Three years of building one of the largest domestic Bitcoin mining operations didn't win CleanSpark many fans. The market saw a commodity business - power costs, hash rate, block rewards - and valued it accordingly. What the market missed is that accumulating a gigawatt of contracted US power is extraordinarily difficult and extraordinarily valuable. You cannot buy it from Amazon. You cannot build it in six months. It takes a decade of relationships with utility commissions, land developers, and grid operators.
CleanSpark now owns that gigawatt. It also earned $766 million in FY2025 revenue at 55% gross margins, repurchased 11% of its own outstanding shares, holds over 13,000 Bitcoin on the balance sheet - mined, not purchased - and issued a $1.15 billion convertible at 0% interest to fund what comes next. What comes next is AI data centers.
The narrative is shifting. In late November 2025, a senior director of site selection from a global hyperscaler called the CEO the evening of the earnings call to confirm they were "still in the running" for the Sandersville, Georgia facility. That call was not from a neocloud. It was from a hyperscaler. The demand is institutional, the timeline is 2026–2027, and the re-rating - when it arrives - will not be gradual.
We're building a position at the inflection point. Downside is hard-floored by Bitcoin mining economics. Upside comes from 535 MW of AI-ready land the market is currently valuing at zero. That's an unusual risk profile for a company this profitable.
1+ Gigawatt of Contracted US Power
The asset that took a decade to build and cannot be replicated in the current grid environment
Bitcoin Mining
The cash engine that finances everything else. 50 EH/s of fully owned, 100% US-based infrastructure generating revenue at a 55% gross margin, with the world's most efficient immersion-cooled fleet.
- FY2025 Revenue$766 million
- Gross Margin55%
- Marginal Cost / BTC~$43,000
- BTC Treasury13,054 coins
- Hashrate50 EH/s
- Fleet Efficiency13.5 J/TH (S21 XP)
AI Data Centers
The optionality the market is currently pricing at zero. $0 AI revenue today - but two dedicated sites are being commercialised for hyperscaler tenants, with industry contracts benchmarking at $1–$1.5M per MW per year.
- Sandersville Capacity250 MW (AI-ready)
- Texas Capacity285 MW (ERCOT approved)
- First TX EnergizationH1 2027 (200+ MW)
- AI Head (Jeff Thomas)ex-President, Humain
- Modular PartnerSubmer (MOU signed)
- Revenue Potential (535 MW)$535M–$800M / yr
Bitcoin Profitability Across Three Market Scenarios
With a $43K cost basis, profitability holds down to $57K Bitcoin. That's a 37% further drop from here with mining economics still in the black - and somehow this doesn't show up in the consensus model.
Reasons to Act Now
Reasons to Wait / Risk Factors
CLSK trades at the cheapest earnings multiple in the peer group while building comparable or larger power capacity. That gap doesn't hold once a lease is signed. The re-rating from 11× to 30×+ has historically fired at the moment of a first hyperscaler contract - not gradually, but immediately.
| Company | FY26 P/E | P/Sales (NTM) | P/Book | AI Revenue | Power (MW) | Assessment |
|---|---|---|---|---|---|---|
| CleanSpark (CLSK) | 11.6× | 2.98× | 1.3× | $0 (building) | 1,000+ MW | Cheapest in group |
| IREN Limited (IREN) | 40× | 8×+ | 3.5× | Yes - signed | ~800 MW | Priced for AI pivot |
| Cipher Mining (CIFR) | 48× | 10×+ | 6×+ | Yes - signed | ~500 MW | Full AI premium |
| MARA Holdings (MARA) | 232× | 4× | 2.5× | No | ~700 MW | BTC treasury premium |
| Riot Platforms (RIOT) | Loss | 5× | 2× | No | ~700 MW | Unprofitable miner |
Our target range of $20–$24 assumes BTC holds above $85K and a signed AI tenant contract by H2 2026. Below are three paths with share price implications.
Bear Case
Crypto winter / AI hype collapses
Base Case
Stable macro, AI contract in 2026
Bull Case
BTC resumes + AI re-rating premium
Listed by share price impact, not chronology. The first signed AI lease is in a category by itself - it's the event that re-rated every comparable peer, and CLSK is still waiting for it.
19,000 S21 XP Immersion Miners Deployed
Adds ~6 EH/s to existing 50 EH/s fleet with industry-leading 13.5 J/TH efficiency. Deployment was deliberately delayed to preserve AI-applicable megawatts - deployment confirms AI-first capital discipline is real.
First Hyperscaler / AI Tenant LOI or Agreement (Sandersville)
The single most important near-term catalyst. CEO confirmed multiple inbound inquiries including direct hyperscaler engagement. A signed lease at Sandersville would validate 250 MW at $1–$1.5M/MW/year and trigger peer-style multiple expansion from 1.3× to 3–5× book.
Texas Lease Signed + Submer Factory Modules Ordered
Sealy, TX (285 MW, ERCOT approved) is being marketed simultaneously. A signed lease here - plus first modular orders from Submer - gives visibility on ~$400–500M of annual AI revenue from 2027. Behind-the-meter gas generation opportunity evaluation also expected.
New Land & Power Acquisitions (Multi-Gigawatt Pipeline)
Management explicitly stated they are hunting for additional power with remaining convertible proceeds. Each new GW acquisition extends the infrastructure moat and raises the long-term platform ceiling. Expect announcements in parallel with AI commercialization.
Texas First 200 MW Energized - AI Revenue Begins
ERCOT approval complete. This is the ultimate proof-of-concept milestone - the moment CleanSpark shifts from being valued as a miner to being valued as an infrastructure company. Sustainable recurring AI/HPC revenue changes the company's fundamental identity.
Texas 240 MW Tranches (×2) - Platform Scale
Two further 240 MW tranches scheduled for 2028 and 2029, bringing total Texas capacity to 725 MW. Combined with Sandersville and future acquisitions, this positions CLSK in the same tier as established neocloud operators. Potential for project-level debt financing at 80%+ LTV from 2027.
Bitcoin Price Dependency
100% of current revenue is tied to BTC price. A sustained bear market below $65K significantly impairs earnings, forces BTC sales from treasury, and likely delays AI capex spending. This is the dominant risk - do not size this position without a view on BTC.
AI Contract Timeline Risk
CEO acknowledged "it's hard to say" whether Sandersville tenant deals close in 2026. If lease agreements slip to 2027 or later, the re-rating thesis is delayed and capital burn accelerates. Contract announcements are binary catalysts - the absence of news is not neutral.
Build Cost & Execution Risk
AI/HPC infrastructure costs $10M per MW - 10× the cost of Bitcoin mining build-out. CLSK has never built an AI data center. First-project execution risk is real, and the Submer partnership, while promising, is unproven at scale domestically.
Dilution & Capital Structure
Share count grew ~50% from 2022–2025 before the 2025 buyback. If AI revenues take longer than expected and BTC weakens simultaneously, management may need to issue equity again. The $1.15B convertible at 27.5% premium matures in 2032 - but conversion risk is real above ~$15.
Q1 FY2026 Results Validate the Thesis
CleanSpark reported Q1 FY2026 revenue of $181 million - below the FY2025 annualised pace of $766 million, reflecting the post-halving reset in mining economics. Expected number. Not a warning sign. The metrics that matter are unchanged: roughly 55% gross margins, $43K/coin cost basis, 50+ EH/s fleet. Profitable across a wide range of Bitcoin price scenarios. Balance sheet strengthening. AI infrastructure timeline on track. Revenue comparisons across halving-distorted periods are noise. What matters is that the cost floor held, the power portfolio expanded, and the hyperscaler conversations are still live.
The core thesis is intact: BTC mining generates the cash, the land-and-power bank is being actively monetised for AI tenants. The first lease agreement remains the re-rating catalyst to watch.
The following scenarios reflect the author’s personal analysis and are not investment recommendations. See our full disclaimer.
CLSK - CleanSpark, Inc. (NASDAQ)
Stop discipline: A close below $8.50 would breach the 2024 demand zone and invalidate the higher-lows pattern, suggesting the BTC cycle has turned more bearish than our base case. Exit the position and re-assess.
Position sizing note: This is a high-volatility, BTC-correlated position. Size accordingly. The 52-week trading range has historically shown a 4× spread between low and high. Options may be a more capital-efficient vehicle for speculative exposure if equity volatility is unacceptable.
Disclaimer: This market tip is for educational and informational purposes only and does not constitute financial or investment advice. All investments carry risk. Past performance is not indicative of future results. Always conduct your own due diligence and consult a qualified financial advisor before making investment decisions. Bellwether Research is not a registered investment advisor.