CNBC interviews Cipher Digital CEO Tyler Page about how the company is pivoting from bitcoin mining toward AI/data-center infrastructure. Page argues the market is finally recognizing that sites once viewed as “tier three” can become valuable AI campuses because they have cheap power, land, fiber, and the ability to add generation faster than waiting for grid hookups.
Watch on YouTube ›Get the market thesis, key claims, assets, contradictions, and follow-up questions from any financial video — then unlock a version personalized to your portfolio, watchlist, and favorite speakers.
Tyler Page’s core message is that Cipher Digital’s old bitcoin-mining footprint has turned into an advantage in AI infrastructure. He says the company spent years building expertise and land in places with abundant power and low demand, and that those locations are now becoming attractive for large AI campuses as hyperscalers need 200–500 MW blocks of power at a single site. In his framing, what used to look like “tier three” geography is being repriced into “tier one” relevance because the bottleneck is no longer just proximity to traditional cloud hubs, but access to power, land, and usable transmission/fiber. He supports that thesis by pointing to several practical elements: Cipher’s deep land portfolio, existing fiber capacity along the I-20 corridor in Texas, and sub-10 millisecond latency to major Texas metros for training or inference use cases. …
Tactically, the setup is a re-rating story tied to AI power scarcity and any evidence that Cipher can turn land and power access into signed projects. Near-term upside likely depends on continued investor focus on data-center power bottlenecks, while the main risk is over-anticipation before permits and deals materialize.
Over the next few months, the base case is that Cipher benefits if AI demand keeps outpacing grid capacity and the company proves it can move faster through direct generation or pipeline-linked power. The thesis weakens if project timelines slip or if the market decides remote campuses cannot capture enough of the AI buildout.
The structural thesis is that AI compute is becoming an energy-location business: whoever controls power, land, and network access can capture value even outside traditional cloud hubs. If this regime persists, bitcoin-mining infrastructure owners may remain important long-duration beneficiaries of AI capex.
Cipher has an extraordinarily robust pipeline for data-center opportunities.
Directly stated in response to whether the pipeline is robust.
Locations once considered tier three can become tier one for AI data centers because AI needs more power and land.
He argues siting economics have changed due to AI’s resource intensity.
Hyperscalers now want single campuses of 200 to 500 megawatts, which makes power-rich non-incumbent locations more attractive.
This is the quantitative demand argument behind the siting thesis.
You have visibility into your pipeline. Is that pipeline robust?
Tyler says the pipeline is extraordinarily robust. He explains that as a Bitcoin miner, they developed expertise in finding locations suited for large power interconnects. With AI growth and demand for larger campuses, what were traditionally tier-three locations are now potentially tier-one locations for AI data centers. They have a very deep land portfolio.
What did you see that got that right to go from tier three to tier one? It's like finding out a Walmart is going in next door and the land value goes up — that's what you're talking about, right?
Tyler explains that to make Bitcoin mining work you need cheap power — places with abundant generation and not much demand. He says they saw the explosive growth in AI with a completely convex adoption curve, and people wanting 200-500MW at a single campus. They asked where you'd find that in Northern Virginia — there's not enough power there — and concluded the only path was to go to places considered tier three. The whole market is starting to realize that may have been backwards.
Is there any latency issue? The farther away these data centers are, does that affect how quickly AI answers come back?
Tyler says the initial pushback from incumbents was about latency. His counter was that a massive fiber line runs east to west across Texas under I-20, and data travels at the speed of light. For training or inference, they're sub-ten milliseconds to the major metros in Texas — not close enough for traditional cloud services requiring really low latency, but sufficient for AI workloads.
Unlock the full claims, asset map, scores, related transcripts, follow-up questions, and AI chat — shaped around your portfolio, watchlist, favorite speakers, and risks.
