Hook
Over the past 72 hours, a curious signal emerged from the semiconductor periphery. Cadence's CEO, Lip-Bu Tan, stood before a room of tech investors and declared his company undervalued. Not a plea. A diagnosis. He pointed to the AI boom, the explosion of chip design starts, and the fact that Cadence sits at the root of every GPU, every ASIC, every silicon brain. Yet the market still priced it like a legacy software vendor. I don't trust narratives that are too clean, but this one had a specific texture: a data point that refused to fit the prevailing story. The CEO's statement wasn't just about Cadence. It was a meta-narrative about how the market misprices "infrastructure" in any exponential boom. And crypto, I suspect, is about to repeat the same mistake.
Context
Cadence is not a household name. It is an Electronic Design Automation (EDA) company—the invisible hand that designs the tools to design chips. Without EDA, no modern semiconductor can be taped out. The industry is a duopoly: Synopsys (32-33% market share) and Cadence (30-31%) control the stack from RTL to GDSII. Together, they generate roughly $150-180 billion in annual EDA and IP revenue, but that number is a fraction of the $6 trillion semiconductor market they enable. The leverage is staggering: every $1 of EDA revenue supports $200-300 in chip production and $5,000-10,000 in downstream tech value. This is the "pick and shovel" of the AI gold rush. Yet the market applies a valuation multiple that belongs to a cyclical software consultancy, not a platform with a 90% gross margin and a 30-year moat. The CEO's undervaluation claim is a distillation of this disconnect.
In crypto, the same pattern recurs. Projects that provide the "tools" for the next wave—compute, data, verification, liquidity—are often dismissed as unsexy, low-growth, or commoditized. The narrative gravitates toward the end-user applications: the AI agent, the meme coin, the hyper-scalable L1. But the infrastructure layers that enable those applications are frequently undervalued until the market realizes the bottleneck. Cadence's story is a crystal ball for crypto infrastructure tokens.
Core Insight
The heart of the undervaluation lies in a narrative misconception. The market sees Cadence as a software company selling licenses to chip designers. But the business is shifting from a "tool license" model to a "platform tax" on AI silicon. Every major AI chip—from NVIDIA's H100 to Google's TPU to Amazon's Trainium—must pass through Cadence's design flow. The company's AI-EDA platform (Cadence.AI) is not just a tool; it's a machine that learns from every chip design, improving its own algorithms. That creates a data flywheel: more designs mean better optimization, which attracts more designs. This is a classic platform dynamics, but the market treats it as a static product.
I hunt for the story the data refuses to tell. Let's look at the numbers. Cadence's revenue for FY2024 is estimated at $46-48 billion, with a gross margin of 88-90% and an operating margin of 35-40%. Its R&D spending is 30% of revenue, aggressively embedding AI into its flows. Meanwhile, the total addressable market for EDA and system-level analysis is about $250-300 billion, with a 10-15% CAGR. The company is investing to capture that expansion, but the R&D expense depresses current earnings. The CEO's argument is that the market is discounting the future payoff of that investment. The hidden information is that the margin expansion once the platform matures could be explosive—imagine a software company with 90% gross margins and a 50%+ operating margin. That's the trajectory.
Now, map this to crypto. Consider a project like Akash Network, which provides decentralized compute for AI workloads. It's an infrastructure layer: the "pick and shovel" for AI model training and inference. The market currently values Akash at a fraction of the large AI model tokens. But as AI agent demand scales, the need for verifiable, censorship-resistant compute will explode. Similar to Cadence, Akash's value is not in the applications it runs but in the platform it provides. The narrative decay is that the market sees it as a "cloud alternative" rather than a foundational layer for autonomous economies. Another example is Chainlink, which provides data oracle infrastructure. It's the EDA for smart contracts—without it, no DeFi, no cross-chain, no AI-agent coordination. Yet its valuation is often tied to speculative narratives rather than the underlying utility of every transaction that depends on it.
The core of the Cadence story is the "superlineard leverage" of infrastructure. In crypto, the same leverage exists: a single piece of infrastructure can enable an entire ecosystem. But the market discounts it because the revenue is indirect (protocol fees, token emissions) rather than direct software subscription. The narrative is that these tokens are "unproductive" or "inflationary"—but they are the R&D expense of the network. The market is pricing the R&D cost as a liability, not as an investment in a platform moat.
Contrarian Angle
Here's where the Cadence narrative breaks down, and where the crypto lesson gets sharper. Cadence's moat is 30 years of design data, deep integration with foundries like TSMC, and switching costs that are measured in hundreds of millions of dollars. No crypto infrastructure project has that. The switching cost for a DeFi protocol to move from Chainlink to a competitor is a few days of code changes and some liquidity migration. The moat is not technical; it's network effects and liquidity aggregation. This is a weaker form of moat. So the undervaluation argument for crypto infrastructure is more fragile.
Moreover, the EDA industry is a duopoly with pricing power and no direct competition from China for at least a decade. Crypto infrastructure faces hyper-competition: there are a dozen decentralized compute networks, a dozen oracle networks, a dozen L1s. The "platform" dynamics are diluted by fragmentation. The Cadence narrative is valid only if the infrastructure project has achieved a clear network effect and a defensible position. Most haven't.
But the contrarian insight is that the market is too pessimistic about the few that have. Projects like Akash, with a 60%+ gross margin on compute (after token costs), and a growing base of AI customers, are starting to show the same characteristics: a platform that improves with usage, a data flywheel from job scheduling, and a lock-in effect from deployed workloads. The valuation gap is real, but it will close only when the narrative shifts from "compute commodity" to "AI infrastructure platform." The Cadence CEO's claim is a warning: the market is systematically mispricing the picks and shovels of the AI boom. In crypto, the same mispricing is happening, but the window to correct it is narrower because the moats are thinner.
Takeaway
Chaos is just a pattern you haven't decoded yet. The Cadence undervaluation is a pattern that will repeat in crypto. The next narrative will not be about the next AI agent token or the next hyper-scalable L1. It will be about the infrastructure that enables those agents to exist. The market will realize that the "tool tax" is the most predictable revenue stream in any exponential cycle. When that realization hits, the valuations of the few established crypto infrastructure platforms will expand. But the clock is ticking. The narrative decay will set in once the market understands that the tools are not just tools—they are the platform upon which the entire AI-crypto symbiosis is built. Decode the script before you bet on the actor.