the 20–30x range. Meanwhile, the AI infrastructure spending spree—expected to reach $300 billion annually by 2027—directly flows through these tools. Every new AI chip design project requires an EDA license. Every custom ASIC for a hyperscaler buys Cadence IP. The CEO’s "undervalued" claim is a direct challenge to the valuation framework the market still uses.
Core: The Three Mechanical Reasons Cadence Is Undervalued
Let me cut through the narrative noise and lay out the three structural drivers that the market is missing.
1. The Business Model Shift from License to Tax
Cadence is transitioning from a perpetual license model to a subscription + consumption model, especially via its cloud platform (Cadence Cloud). This changes the revenue recognition from lumpy, large upfront payments to recurring, predictable streams. But more importantly, it converts EDA from a cost center into a variable cost tied to chip volume. As AI chip volumes explode, each additional GPU sold implicitly generates more EDA consumption through design iterations, updates, and new projects. The market still values Cadence as a "software" company, but its revenue elasticity to AI chip output is approaching that of a royalty. Yield is the lie; liquidity is the truth. The liquidity of AI capex is now flowing into EDA as a continuous tax, not a one-time purchase.
2. The AI-EDA Flywheel
Cadence’s own AI platform, Cadence.AI, is embedding machine learning into the design flow. This does two things: it reduces design time (which could reduce per-project revenue in a static world) but dramatically increases the number of design starts. Shorter time-to-market means more projects per year. The AI-EDA flywheel is self-reinforcing: AI chips need better design tools, which require more AI to optimize, which generates more EDA revenue per chip. The CEO’s point is that the market sees the capex on AI hardware but ignores the multiplicative effect on the tools that enable that hardware. Auditing the code, not the charisma. The code here is the hook: each design iteration on a 2nm node costs $500 million+ in total, and EDA’s share is ~25-30%. Every new node brings a step-function increase in EDA spend per chip.
3. The System-Design Platform Expansion
Cadence is no longer just a chip design tool. It is aggressively expanding into system-level design: PCB, multi-physics simulation, 3D-IC packaging, and chiplet integration. The acquisition of Invecas and OpenFive, combined with the failed but still influential partnership with Ansys, signals a push into a $300 billion total addressable market (TAM) that includes entire electronic systems, not just chips. The market still prices Cadence against the ~$150 billion EDA TAM. If the transition succeeds, the stock should rerate toward a platform multiple. The CEO’s undervaluation thesis is a bet on that TAM expansion being recognized.
From my experience auditing the ICO mania of 2017—where 80% of utiliy tokens were dead on arrival—I learned to spot when a narrative is lagging reality. Here, the narrative is still stuck in "EDA is a cyclical software supplier." The reality is "EDA is an AI infrastructure royalty." The data reveals the path: Cadence’s revenue growth has accelerated from 8% CAGR in 2019-2022 to 15-20% in 2024-2025, driven entirely by AI chip design demand. The margin structure is improving: gross margins above 88%, operating margins above 35%, and free cash flow conversion above 100%. Yet the forward P/E sits at ~30x, barely above the pre-AI era. The market is discounting the structural shift.
Narrative follows logic, never precedes it. The logic is simple: AI chip design complexity is growing exponentially, and EDA is the only way to manage that complexity. As long as the number of transistors per chip doubles every two years, EDA spend per chip grows. And as long as hyperscalers design custom silicon, the number of design starts grows. Cadence is the toll booth on both axes.
Contrarian: The Risks the Market Is Overpricing
The bear case on Cadence centers on three points: China export controls, competition from Synopsys, and the risk of AI capex slowdown. All three are real, but all three are overpriced in the current valuation.
China Risk: China accounts for ~14-17% of Cadence’s revenue. Export controls limit the sale of advanced EDA tools to Chinese customers. The market treats this as a permanent drag. But the drag is largely priced in. The revenue growth from the rest of the world—especially the US CHIPS Act, EU Chip Act, and Japan’s semiconductor revival—is more than offsetting the China stagnation. Furthermore, any geopolitical thaw (unlikely but possible) would unlock a massive re-rating. The "China premium" for risk is currently a discount on the stock. Pivot not panic: The data reveals the path. The path is global diversification, not dependence.
Synopsys Threat: Yes, Synopsys is larger and acquired Ansys for $35 billion, giving it a stronger system-simulation suite. But EDA is not a winner-take-all market. The duopoly is stable because most customers want multiple suppliers to avoid vendor lock-in. Cadence’s strength in analog/mixed-signal, custom design, and PCB gives it a differentiated moat. The market often frames the competition as a zero-sum game, but the TAM is growing so fast that both can win. The mistake is to assume that Synopsys’s growth implies Cadence’s stagnation. The data shows Cadence’s market share has actually been stable or slightly increasing in the AI segment.
AI Capex Slowdown: The most common bear argument is that AI infrastructure spending is a bubble, and when it pops, EDA will crash. But this ignores the long lead time of chip design. AI capex decisions today drive EDA revenue 2-3 years from now. Even if hyperscalers cut capex in 2025, the design projects already in the pipeline will sustain EDA revenue through 2027. Moreover, the secular trend of chip complexity rising ensures that the per-chip EDA spend remains elevated even if unit volumes dip. The market is overestimating the cyclicality of the business. In reality, Cadence’s revenue has been more resilient than analog chipmakers during the last three downturns.
Takeaway: The Next Narrative
The market will eventually reprice Cadence not as a software vendor but as an AI infrastructure royalty. The catalyst will be the next earnings call where the company shows that AI-related revenue (directly tied to AI chip design wins) exceeds 50% of total revenue. That moment will force a reframe. Until then, the CEO’s "undervalued" call is a reliable signal for those who understand structural leverage. The question is not whether Cadence is cheap. The question is whether you are willing to hold through the narrative lag. Floor prices bleed, but structure remains. The structure here is the permanent tax on the AI era.
Pivot not panic: The data reveals the path. The path is to buy the infrastructure that enables the infrastructure. Cadence is the pickaxe in the AI gold rush. The drillers get the headlines. The pickaxe seller gets the recurring revenue.
