I haven't heard much from Rob Hamilton (AnchorWatch) lately. There was a stretch where he was everywhere, in every conversation. Then around December last year -- I stopped seeing his posts. Maybe the X algo turned against him.
Anyhow, he had chat put this summary out about the Nashville Energy & Mining Summit 2026. I sliced it down because it reeked of slop a vacuous statements. The later points are better than the earlier ones.
1. Bitcoin ASICs now compete for the same upstream inputs as AI silicon1. Bitcoin ASICs now compete for the same upstream inputs as AI silicon
ASIC vendors aren't operating in a "miners only" lane anymore. They're fighting for the same scarce stuff as the companies pushing AI chips
2. Energy is the binding constraint now, not just an operating line item2. Energy is the binding constraint now, not just an operating line item
The winners aren't just the ones who can buy chips. They're the ones who can energize capacity and actually run it.
3. Difficulty can't compound forever if machines and power both get tight3. Difficulty can't compound forever if machines and power both get tight
Bitcoin difficulty has been compounding at something like 40–50% annually for years now.
I'm not predicting an imminent decline. I'm just pointing out that exponential curves need real world inputs, and those inputs have lead times that don't bend to anyone's roadmap.
4. The network may already be over secured on raw hashrate4. The network may already be over secured on raw hashrate
If hashrate growth slows or plateaus, that's not automatically a security crisis.
It might just be the subsidy finishing the job it was designed to do, and the network settling into a steadier regime where the interesting question isn't how much hashrate exists, but where it sits and who controls it.
There's no competing industrial use for SHA-256 that forces a permanent escalation race. No nation state is secretly stockpiling SHA-256 ASICs for some other purpose.
5. Miner strategy will matter more than miner hashrate5. Miner strategy will matter more than miner hashrate
My base case isn't "miners get replaced." It's "miners get recruited."
If you control land, interconnect, and operations, you can monetize electrons in more than one format. Bitcoin becomes the flexible floor load. HPC becomes the premium tenant when you can deliver the full product.
Bitcoin mining is a single global contest. Every 2,016 blocks, difficulty adjusts to keep the network near a 10 minute block time. When new hashrate shows up anywhere, everyone feels it. Your new machine doesn't just help you, it pressures everyone else.
AI doesn't work like that.
Your inference workload doesn't directly compress someone else's GPU economics on the other side of the world. AI isn't a global difficulty function. The value of a GPU is local to the workload and the contract. That means older hardware can stay useful longer than old gen Bitcoin ASICs ever could, even when the frontier moves fast.
Some [bitcoin miners] will become the on ramps, because they already figured out how to industrialize power.
Not an expert in either, but been reading about NPU's lately (a more energy efficient way to do AI inference).
What struck me was that, NPUs are to GPUs as ASICs were to GPUs....that is to say in AI inference, the GPU is more of the "general purpose" chip, while an NPU is specialized just for inference.
That means that you really can't use NPUs for anything else, and also you cant use them for things like training.....but the vast majority of the worlds compute need for AI is for inference (powering your ChatGPT chat session)....
This gets me to think, could a NPU be expanded to support SHA256 hashing as well? That is what if the same "miner" could simultaneously mine Bitcoin and/or do AI inference?
A couple years ago I spoke to a former Broadcom designer that for lulz developed a PCIe board that he built in 2 versions: SHA256d and a bespoke tensor unit for inference. His thinking was that he could just put sets of both in a container and during NPU demand downtime, mine Bitcoin as grid booking permitted.
I don't know how much inference demand downtime there is right now, but judging by the "batch" APIs being promoted for a while now, there must be non-constant demand.
I guess in an ideal world it would be like multi-tasking. Moment to moment context switching (maybe not at the microsecond level, but maybe at the seconds level). So run an inference request, then guess a few hundred million hashes, then switch back to inference....goal would be to keep NPU constantly occupied
Exactly.
I like this sentence. Short but impactful sentiment
It will be interesting to see if there will be a future in which governments ditch their CBDCs and actively recruit Bitcoin miners to work for their benefit.
https://twiiit.com/Rob1Ham/status/2014763148717277600