Share article

Bitcoin$62,716.03 miners are getting a fresh narrative bid after VanEck argued they are "sitting on a gold mine": not in BTC, but in power. With AI data centres scrambling for megawatts and grid access, mining sites are starting to look like ready-made real estate for high performance compute (HPC). [1]
US Bitcoin ETFs pull in $167M as altcoin ETP outflows deepen
Bitcoin$62,716.03 itself is trading around $70,120 on Saturday (up roughly 0.93% on the day), putting its market capitalisation at about $1.3 trillion using circulating supply maths. The interesting part is that miners now have two potential revenue lines: hashing for block rewards, and leasing electrons (and infrastructure) to AI.

Enjoy articles without ads?

Register for free and get unlimited access to all articles.

Why AI is suddenly bidding for miners' turf

VanEck's core point is simple: the bottleneck for AI is not only GPUs, it is power delivery. Anyone can order racks, fewer can get the grid to sign off on large, near-term capacity. [2]
Bitcoin$62,716.03 miners already did the unsexy work years ago:
  • Grid interconnections and substations built for tens to hundreds of megawatts
  • Land, permits, and power purchase relationships
  • Operational muscle running industrial load 24/7, plus curtailment programmes in some regions

That stack is precisely what hyperscalers and AI compute firms are tripping over. New data centre builds can spend years stuck in queues for interconnection studies, transmission upgrades, and local approvals. A mining site with spare capacity, or a pathway to expand it, is a shortcut.

VanEck's "gold mine" framing is less about miners becoming Nvidia overnight and more about miners monetising something they already own: the right to draw power at scale. [3]

Meta Reportedly Acquires AI-Agent Social Network Moltbook as Unofficial MOLT Token Spikes 258%

The mining business post-halving: power arbitrage matters more

By March 2026, the Bitcoin block subsidy is 3.125 BTC per block (post-2024 halving), so miners live and die on efficiency, uptime, and the BTC price. At $70,120, the subsidy component alone is roughly $219,000 per block, before fees. Multiply that by 144 blocks a day and you get why miners still care about BTC price, even if fees swing.

But halving cycles also make the miner playbook more ruthless:

  • Top operators hunt cheaper energy, better machines, and better financing.
  • Marginal operators get forced into selling coins, shutting down, or consolidating.
  • Everyone looks for non-hash revenue that is less correlated to difficulty and subsidy cuts.

AI colocation and HPC hosting is attractive because it can, in theory, convert a miner from a pure commodity producer into something closer to an infrastructure landlord. That is what investors tend to pay up for, when it is real. [4]

What "pivoting to AI" actually means (and what it does not)

CT (crypto Twitter) tends to hear "AI" and immediately ape into anything with a datacentre photo and a ticker. Reality is more granular.

The bullish version

A miner signs long duration hosting contracts, funds capex to retrofit or expand, and turns energy and interconnects into stable cash flow. Some public miners have already tested this model via HPC partnerships, and the market has clearly been willing to re-rate credible announcements.

The best positioned sites typically have:

  • High quality power (stable, scalable, contracted sensibly)
  • Fibre connectivity (or a realistic path to it)
  • Cooling and building envelope suitable for high density racks
  • Room to expand without a planning war

The awkward version

A lot of mining sites were built for ASICs, not GPUs. That matters. HPC wants consistent uptime, tight SLAs, and redundancy. Many miners built their advantage on interruptibility (curtail when power prices spike, then spin back up). That is brilliant for grid services and mining margins, but it is not automatically compatible with AI clients who pay for always-on compute.
Retrofitting can be capital intensive. So can the boring bits: transformers, switchgear, backup generation, chillers, and security standards that enterprise clients require. If a miner has to issue equity at ugly prices to fund the rebuild, shareholders can still get diluted into oblivion even if the "AI" headline pumps the stock.
Morning Crypto Report: Ripple's Jed McCaleb Joins Forbes Rich List, SHIB Tests $3.5B Mark, MARA Starts Selling Bitcoin

On-chain reality check: miners still follow BTC

Even with the AI narrative gaining traction, miner fundamentals are still chained to network economics.
A few on-chain and network indicators to keep in view while this theme plays out:
  • Hashrate and difficulty: if hashrate keeps climbing, competition rises, and "AI pivot" stories start to look like a scramble for survival rather than strategic expansion.
  • Fee environment: fee spikes can rescue weaker miners temporarily, but fees have historically been bursty. Sustainable fee markets matter more than one-off crazes.
  • Miner balance behaviour: heavy exchange inflows from miner-tagged wallets often precede sell pressure. Steadier reserve behaviour suggests operators are not being forced to dump into the bid.

None of that kills VanEck's thesis, it just keeps it grounded. A miner can be right about AI demand and still get wrecked by a bad debt stack or a fleet that is too inefficient for the next difficulty leg up.

What the market is likely to price next

If VanEck is right, the trade is not "miners become AI companies." The trade is that the market starts valuing certain miners as power and interconnection assets with optionality. [5]

Watch for these signals, because they separate proper deals from vibes:

  1. Signed contracts with named counterparties (not "non-binding LOIs")
  2. Clear capex plans and timelines (how many MW, by when, at what cost)
  3. Unit economics (revenue per MW, margin expectations, who pays for upgrades)
  4. Balance sheet survivability (can they fund it without constant dilution)
  5. Operational fit (fibre, cooling, redundancy, and regulatory posture)

If disclosures stay fuzzy, assume the "AI" tag is marketing. That is where things get dodgy fast, especially in small caps looking for a new story after a tough halving year.

Arthur Hayes Hits the Brakes on Bitcoin: 'I Wouldn't Bet $1 Right Now'

Risks and invalidation (read this bit twice)

Key risk: AI data centres do not just need cheap power, they need reliable power, connectivity, and enterprise-grade infrastructure. Many mining sites are not plug-and-play for that.

What would invalidate the bull case:

  • AI customers refuse to sign meaningful long term contracts at attractive rates, or demand terms that make the economics mediocre after retrofit costs.
  • Miners fund pivots with excessive dilution or expensive debt, turning "new revenue" into "new bagholders."
  • Grid constraints tighten further (or policy shifts), limiting expansion even at existing sites.
  • BTC network competition rises faster than miners can diversify, keeping them trapped in low-margin hash wars.

VanEck's point stands: miners have a head start on the hardest part of data centres, power. The only question that matters for investors is whether that head start turns into contracted cash flows, or just another narrative pump that fades once the capex bill lands.

Tags

#AI data centers#Bitcoin miners#High-performance compute#Power infrastructure#VanEck#Grid interconnection#Mining revenue#Energy demand