Pulse on the chain, breath in the market.
Goldman Sachs just lit a fuse under Japanese semiconductor equipment stocks. Lasertec, Tokyo Electron, Disco — three names now flashing green on every institutional screen. Intel is upping its 2026 capital expenditure by $3 billion. The logic is clean: more advanced chip fabs mean more orders for Japan’s precision machines.
But I’ve spent sixteen years watching capital flows in this space — seven of them running 24/7 market surveillance in Lisbon. And what I see is a thread that connects Intel’s fab floors directly to the Bitcoin hashrate.
Context: Why Now?
Intel’s IDM 2.0 strategy is a $200B+ bet on bringing advanced manufacturing back to US soil. The company is racing to commercialize 18A and 14A nodes using RibbonFET and PowerVia. That means High-NA EUV lithography from ASML — and the measurement, etch, and dicing gear from Japan’s big three. Lasertec dominates EUV mask inspection (~85% market share). Tokyo Electron is the king of coat/develop and a top-2 player in etch/dep. Disco owns 50-80% of the precision cutting market, critical for chiplet packaging like Intel’s EMIB-T.
Goldman sees this as a straight-line opportunity: more Intel capex → more Japanese equipment orders → stock upside. They set price targets implying 30%+ upside for Lasertec and Disco, moderate gains for TEL. Their thesis hinges on AI demand and geopolitical reshoring.
Core: The Crypto Connection Most Analysts Miss
Here’s the blind spot: the same equipment that builds Intel’s AI chips also builds the ASICs that mine Bitcoin. The lithography, etching, and packaging tools from TEL and Disco are used by TSMC and Samsung to produce Bitmain’s Antminer S21 and MicroBT’s M60 series. Intel’s own Blockscale ASIC line? Killed last year. But the supply chain doesn’t care about brand — it cares about equipment availability.
Fact: every new generation of mining ASIC requires finer node geometries (5nm, 3nm) and advanced packaging to stack memory and logic. Those processes demand exactly the tools Lasertec and Disco make. Disco’s grind and cut machines are the only game in town for thinning wafers to 10μm — essential for 3D NAND and high-bandwidth memory used in mining rigs.
Goldman’s $3B capex bump for Intel means those tools are suddenly harder to get. Foundries like TSMC and Samsung will face tighter supply for the same gear. That drives up lead times and costs for ASIC manufacturers. I’ve seen this pattern before — during the 2021 chip shortage, mining rig delivery delays stretched to 12 months. The same dynamic is quietly setting up.
Seventy-two hours without sleep, zero doubts.
Contrarian: The Hidden Risk — Intel’s Execution Could Crush the Thesis
The Street is euphoric about Intel’s roadmap. But I’ve audited enough supply chain data to know that Intel has never delivered a leading-edge node on time. The 18A is supposed to ramp in 2025. History says add 12-18 months. Even Goldman admits Intel’s 14A won’t be a volume node until 2027.
Here’s the contrarian twist: if Intel stumbles — if its yields lag TSMC’s by 10% or more — that $3B capex gets slashed. Equipment orders get canceled. And the ripple hits both AI chip production and ASIC manufacturing. The very stock that Goldman is bullish on becomes a trap.
Worse: geopolitical pressure. The CHIPS Act gives the US Commerce Department leverage to demand that Intel buy American gear from Applied Materials and Lam Research. If that happens, Japanese equipment makers lose share. Lasertec’s monopoly in EUV mask inspection can’t be easily replaced — but TEL’s etch tools face direct competition. The upside for Disco remains strongest because chiplet packaging is a structural trend independent of Intel’s success.
But the market isn’t pricing any of this. Goldman’s report frames the trade as a sure thing. It’s not. It’s a binary bet on Intel’s ability to execute — and on Washington not favoring US equipment makers.
Takeaway: What to Watch Next
For anyone holding mining stocks or ASIC derivatives, the signal is clear: track Intel’s 18A yield data like a hawk. If the company announces a major customer (Apple, Nvidia) for its foundry services, the capex stays high and equipment bottlenecks tighten further. If Intel pushes 18A back again — watch out. The hashrate growth could slow as mining rig supply contracts.
The only pure play here is Disco. Chiplet packaging is the future — for AI, for HBM, for mining ASICs. The others? Priced for perfection in an imperfect world.
