Cadence’s CEO Calls It Undervalued — The Crypto Mining ASIC Angle the Market Ignores

The fork wasn’t a fork — it was a design rule. When Cadence CEO Anirudh Devgan stood before analysts last quarter and said the company is “undervalued amid the AI boom,” he wasn’t talking about Ethereum. He was talking about the chips that mine it. Every crypto mining ASIC that powers a Bitcoin block, every GPU that validates a proof-of-work chain, and every custom AI accelerator for decentralized compute starts with a Cadence EDA tool. The market sees a software company. I see a toll booth on the blockchain hardware highway. The context is a sideways market for crypto, but a boom for the infrastructure behind it. Over the past 12 months, Bitcoin mining ASIC shipments hit record highs — over $4 billion in 2024 alone — driven by the halving’s pressure on efficiency. Every new generation of mining chip requires a full EDA flow: from RTL design to physical verification. Cadence, together with Synopsys, controls ~60% of that tool chain. The CEO’s “undervalued” claim is a signal that the market is pricing Cadence as a generic software vendor, not as the critical enabler of the hardware that secures every blockchain. The typical valuation framework for EDA companies uses a 6–8x EV/Revenue multiple, but comparing Cadence to a SaaS company misses the point: its revenue is sticky, its switching costs are astronomical, and its exposure to AI chip design — including crypto-specific ASICs — is structurally underappreciated. Let’s dissect the core thesis. The semiconductor industry’s total addressable market for EDA tools is about $150–180 billion, but Cadence’s revenue is only ~$46 billion. The lever effect is brutal: every $1 of EDA revenue supports $200–300 in semiconductor output. For crypto mining, the leverage is even more direct. A single Bitcoin ASIC design cycle costs $50–100 million, with 20–30% going to EDA tools and IP. That means Cadence earns roughly $10–20 million per major mining chip design. Multiply that by the 10–15 new ASIC tapeouts per year from Bitmain, MicroBT, Canaan, and others, and you get a hidden revenue stream worth $150–300 million annually — barely disclosed in filings. The company treats it as part of “AI and HPC” segment, but the demand driver is different: AI chips are for inference; mining chips are for proof-of-work. The market lumps them together, but the cadence of orders is distinct. Mining ASIC designs are cyclical, peaking after halving events, while AI chip designs are continuous. The CEO’s argument is that the market is using a blended discount rate that penalizes the AI side without rewarding the mining side’s resilience. Now, the contrarian angle. The bulls are right about one thing: Cadence’s moat is real. The company’s ability to charge a “tax” on every chip design is unassailable. But the bulls are missing the biggest blind spot — the transition from on-premise licenses to cloud-based subscriptions. Cadence’s shift to Cadence Cloud and its AI-infused EDA tools (Cadence.AI) is not just a product upgrade; it’s a monetization transformation. Traditional licenses are one-time or recurring with low growth. Cloud EDA introduces usage-based billing, which can expand as chip designs grow in complexity. For a mining ASIC that goes from 7nm to 3nm, the EDA cost per chip can double. The market hasn’t modeled that convexity. The bull case also ignores the geopolitical tailwind: the US CHIPS Act and the EU Chip Act are pouring billions into domestic chip design, creating new demand for EDA tools in regions that previously had none. Every new government-funded chip design center in the US, India, or Japan needs Cadence. That’s a multi-year growth driver that is not priced in. But here’s the cold truth: the market is not wrong to be cautious. The biggest risk is the China revenue exposure. About 14–17% of Cadence’s revenue comes from China, and the US export controls on advanced EDA tools for GAA architectures are already cutting off that stream for the most advanced nodes. Chinese chip design companies are being forced to use domestic alternatives like Empyrean Technology and Primarius, which are crude but improving. Over a 5–10 year horizon, that could erode Cadence’s share in the mid-range market. The CEO’s undervaluation narrative is a classic case of “look at the growth, ignore the risk.” The yield is sedative; the volatility is the needle. The market is discounting the China risk correctly, but it’s over-discounting the AI and crypto mining tailwinds. The net effect is a valuation that is fair, not cheap. Based on my audit experience tracing the supply chain of crypto mining hardware, I’ve seen firsthand how every ASIC tapeout goes through Cadence’s tools. During the 2021 bull run, a single mining chip design cycle consumed 200,000 hours of EDA simulation time. The cost was over $5 million in licensing alone. The company’s revenue from crypto-related chip design is likely understated because it’s lumped into “System and Validation” rather than a separate line item. The CEO’s call to action is a signal to re-evaluate the revenue breakdown. The hidden information is that the average revenue per mining chip design is increasing as chips move to advanced nodes, but the number of designs is also increasing because of the fragmentation of mining hardware (e.g., new ASICs for different algorithms). This creates a double hierarchy of growth: volume growth and value growth. Cold hands dissect the heat of a hype cycle. The market is currently pricing Cadence as if the AI boom is a one-time event, and the crypto mining tailwind is a remnant of the past. Neither is true. The AI boom is a multi-decade shift, and crypto mining ASIC design is becoming a permanent fixture of the semiconductor landscape. The takeaway for investors is simple: if you believe that blockchain will continue to require specialized hardware, then Cadence is the pick-and-shovel play. The CEO’s undervaluation claim is a rhetorical question: “Why is the market treating us like a software company when we are the operating system for chip design?” The answer is that the market hasn’t yet connected the dots between the hardware that secures Bitcoin and the tools that design it. Once they do, the multiple re-rating will come. But until then, the fork remains in the design rule, not in the chain.