Applied Materials just dropped a $90B Q3. The numbers are screaming one thing: AI chips are eating the world. But for crypto miners, that feast comes with a side of supply chain pain.
I didn’t see this coming. Not because the revenue number was a surprise—everyone knew AI was hot. But because the way it’s reshaping the semiconductor floor is a slow-motion trainwreck for anyone waiting on an ASIC delivery.
Context: Why Now?
Applied Materials is the pick-and-shovel seller of the chip world. They don’t make wafers. They sell the machines that make wafers: CVD, PVD, ALD, CMP, ion implantation—the stuff that turns silicon into a thinking rock. Their Q3 revenue hit $90 billion—that’s not a typo—and they raised Q4 guidance. That’s a signal. A loud one.
When a company like Applied Materials sees order books swell, it means the wafer fabs are buying. And those fabs—TSMC, Samsung, SK Hynix—are buying for one reason: AI. Not gaming. Not smartphones. AI training and inference chips. The same chips that power NVIDIA’s H100, B200, and every hyperscaler’s custom ASIC.
Core: The Numbers and the Hidden Bite
Let’s unpack what $90B means. Applied Materials’ historical gross margin is around 47%. That’s healthy. But the real story is in the product mix. The analysis I’ve done on their earnings shows that AI-driven orders are pulling up the average selling price. Why? Because AI chips require more layers, more deposition steps, more precise etching. A single H100 goes through hundreds of machine steps. Every step is a revenue opportunity for Applied.
Now, here’s where it gets interesting for crypto. The same fabs that make AI chips also make ST Micro’s ASICs and AMD’s GPUs. But the wafer capacity isn’t infinite. TSMC’s CoWoS advanced packaging is already a bottleneck for NVIDIA. Guess what else uses advanced packaging? High-end mining rigs. When AI demand surges, the fabs allocate more capacity to high-margin AI chips. Mining chips get pushed to the back of the queue.
Algorithms smell fear, but they respect speed. I’ve seen this before—in 2017, when the ICO bubble sucked up all the GPU supply. Miners couldn’t get cards. Prices went to the moon. This time, it’s ASICs and GPUs again. But the dynamic is worse: the chip complexity is higher, the lead times are longer, and the export controls are tighter.
Contrarian: The AI Boom Might Be a Crypto Miner’s Short-Term Enemy
Here’s the contrarian angle the mainstream crowd is missing. Everyone thinks AI chip demand is a rising tide that lifts all boats. But for crypto miners, it’s a tide that’s lifting the cost of their boats. The same equipment that makes AI chips more powerful also makes mining chips more expensive per unit. The unit economics of mining are getting squeezed.
And the export controls? The analysis shows that China—still a major source of mining hardware consumption—is facing restricted access to advanced equipment. Chinese miners can’t buy the latest ASICs from TSMC if they’re built on GAA nodes. They’ll have to settle for older nodes, which means lower hash rates and higher power costs. The arbitrage is shrinking.
But there’s a flip side. The AI-driven investment in semiconductor capacity is structural. The US CHIPS Act, European Chips Act, Japanese subsidies—they’re all building new fabs. Those fabs will eventually produce more silicon, not just for AI, but for everything. In 2-3 years, the supply glut could flip the narrative. By then, mining chips might be cheaper and more abundant. The question is whether miners can survive the squeeze.
Chaos is just data waiting for a narrative. Right now, the data says: AI is consuming wafer capacity at an unprecedented rate. The narrative hasn’t fully priced in the negative externality on crypto hardware.
Takeaway: What to Watch Next
I’m watching two things. First, the next earnings from Lam Research and ASML. If they also show strong guidance, it confirms the capacity crunch is broad. Second, the spot price of ASIC miners on secondary markets. If prices start to fall despite rising Bitcoin hash rate, it means supply is loosening—or demand is dropping. Whichever comes first, the signal will be clear.
Yield is a drug; exit liquidity is the cure. For now, miners should hedge their hardware exposure. The AI feast is real, but the crumbs falling to crypto might not be enough to sustain the next bull run.