TeraWulf is a digital infrastructure company that mines Bitcoin and is currently expanding into high-performance computing (HPC) for AI. It generated $27.1 million in revenue in the most recent quarter, up 43% from a year ago, while scaling its total hashrate capacity to 10.0 exahashes per second. The company is in a transition phase, moving from a pure-play Bitcoin miner to a provider of power and space for energy-hungry AI data centers.
The investment thesis on TeraWulf is that its true value lies in its 750 megawatts of zero-carbon power capacity, which is worth significantly more as a data center than as a Bitcoin mining rig. While mining is volatile and subject to the Bitcoin halving cycle, AI hosting provides stable, long-term contracts with higher margins. If management can successfully convert its existing power infrastructure at Lake Mariner into Tier 3 data centers, the business shifts from a commodity miner to a critical infrastructure provider.
We lean cautious on TeraWulf because the current stock price already assumes a near-flawless execution of the AI pivot, leaving little room for error. The business is doing the right things by selling its minority stakes to focus on its own infrastructure, but the stock has moved ahead of the actual cash flow from these new data centers.
TeraWulf's stock stayed flat for a while but has recently soared to new highs. The company used to focus only on mining Bitcoin, but it is now shifting toward building massive power plants for the energy-hungry computers that run artificial intelligence. Investors are excited because those power sites are becoming much more valuable for AI than for mining.
What does it do?
TeraWulf is a growth business that earns money by mining Bitcoin and renting out its zero-carbon power infrastructure to high-performance computing (HPC) firms. The company develops and operates large-scale facilities that take low-cost, carbon-neutral energy (like nuclear and hydro) and turn it into computing power. In its mining business, it earns Bitcoin directly by solving complex mathematical problems and then sells that Bitcoin for cash. In its newer HPC business, it functions more like a landlord, providing the specialized power, cooling, and space that AI companies need to run their heavy-duty graphics chips.
Where does revenue come from?
The vast majority of revenue today comes from Bitcoin self-mining, but management is aggressively shifting the mix toward HPC hosting. Currently, Bitcoin mining accounts for nearly all sales, driven by the total amount of computing power (hashrate) the company can deploy and the current price of Bitcoin. Geographically, all revenue is generated from its two primary domestic sites: Lake Mariner in New York and the Nautilus facility in Pennsylvania. After the quarter ended, the company sold its interest in the Nautilus joint venture to focus entirely on its owned Lake Mariner site.
Who are its customers?
TeraWulf serves the global Bitcoin network as its primary customer today, but its future relies on attracting large-scale AI and enterprise cloud providers. In its mining business, its "customer" is the decentralized Bitcoin blockchain, which pays the company in newly minted Bitcoin and transaction fees. As it expands into HPC hosting, its customers will be firms needing specialized data center space for AI training and complex simulations. In the most recent quarter, the company operated at a capacity of 10.0 exahashes per second and self-mined 555 Bitcoin, highlighting its scale as one of the larger publicly traded miners in the United States.
What gives it staying power?
TeraWulf's durability comes from its exclusive rights to 750 megawatts of zero-carbon power capacity at its Lake Mariner site. Power is the single biggest cost in digital mining and AI, and having long-term access to clean energy at scale creates a hurdle for competitors. This infrastructure is difficult to replicate because of the time and permits required to secure such large power loads.
Where is it headed?
The company is making its biggest strategic bet on building out Tier 3 data centers for AI hosting at Lake Mariner. Management is currently constructing CB-1 and CB-2, which are dedicated HPC facilities expected to come online by the first and second quarters of 2025. This shift is intended to move the company away from the boom-and-bust cycle of Bitcoin and toward the steady, recurring cash flow of the data center industry.
The business is growing revenue quickly but facing higher costs following the recent Bitcoin halving. Revenue reached $27.1 million in the most recent quarter, a 43% increase over the prior year, mostly because the price of Bitcoin rose enough to offset a lower number of coins mined. However, the mining business is becoming more expensive, with power costs per Bitcoin jumping from $9,322 to $30,448 year-over-year.
Cash generation is currently being redirected into massive infrastructure investments for the AI pivot. While the company generated $27.1 million in revenue, it is spending heavily on constructing new data center facilities at Lake Mariner. This high CapEx is necessary to transform the business model, but it means the company is not yet producing consistent free cash flow for shareholders.
The balance sheet was recently bolstered by a massive $500 million capital raise to fund the expansion. TeraWulf holds a significant cash position following a convertible note offering, which gives it the runway to build its first 70 megawatts of HPC capacity without needing immediate mining profits. This "war chest" is the company's primary defense against the volatility of the crypto market.
TeraWulf is a high-growth infrastructure play with a balance sheet that now matches its aggressive expansion plans.
The company grew its hashrate capacity to 10.0 exahashes per second, a significant scale milestone for its mining operations. This increase in capacity allowed TeraWulf to capture more Bitcoin value during a period where the coin's price rose by 117%.
Power costs per Bitcoin have tripled over the last year, which could eat into margins if Bitcoin prices stall. Management needs to bring the new HPC hosting business online quickly to diversify away from these rising mining costs.
The digital infrastructure market is splitting into two paths: volatile Bitcoin mining and the explosive $200 billion AI data center market. AI infrastructure is growing roughly 30% annually and is on track to exceed $500 billion by 2028 as companies race to build the "factories" that train AI models. The most valuable asset in this industry is no longer the computer chips, but the multi-year access to hundreds of megawatts of power. TeraWulf is currently a niche player in the mining world that is attempting to become a key infrastructure provider in the AI hosting space.
The competitive dynamic for data center power has become brutal as Big Tech companies buy up every available megawatt of grid capacity. Barriers to entry are high because securing large-scale power permits takes years, creating a "land grab" environment. In this market, the ability to deliver physical capacity on schedule is the only thing that preserves pricing power.
CoreScientific is the most direct threat because it has already signed multi-billion dollar AI hosting deals, proving the model that TeraWulf is just beginning to build. MARA and Riot remain larger threats in the mining space, using their massive balance sheets to squeeze smaller players through sheer scale. CoreScientific is the dangerous threat because it has already successfully bridged the gap from mining to high-margin AI hosting.
TeraWulf is holding its ground by concentrating its focus on its 100% owned New York site, but it is currently racing to catch up with larger peers who have already secured AI contracts.
TeraWulf's protection comes from its "efficient scale" and rights to 750 megawatts of zero-carbon power infrastructure. Securing a power load of this size at a single site is a process that takes years and creates a local monopoly on large-scale digital hosting. This physical asset is far harder to replicate than the software or the mining rigs themselves.
The company's numbers show a business in transition: while revenue is up 43%, the ROIC is currently suppressed by heavy spending on new data centers. The combination of rising power costs and massive capital spending proves that the company's mining edge is narrow, and its long-term survival depends entirely on the AI pivot.
The forward-looking verdict is that TeraWulf's moat is strengthening as it converts commodity mining power into specialized AI data centers.
Hashrate hit 10.0 EH/s, but power costs per Bitcoin tripled post-halving.
Sold Nautilus JV interest for $200M to focus on Lake Mariner.
CEO Paul Prager founded the company and holds a significant equity stake.
Capital Allocation Track Record
Management has shown strong strategic foresight by selling its minority interest in the Nautilus JV to concentrate capital on its 100% owned Lake Mariner facility. This move simplifies the business and ensures that the upside from the AI pivot belongs entirely to TeraWulf shareholders. CEO Paul Prager has a long history in energy infrastructure, which is a critical skill set as the company transitions from mining to complex data center construction. While the recent $500 million debt raise adds financial pressure, it was a necessary step to secure the funding needed for the 2025 expansion.
The primary risk is the company's high dependence on Paul Prager's vision and the execution risk of building Tier 3 data centers, which are far more complex than simple mining sheds. Building for AI customers requires much higher standards for reliability and cooling than mining, and the company is still in the "proof-of-concept" stage for this business. If Prager or key infrastructure talent were to leave, the company would struggle to maintain the technical credibility required to sign major enterprise contracts. There is currently no obvious deep bench of data center veterans within the public leadership team, making this a key-person risk for investors.
We expect revenue to grow from $0.50B in FY2026 to $1.88B in FY2031 (~30% CAGR), with EPS growing from $0.25 to $1.45 (~42% CAGR). Revenue scales as the company transitions its power capacity from Bitcoin mining to higher-value AI and high-performance computing hosting. Profit margins expand as the company shifts toward long-term HPC contracts which offer more stable and higher pricing than volatile mining rewards. Operating margin expected to reach ~35% by FY2031.
Successful delivery of CB-1 and CB-2 facilities in 2025. Completing these first 70 MW of HPC capacity would prove TeraWulf can build Tier 3 data centers, likely leading to a massive valuation re-rating.
Signing a major AI anchor tenant for Lake Mariner. A long-term contract with a reputable AI firm would validate the infrastructure and secure years of predictable, high-margin revenue.
Expanding Lake Mariner capacity beyond the initial 750 MW. If the site can scale further, TeraWulf becomes one of the few single-site gigawatt-scale data center operators in the country.
Construction delays or cost overruns on HPC facilities. Failure to deliver CB-1 by Q1 2025 would damage management's credibility and delay the transition to higher-margin revenue.
Bitcoin price crash before HPC revenue comes online. If Bitcoin drops sharply, the cash flow from the mining arm might not cover the costs of the data center buildout.
AI demand cools before long-term contracts are signed. A slowdown in AI infrastructure spending could leave TeraWulf with expensive, half-finished data centers that have no tenants.
Below is our estimate of current and future fair value, with detailed reasoning and assumptions. Fair value is a judgment, not a fact, and other analysts will likely land on different numbers. Use it as one data point in your research, and apply your own discretion in any investing decision.
We use a Forward P/E framework anchored to the FY2028 earnings inflection point. This fits TeraWulf because current GAAP earnings are negative due to heavy infrastructure build-out costs, making 2028 the first year that truly reflects the recurring profitability of the new AI-focused business model. Price-to-earnings is the most effective way to value a business once it shifts from "capacity growth" to "cash flow generation."
Applying a 35x multiple to the FY2028 EPS estimate of $0.72 yields a future value of $25.20, which discounts back to a present fair value of $21. The 35x multiple sits at the high end of the peer range (IREN 28x, Applied Digital 35x, Core Scientific 22x), a position we believe is justified by TeraWulf’s industry-leading $3.1 billion liquidity and 1 GW power pipeline. The EPS basis of $0.72 is sourced directly from the deterministic projections for the first full year of ramped HPC operations.
Cross-checked with an EV/Revenue approach (FY2028 projected $630M revenue × 15x infrastructure multiple), we get a fair value of $19 per share—within 10% of our $21 P/E answer. This second framework focuses on the top-line "rent" generated by the megawatt capacity. The 15x revenue multiple is consistent with high-growth digital infrastructure platforms that own their power assets, confirming that our primary valuation is not overly reliant on aggressive margin assumptions.
We're assuming TeraWulf successfully converts 500 MW of its pipeline to high-margin HPC hosting by FY2028. This matches management’s guidance to bring 250–500 MW online annually and is supported by existing tenant signings with Core42 and Fluidstack, which demonstrate immediate market demand for their liquid-cooled sites.
We're assuming a 35x Forward P/E multiple for the 2028 earnings inflection point. While traditional data center REITs trade at 20–25x, TeraWulf’s vertically integrated power access and specialized AI infrastructure command a premium similar to high-growth peers like IREN or Applied Digital during their ramp phases.
We're assuming the 10% discount rate accurately reflects the risk of this pivot. Given the company's significant $3.1 billion cash position and lack of immediate debt pressure, a 10% rate is reasonable despite the high 4.26 beta, as the capital structure is now better aligned with contracted future cash flows.
The biggest risk is a construction or power-interconnection delay at the newly acquired 1 GW Kentucky campus. This would stall the high-margin pivot to High Performance Computing (HPC) revenue, keeping the business tied to the volatile Bitcoin cycle and likely compressing the multiple from 35x to 15x. Such a delay would knock roughly $12 off the per-share fair value as investors lose confidence in the execution timeline. Watch for quarterly updates on "megawatts under construction" versus "megawatts online."
Bear case ($14): Delays in energizing the 1 GW Kentucky campus push recurring HPC lease revenue back by more than 12 months; or Bitcoin prices drop below $45k, forcing the company to liquidate cash to cover high fixed operating costs at Lake Mariner.
Bull case ($34): Accelerated conversion of the 750 MW pipeline results in FY2028 EPS exceeding $1.10; or Hyperscaler demand for liquid-cooled data centers drives lease rates 20% above the currently modeled $630 million run-rate.
Clearthesis wrote this report from 28 sources, including SEC filings, industry research, and recent news.
How did you like this thesis?
Your feedback helps us make reports better for you
© 2026 Clearthesis.ai · Report generated on June 23, 2026
This is an AI-generated analysis for informational purposes only and does not constitute financial advice. Data and analysis may not reflect recent developments if viewed significantly after the generation date. Always conduct your own due diligence before making any investment decisions.
The market is bullish because TeraWulf owns rare power infrastructure that is far more valuable for powering AI data centers than for mining Bitcoin. By pivoting its 750 megawatts of zero-carbon capacity toward high-performance computing, the company shifts from a volatile commodity business to a stable, essential landlord for energy-hungry AI hardware.
Skeptics think that this transition carries significant execution risk as the company leaves its proven mining model behind. Turning a mining site into an AI data center requires complex infrastructure upgrades and new customer contracts that may not generate returns as quickly or reliably as the existing mining business.