Signed a 19 billion yuan order, but dropped 8.87%

NEXT PICK · Market Insights

Signed a 19 billion yuan order, but dropped 8.87%

Tuesday, July 7, 2026

A Bitcoin mining company just announced its biggest deal in twenty years—signing a 20-year data center lease with AI giant Anthropic worth about $19 billion. However, not only did the stock fail to soar, it opened low but rose high, then turned downward, closing at $20.24, plunging 8.87%.

On one side are contracts that are almost written into company history; on the other, the confidence of holding about $3.1 billion in cash on hand, yet the market chose to vote with its feet on this day. Is this a classic scenario of "all the good news being spent," or has smart money seen through the unclosed crack behind the contract?

Latest price $20.24 ▼ -8.87%

Anthropic contract $19 billion ▲ 20-year term

Cash reserves Approximately $3.1 billion — End of Q1

Why was it sold off on the day the epic contract was signed?

First, let's clarify today's unusual movement. On July 6, TeraWulf officially announced a 20-year lease with Anthropic, providing about 401 megawatts of AI computing power for its "Justified Data" campus in Hawesville, Kentucky. The contract is expected to generate approximately $19 billion in revenue, with the first batch of capacity planned to go live in the second half of 2027 and full capacity in early 2028.

This marks a key leap from being a "Bitcoin miner" to an "AI infrastructure landlord." As soon as the news broke, the stock surged as much as 16% in pre-market trading. But on July 7, the situation took a sharp turn, and the stock price closed down 8.87%.

Where does the contrast come from? Wall Street analysts have collectively raised their target prices, but KBW has pointed out the thorn—the investment-grade credit guarantee supporting this lease has not yet been finalized, and it depends on Anthropic's choice of hardware suppliers within four months; Moreover, the company sold the majority stake in the Abernathy joint venture project to Fluidstack, and lacked new catalysts in the short term, so the funds to "sell the news" flowed first.

So here's the question

For a massive contract that hasn't yet secured its foothold in financing, should the market price it based on what has already been received, or should it discount it because there are still uncertainties? Today's drop is the first vote on this issue.

Half relies on Bitcoin, half on AI—who exactly is it?

TeraWulf is headquartered in Easton, Maryland, USA, and was founded in 2021, specializing in digital infrastructure. Opening its 10-K, the core narrative is just one sentence—to power both Bitcoin mining farms and high-performance computing (HPC) data centers with clean, low-cost, and reliable energy.

Its income thus grew two curves. One is Bitcoin mining, which is directly tied to the roller coaster of coin prices; The other is cloud computing and hosting services for the AI industry, catering to the nearly greedy appetite for computing power driven by large model training.

This dual-engine steering is accelerating. In the first quarter of 2026, HPC leasing revenue contributed over 60% of total revenue, surging 117% quarter-on-quarter, while mining revenue was halved by 50% during the same period. In other words, the old label of "Bitcoin miner" is being gradually torn off by the company itself.

Where does its irreplaceability come from? It's not some advanced patent—mining machines are sourced from Bitmain, and servers use standard NVIDIA GPU clusters. The true moat is low-cost, proprietary clean energy and a compliant land located in the U.S. that can accommodate government and AI-sensitive clients. You can think of it as a "landlord" holding cheap electricity and prime locations, not a tech geek.

The other side of the 19 billion yuan contract is a valuation vacuum with a price-to-sales ratio of 60 times

Let's first look at that glaring contrast. The company's current market value is about $10 billion, while FY2025 full-year revenue is only $168.5 million, resulting in a price-to-sales ratio as high as about 60 times—by comparison, mature data center operator Equinix typically has only 8 to 10 times, and mainstream Bitcoin miners only 2 to 5 times.

Traditional valuation anchors have almost completely failed here. Common metrics like negative price-to-earnings ratios, enterprise value multiples (EV/EBITDA), and free cash flow yield cannot produce decent figures because the company continues to incur losses. FY2025 net loss reached $661.4 million, but it's important to see that about $429.8 million is non-cash fluctuations in the fair value of warrants and derivatives liabilities, not operating losses burned through real money.

This must be done fairly on behalf of this company

The market's panic that "cash is about to run out" actually doesn't hold up. By the end of 2025, the company's cash on hand plus restricted cash reached about $3.7 billion, with about $3.1 billion still at the end of the first quarter. What it lacks is not short-term survival money, but the execution capability to complete the AI park on schedule in 2028.

So why does the market give a company with revenue under 200 million yuan a 60-fold price-to-sales ratio? The answer lay in that $19 billion contract. But if the credit guarantee for this contract is delayed, who will pay this 60-fold premium?

The critical point from "government validation" to "commercial implementation."

Before the contract was officially announced, the company was still stuck in an awkward phase—although Bitcoin mining was commercially operational, its AI data center business was still small in scale. The amount of DoD-related contracts previously secured was negligible, spanning both concept and proof-of-concept levels, and did not meet the threshold for substantial commercialization.

And the July 6th Anthropic lease was exactly that knocking at the door. A 20-year contract, officially disclosed in SEC filings and company press releases, with investment-grade credit expectations, marks the tipping point of moving from "government validation" to "commercial implementation."

But market attitudes remain divided. The stock price is $20.24, having climbed above the 200-day moving average of $16.84, yet remains below the 50-day moving average of $24.29—indicating that the medium- to long-term trend has not yet deteriorated, but short-term sentiment has clearly cooled. The market admits the narrative has changed, but it's just not ready to pay in full for every step before cashing out.

Between the landlord and the contractor, which side does it stand on?

On the dual track of AI infrastructure and Bitcoin mining, TeraWulf is surrounded by rivals on all sides. The most subtle competitor is CoreWeave—a partner in the Anthropic ecosystem, holding massive H100/H200 clusters, and a direct competitor in data center hosting business, leaving TeraWulf far behind in computing power density and technical depth.

On the mining side, North American mining companies like CIFR and HUT are locked in fierce competition. HUT has larger revenue scale and computing power deployment, and is also transitioning to AI HPC, moving faster while converging in direction; CIFR partially achieved positive free cash flow, which is exactly the answer TeraWulf cannot provide for now.

So what exactly is TeraWulf's moat? Ultimately, it's the cost of switching and that piece of land. Once Anthropic's servers are physically deployed, the cost of migration will be nearly impossible; Moreover, the 12 to 24-month construction cycle for self-built clean energy parks makes it difficult for clients to find alternatives during the contract period. It may not be the most skilled contractor in this field, but it could be the landlord with scarce electricity and land, making it the hardest to replace.

The ceiling is high enough, but can it really reach it?

TeraWulf's imagination spans two major markets. On the Bitcoin mining side, the industry's annual revenue fluctuates between $50 billion and $80 billion with coin prices, but its share of total network hash power is less than 1%, so its theoretical contribution is limited. The real sea of stars lies on the other side—global data center construction spending has exceeded $300 billion annually, and many institutions expect the hyperscale data center market to surpass $500 billion by 2027, with a three-year compound annual growth rate of about 20%.

The problem lies in penetration rates. Even conservatively estimates that the market for customized AI hosting is worth tens of billions of dollars, TeraWulf's current quarterly revenue is about $34 million, with a market share so small it can almost be rounded to zero. Anthropic's $19 billion contract means it has overcome a hurdle it has failed to overcome in recent years—but only if it can deliver 401 megawatts in full by 2028.

The bulls' trump cards: contracts, cash, and clean energy

If the bulls used to tell stories about "whether rumors come true," after July 6, the foundation of the story has shifted to concrete and steel.

▲ Bull Case

Bears' counterattack: triple pressures of valuation, execution, and dilution

The bulls' foundation is even harder, but the bears' firepower hasn't died out; they're targeting the long gap between contracts and reality.

▼ Bear Case

Keep a close eye on these three fuses in the coming year

The next pace of the company's stock price will be driven by three types of events: whoever pays first will gain pricing power for the next round.

•	About four months ahead: Anthropic finalizes hardware suppliers and implements investment-grade credit guarantees—this is the key to unlocking today's decline
•	Second half of 2027: The first batch of production capacity in the Justified Data Park will go online, marking the first hard measure of execution
•	Next quarterly financial report: Can HPC rental revenue maintain triple-digit quarter-on-quarter growth, and will the ebb and flow of mining business continue?

It has more than tripled in a year, and prices have long been ahead of the fundamentals

Looking back at this ticket, the story is more exciting than its performance. In the early days of its IPO in 2022, it hovered below $1 during the crypto bear market, surviving through debt financing and asset restructuring. The turning point came after Bitcoin's 2024 rebound, with the real main rally concentrated from 2025 to early 2026—with gains exceeding 300% in the past 12 months and about 95% within the year.

What drives this rally is less about the financial reports and more about the grand narrative of AI infrastructure and the layers of expectations surrounding Anthropic's contract. Interestingly, when the contract was actually delivered in July 2026, the stock price actually pulled back—because the market had already swallowed up the optimistic pricing during the rumor phase. Prices have consistently outperformed fundamentals, and from 2026 to 2028, this is the key window to test whether this lead is foresight or overdraw.

⚠️ Risk Notice • The investment-grade credit guarantee supporting the 19 billion yuan contract has not yet been finalized, posing a risk of default • A price-to-sales ratio of about 60 times lacks current earnings support, and negative news can easily trigger valuation cuts • The park will not reach full production until 2028, and the massive capital expenditure may lead to significant equity dilution • Significant fluctuations in Bitcoin prices directly impacted mining revenues that were still operating • Beta 4.26, short-term volatility is magnified exponentially by market sentiment

🟡 Neutral The positive news is now a reality, but the cracks between financing and execution have yet to close, and both bulls and bears are still missing one more opportunity.

💬 Discussion Whether it's "all the good news is gone" or "the night before getting on board" depends on the credit card four months later. Its current real-time buy and sell prices can be viewed via WULF on the official account.

Data source

•	Source: NextPick real-time snapshot + TeraWulf earnings and press releases + SEC 8-K announcement + cross-verification by mainstream financial media (Reuters, CNBC, DCD, etc.).