TeraWulf Wants to Double Muskie. Its Balance Sheet Says Good Luck.

I always keep an eye out for bullish headlines that skip the arithmetic, and this week's news from TeraWulf is a textbook example. On Wednesday, Chairman and CEO Paul Prager told investors the company is looking to double the power of what it calls the Muskie Data Campus.

The market's reflexive response to words like "double" and "gigawatt" is to assume growth is inevitable. That is the false narrative. Doubling a data center project doesn't happen by executive wish. It requires securing additional power contracts, funding construction that doesn't come online for three years, and surviving a capital structure that is already stretched to its limits. Let's look at the numbers.

The arithmetic no one is quoting

TeraWulf's Q2 2026 results, released on August 5th, paint a company that is growing its revenue story while hemorrhaging cash. Total revenue for the quarter was $44.8 million, with $31.9 million — about 71% — now coming from high-performance computing lease revenue. That is a meaningful shift from the company's former bitcoin-mining days, and it's the reason the stock is still being taken seriously.

But beneath the headline transition sits a GAAP net loss attributable to TeraWulf of roughly $939.9 million for the quarter and a non-GAAP adjusted EBITDA loss of $18.3 million — a reversal from the $14.5 million adjusted EBITDA profit the same quarter delivered a year earlier. The GAAP loss was inflated by a $755.7 million change in the fair value of warrants, which is a mark-to-market accounting hit, not a cash outflow. Still, the adjusted EBITDA flip from profit to loss tells you the operating model is not yet self-funding.

Over the first six months of 2026, TeraWulf spent $1.38 billion on plant and equipment purchases and deposits. That's the kind of burn rate that makes "doubling" a project sound less like a plan and more like a capital-raising pitch.

The balance sheet

As of June 30th, TeraWulf reported approximately $3.0 billion in cash and restricted cash against total debt of roughly $5.3 billion — combining $1.1 billion in short-term convertible notes, $1.0 billion in long-term convertible notes, and $3.02 billion in long-term debt. Total stockholders' equity is in deficit territory. The company had a stockholders' deficit of nearly $79 million as of March 31st, with an accumulated deficit of $1.42 billion.

In plain English: the company's liabilities exceed its assets. The equity holders have negative book value. This is not a balance sheet that has room for an open-ended expansion option. Every additional megawatt of committed construction has to be funded by more debt or more equity — both of which dilute the position of existing shareholders.

And the dilution is already baked into the cap table. There are 425 million common shares outstanding plus 80.9 million warrants with a weighted-average strike of $0.17 — essentially in-the-money warrants that will convert into additional shares at virtually no cost to the holder. Add the convertible notes on the books, and the fully diluted share count could be substantially higher than what the headline number suggests.

The Anthropic lease doesn't solve the timing gap

The ~$19 billion Anthropic lease — a 20-year agreement for 401 megawatts at the Justified Data Campus in Hawesville, Kentucky — is real, and it's a legitimate long-term revenue anchor. But initial delivery isn't expected until the second half of 2027, with full delivery in early 2028. The money doesn't start flowing for another 14 months at the earliest.

Muskie's timeline is even later. The first 500 megawatts are targeted for the second half of 2028. A doubled 2 GW version would push the second half of that capacity even further out. Meanwhile, the company is already planning to raise approximately $3.5 billion in additional debt — including its first leveraged loan — led by Morgan Stanley, specifically to fund its Kentucky data center buildout.

The sequence here is important: TeraWulf needs to borrow billions to build facilities that won't generate revenue until 2027 and 2028, while carrying an existing debt load of over $5 billion and negative equity. That's not a growth story. It's a leveraged bet on the assumption that AI infrastructure demand stays insatiable and capital markets stay cooperative for the next three years.

What would make this work

I'm not saying TeraWulf is a bad company. I'm saying the "double Muskie" headline is a narrative, not a fact, and the gap between the two is where investors need to pay attention. For this thesis to hold, several things have to go right:

The $3.5 billion debt raise needs to close on favorable terms. Leveraged loans and high-yield bonds are expensive capital, and the interest service on an additional $3.5 billion on top of $5.3 billion of existing debt will be a significant drag on cash flow. The company hasn't disclosed the pricing or covenants, which is a data gap that investors should treat as a risk, not an oversight.

The Anthropic lease and future customer contracts need to materialize on schedule. A 20-year lease is only as good as the tenant's ability to pay over 20 years. Anthropic is a credible counterparty, but the lease is a projection, not a revenue stream, for the next 14 months.

Power availability in Eastern Kentucky needs to hold. TeraWulf has signed electric service agreements with Kentucky Power Company for up to 1 gigawatt of data center capacity at Muskie. Doubling to 2 GW would require a new or expanded power agreement — and there's no public filing confirming that additional capacity is secured. Power contracts are the bottleneck in this industry, not acreage or ambition.

The stock's position

WULF is trading at $17.08 today, down 3% on the session, with daily volume over 40 million shares and turnover approaching $700 million. That's a stock where institutional money is flowing in both directions, and the volatility is real. The shares have been hovering in the high-teens range after running much higher earlier in the year.

Bottom line

Despite the headline about doubling Muskie, I believe the market is over-reading CEO optimism as a commitment. The "may double" language in the original reporting is conditional — and the balance sheet doesn't support an open-ended expansion thesis. TeraWulf is transitioning from a bitcoin miner to a data center owner-operator, and the Anthropic lease is a genuine structural step in that direction. But the capital structure required to execute on a 2 GW Muskie, on top of the Justified campus and ongoing Lake Mariner buildout, demands billions more in debt or equity that will dilute shareholders further.

That being the case, I rate TeraWulf as a Hold. The company has a credible long-term revenue trajectory through the Anthropic lease and a real footprint in energy-advantaged Kentucky locations. But the balance sheet risk — negative equity, over $5 billion in debt, $1.38 billion in first-half capex, and a pending $3.5 billion debt raise — makes me uncomfortable with new positions at current levels. Investors who already own the stock should watch the debt raise pricing and the Muskie power contract status before adding. For new money, there are cleaner entries into the AI infrastructure theme where the balance sheet isn't the primary risk.

I rate TeraWulf as a Hold.