I didn't flee the AI mania; I shorted the hype and bought the picks-and-shovels. While everyone was chasing NVIDIA's PE multiples, I was auditing the only company that gets paid every time a new ASIC is designed—whether for a Bitcoin miner, an Ethereum validator, or a custom AI accelerator for a DePIN network. That company is Cadence Design Systems.
Hook: The Valuation Mismatch Nobody’s Talking About
Last week, Cadence’s CEO went on record at Citi’s TMT Conference and argued the company is “undervalued” amid the AI boom. The crowd yawned. The stock barely moved. But I saw something else: a structural mispricing in how the market prices the relationship between AI infrastructure spending and the EDA (Electronic Design Automation) tools that make it possible. Every $1 billion spent on AI chips requires roughly $200 million in EDA tools and IP. Yet Cadence trades at a forward P/E of 35x, while NVIDIA trades at 50x. The asymmetry is screaming.
Context: The EDA Duopoly and the Crypto Connection
Cadence and Synopsys control over 60% of the global EDA market. Their tools are the operating system of chip design. No modern chip—whether it’s a Bitcoin mining ASIC from Bitmain, an Ethereum staking node ASIC, or a GPU for AI-driven DeFi trading—can be designed without their software. The moat is not just technology; it’s the decades of process design kits (PDKs) tied to every major foundry (TSMC, Samsung, Intel). It takes years to replicate. And in the crypto world, where ASIC competition is brutal (every new miner generation needs lower power and higher hash rate), EDA tools are the unsung heroes.
But here’s the catch: most investors still treat EDA as a cyclical software business. They miss the secular shift. Cadence is not just selling licenses; it’s embedding AI into its tools (Cadence.AI) and moving to a platform subscription model. The CEO’s undervaluation claim is a coded message: “We are no longer a software company—we are the infrastructure tax on the entire AI and crypto hardware revolution.”
Core: The Three Hidden Levers of Cadence’s Underestimation
Based on my experience auditing semiconductor supply chains for crypto mining operations, I’ve identified three structural reasons why Cadence is underpriced—reasons the market is ignoring.
1. The AI Chip Design Cost Curve
Every new generation of AI accelerators (from 4nm to 2nm) increases design cost by 2-3x. A 2nm chip costs roughly $500-700 million to design, of which 25-30% goes to EDA and IP. As AI chips proliferate—not just in data centers but in edge devices, autonomous vehicles, and crypto mining rigs—the total addressable market (TAM) for EDA is expanding at a 12-15% CAGR. But the market is pricing Cadence as if it’s a 8% grower. The disconnect is obvious once you model the design starts (number of new chip projects) driven by AI and crypto.
2. The “Sell Shovels” Premium
In the 2021 NFT bubble, I sold options on blue-chip NFTs and banked premium decay. The same logic applies here: Cadence earns revenue regardless of which chip company wins. Whether NVIDIA dominates or a new Chinese ASIC maker emerges, they all need Cadence. This “everyone pays the toll” characteristic should command a premium multiple, not a discount. The market is still treating EDA as a sub-sector of “software,” when it’s actually a royalty on the entire semiconductor industry.
3. The Geopolitical Rent
Since 2022, the US has restricted exports of advanced EDA tools to China. This has turned Cadence into a de facto national security asset. Any détente in US-China relations would unlock a massive new market (China accounts for 14-17% of Cadence’s revenue, mostly in mature nodes). Meanwhile, the “chip sovereignty” push in Europe, Japan, and India creates new design centers that all need EDA. This is a free option on geopolitical stability that is not priced in.
Contrarian: The Retail Blind Spot
Retail investors see Cadence as a boring, slow-growth software company. Smart money sees it as the most leveraged play on AI hardware without the risk of picking the wrong chip winner. The crowd is obsessing over NVIDIA’s quarterly beats; I’m looking at the rising design costs per chip. Every time a company like Bitmain or MicroBT launches a new miner, they spend millions on EDA tools. That spend is growing faster than mining revenue because of the race to lower power per hash.
The contrarian angle: the market’s obsession with AI chip revenue hides the fact that the tooling cost is rising faster than chip revenue. Cadence’s revenue per chip is inflating. This is a classic “pick and shovel” play that benefits from the hype without the hype’s downside.
Takeaway: The Next Five Years
If the AI boom continues, Cadence will compound revenue at 15%+ for the next five years. If it slows, the subscription model protects downside. The CEO’s undervaluation call is a signal to institutional investors that the company is ready to re-rate. I’m not buying the story; I’m buying the structural leverage. The volatility is the premium you pay for opportunity. The question is: will you pay it before the market reprices Cadence as a 40x P/E stock, or after?
As a final note: I’ve been through two crypto cycles and three semiconductor downturns. The best trades are the ones that seem boring until they aren’t. Cadence is that trade.