TeraWulf signs $19 billion Anthropic lease

Bernstein reaffirmed its $36 price target on TeraWulf after the Bitcoin mining and data infrastructure company signed a 20-year lease with Anthropic that could generate about $19 billion in revenue and sold its majority stake in a Texas joint venture for $530 million, moves that deepen the company’s shift toward artificial intelligence infrastructure.

The brokerage said the two transactions support TeraWulf’s strategy of prioritizing fully owned data center assets and direct contracts with major AI customers. Bernstein kept its valuation largely intact after reviewing the impact of the deals, while raising its long-term earnings expectations for the company’s high-performance computing business.

TeraWulf shares closed Monday at $22.21, up 4.9%, as traders continued to reassess the company’s transition from a digital asset mining operator into a large-scale provider of power-backed computing infrastructure.

The Anthropic agreement covers 401 megawatts of IT capacity at TeraWulf’s campus in Hawesville, Kentucky. Bernstein said the contract carries an expected average annual yield of $2.4 million per megawatt, above its previous model estimate of $1.9 million per megawatt. That higher yield is central to the brokerage’s view that the deal improves the quality and visibility of TeraWulf’s future revenue base.

The lease is one of the longest announced so far in the fast-growing AI infrastructure market. It runs for 20 years and includes two renewal options that could extend the relationship by another 10 years. If fully realized, the agreement would provide TeraWulf with a multi-decade revenue stream tied to Anthropic’s need for large-scale computing capacity.

Initial capacity under the lease is expected to come online in the second half of 2027, with full completion targeted for early 2028. Once deployed, the Anthropic contract will significantly expand TeraWulf’s contracted AI business and add another major customer to its growing data center platform.

A major step away from mining volatility

The latest announcements mark a major point in TeraWulf’s transformation. The company built its foundation in digital asset mining, a business closely tied to Bitcoin prices, network difficulty, energy costs and broader cryptocurrency market cycles. The Anthropic lease gives TeraWulf a different kind of revenue profile: long-term, contracted income from a customer in the AI sector.

That distinction is important because the mining business can produce sharp swings in revenue and margins. When Bitcoin prices rise, miners can generate strong cash flow. When prices fall, or when mining difficulty increases, profitability can compress quickly. AI infrastructure contracts, by contrast, can provide more predictable revenue if the customer relationship remains intact and the data center assets are delivered on schedule.

Bernstein’s reaffirmed target reflects confidence that TeraWulf is moving toward a business model that the market may value differently from traditional mining operations. Rather than being judged mainly on cryptocurrency production, hash rate and Bitcoin exposure, the company is increasingly being assessed on power access, data center execution, contracted megawatts and tenant quality.

The brokerage said TeraWulf’s AI orderbook has expanded to roughly $27 billion in contracted revenue across three tenants. Together, those contracts cover about 839 megawatts of IT load. That scale places the company among the more visible public market names trying to convert mining infrastructure and power rights into AI-ready computing capacity.

Anthropic deal lifts revenue expectations

Bernstein raised its 2030 adjusted EBITDA forecast for TeraWulf by $43 million after incorporating the new lease and the Texas asset sale into its model. The firm now expects TeraWulf’s net AI revenue to rise from $209 million in 2026 to $1.7 billion in 2030, implying a 70% annualized growth rate over that period.

The brokerage also estimates that TeraWulf’s high-performance computing EBITDA could grow at a 94% compound annual rate, reaching $1.5 billion by 2030. Bernstein expects margins for that business to approach 85%, reflecting the economics of long-term, power-backed data center contracts once capacity is built and leased.

Those forecasts underline why the AI infrastructure pivot has become so important to TeraWulf’s market story. The company is no longer being viewed only as a miner with exposure to Bitcoin. It is increasingly being treated as a power and data center platform with the ability to serve customers that need large blocks of energy and computing capacity.

Bernstein’s valuation places TeraWulf at 21 times one-year forward EV/EBITDA. That produces an estimated enterprise value of $25 billion and supports the firm’s $36 per-share price target after accounting for $6.6 billion in net debt.

The price target implies meaningful upside from Monday’s closing price, though the brokerage also noted that execution risk remains. TeraWulf must build and deliver the necessary capacity, manage its balance sheet, and continue attracting high-quality customers in an increasingly competitive AI infrastructure market.

Texas stake sale supports full ownership strategy

Alongside the Anthropic lease, TeraWulf sold its 50.1% interest in the Abernathy joint venture for $530 million to a buyer group led by Fluidstack. The joint venture was formed in 2025 to develop a 168-megawatt data center in Abernathy, Texas.

The sale produced an 18% return on TeraWulf’s $450 million equity contribution, according to Bernstein. More importantly, it gives the company capital it can redirect toward projects where it has full ownership and operational control.

Bernstein said the payments are expected to be made in three installments. TeraWulf is set to receive $250 million within two weeks of signing, $150 million by the end of 2026, and the remaining $130 million by April 2027, subject to adjustments.

The decision to sell the majority stake in Abernathy is not being interpreted as a retreat from AI infrastructure. Instead, Bernstein said it fits TeraWulf’s preference for owning and controlling the assets that support its customer contracts. Full ownership can give the company greater influence over development timelines, operating standards, customer relationships and long-term economics.

The proceeds also provide non-dilutive funding at an important time. Building AI data centers requires large upfront capital commitments, especially when projects involve power infrastructure, cooling systems, high-density computing environments and long-duration customer contracts. By monetizing the Texas stake, TeraWulf gains cash that can be redeployed without immediately relying on new equity issuance.

Power access becomes the key asset

TeraWulf currently controls 3.6 gigawatts of power capacity across New York, Kentucky and Maryland. Bernstein said that foundation could support 250 to 500 megawatts of new IT deployments each year, depending on development pace, customer demand and infrastructure readiness.

That power base is central to the company’s appeal. As AI models become larger and more compute-intensive, access to reliable electricity has become one of the biggest bottlenecks in the data center market. Companies such as Anthropic, OpenAI, Microsoft, Google and other AI-focused firms require enormous amounts of power to train and run advanced models.

Digital asset miners are increasingly positioned as potential beneficiaries of that bottleneck. Many already control large energy sites, power contracts and operational teams capable of managing energy-intensive computing environments. While mining infrastructure is not identical to AI data center infrastructure, the overlap has become valuable as demand for high-performance computing grows faster than new power capacity can be brought online.

TeraWulf’s strategy reflects that industry-wide shift. The company is using assets originally associated with mining to build a business that serves AI tenants. If successful, the repositioning could move TeraWulf into a different valuation category, closer to infrastructure and data center operators than to pure-play cryptocurrency miners.

For traders following the sector, this shift changes the key questions around the company. Bitcoin production and mining margins still matter, but the larger debate is now about how much value the market should assign to contracted AI revenue, long-term power access and the ability to deliver large data center campuses on schedule.

A market rotation toward AI infrastructure

The timing of TeraWulf’s announcements is notable because the broader digital asset market has faced pressure. Bitcoin fell sharply in the first half of 2026, with a 33% decline over that period, while spot Bitcoin ETFs saw a record $4.5 billion in outflows in June.

That backdrop has reinforced the appeal of companies with exposure to AI infrastructure rather than direct reliance on cryptocurrency prices. Capital has increasingly rotated toward businesses tied to AI computing, power availability and data center development.

TeraWulf is one of several mining-linked companies trying to benefit from that rotation. The market has begun to separate firms with credible AI infrastructure pipelines from those still viewed mainly as Bitcoin mining operators. Companies with large power portfolios, strong sites and signed AI customers have attracted more attention as power constraints become a defining issue for the technology sector.

The Anthropic lease gives TeraWulf a high-profile anchor customer and a long-term contract structure that can be modeled more easily than mining revenue. That can matter for valuation, because predictable cash flows generally support higher multiples than revenue streams tied to volatile commodity or cryptocurrency prices.

Still, the transition is not without risk. AI data center development is capital intensive, technically demanding and highly dependent on timely delivery. Delays in construction, equipment procurement, grid interconnection or customer readiness could affect revenue timing. The company’s rising debt load also remains a point of focus, particularly as it funds large infrastructure projects ahead of full cash flow contribution.

Customer concentration remains a key risk

Bernstein noted that TeraWulf’s revenue remains concentrated among a small number of clients. That concentration could create exposure if one customer delays deployments, renegotiates terms, reduces demand or faces financial stress.

The Anthropic deal helps validate TeraWulf’s platform, but it also increases the importance of successful execution with a limited group of large tenants. In infrastructure businesses, long-term contracts can reduce revenue uncertainty, but they do not eliminate counterparty risk or project delivery risk.

Diversification will therefore remain a critical issue. If TeraWulf can add more AI customers over time, it may reduce dependence on any single tenant and strengthen the durability of its revenue base. If diversification lags, the company could remain vulnerable to changes in the plans or financial conditions of a few major customers.

The company also faces competition from established data center operators, utilities, private infrastructure funds and other mining companies seeking to repurpose power assets for AI workloads. The shortage of power gives TeraWulf an advantage, but it does not guarantee that every site will attract premium contracts or be developed without cost pressure.

Valuation now hinges on execution

Bernstein’s unchanged $36 target suggests the firm views the Anthropic lease and Abernathy sale as strategically positive but broadly consistent with its prior long-term expectations. The higher yield from the Anthropic contract improves the outlook, while the Texas sale supports the company’s full-ownership strategy and provides additional capital for future projects.

The bigger market question is whether TeraWulf can deliver on the scale now implied by its AI orderbook. A contracted revenue backlog of roughly $27 billion is substantial, but the value of that backlog depends on construction, financing, customer performance and operating execution over many years.

For traders, the stock is increasingly tied to a new set of drivers. Bitcoin prices may still influence sentiment, but TeraWulf’s valuation is now more closely linked to AI demand, power scarcity, data center development timelines and the credibility of its long-term customer contracts.

The recent transactions show how quickly the identity of a mining company can change when it controls large-scale energy infrastructure. TeraWulf is using that foundation to position itself as a landlord and operator for the AI economy, where access to power may be as valuable as the computing hardware itself.

If the company succeeds, the Anthropic lease could be remembered as a defining moment in its transition away from mining cyclicality and toward contracted AI infrastructure revenue. If execution falls short, the same long-term commitments that currently support the bullish case could become harder to justify.

For now, Bernstein remains constructive, keeping its $36 target in place and pointing to TeraWulf’s growing AI backlog, expanding power platform and strategy of direct customer relationships as reasons the company may continue to command a rising premium in the market.