Wu Jihan's $1.3 Billion Bet: BitDeer's Gamble on the AI Infrastructure Gold Rush

On February 20, 2026, BitDeer announced a stark reality: zero Bitcoin in reserves. The previous week’s output of 189.8 BTC had been sold, joining 943.1 BTC from inventory—all liquidated. For a mining company founded on the principle of accumulating digital assets, this move signaled a fundamental pivot. Wu Jihan, the architect behind Bitmain and now BitDeer, had made a choice: exchange the Bitcoin miners’ traditional wealth accumulation strategy for a far more ambitious and uncertain gamble on artificial intelligence infrastructure.

The mathematics are straightforward but sobering. BitDeer carries approximately $1.3 billion in debt as of early 2026, with interest payments exceeding $65 million annually. Yet AI-related revenue barely registers—less than $10 million across the entire year of 2025, representing less than 2% of total income. The gap between financial obligation and current revenue generation is a chasm that Wu Jihan hopes to bridge through aggressive expansion over the next three years. Whether that hope transforms into reality will likely not be determined until 2029 or later.

Building the Infrastructure Fortress

Wu Jihan didn’t enter the AI data center race empty-handed. Over ten years in Bitcoin mining, BitDeer accumulated strategic assets that few competitors possess: established power infrastructure, operational expertise with large-scale energy systems, and crucially, the technical capability to develop proprietary hardware.

BitDeer’s global power pipeline capacity totals 3,002 megawatts, with 1,658 MW already operational and 1,344 MW under construction. To contextualize this scale: a typical Google hyperscale data center consumes 100 to 300 megawatts. BitDeer’s infrastructure is equivalent to packaging 10 to 30 such facilities into a single operating entity.

The company’s three core projects define this transformation. Rockdale, Texas operates 563 MW of capacity, primarily deployed for traditional Bitcoin mining—the cash cow generating stable returns. Tydal, Norway represents the fastest-moving initiative: 175 MW of existing mining infrastructure undergoing conversion to HPC/AI deployment, expected to deliver 164 MW of IT load capacity by year-end 2026. The conversion leverages existing hydroelectric power advantages and costs significantly less than greenfield development.

Then there’s Clarington, Ohio. This 570 MW facility with a 30-year power contract was positioned as the crown jewel, but has become the timeline’s greatest threat.

Wu Jihan’s silicon strategy mirrors his Bitmain playbook. BitDeer’s internally developed SEAL series mining chips have reached third-generation maturity. The SEAL03 achieves 9.7 joules per terahash, with the mass-produced A3 Pro now competing at global elite levels. The SEAL04 target is 5 joules per terahash—a threshold that would establish global dominance. These chips generate gross margins exceeding 40%, dwarfing the 4-7% returns from mining operations themselves. The pattern is identical to his prior venture: transition from purchasing others’ tools to manufacturing proprietary solutions.

The Debt Machine and the Stock Price Roulette

Financing Wu Jihan’s transformation required aggressive capital raises. From May 2024 through February 2026, BitDeer issued multiple tranches of convertible bonds totaling over $1.4 billion. Tether became the second-largest shareholder with a $100 million commitment in May 2024. Subsequent bond issuances carried interest rates initially at 8.5%, later declining to 5.25%, with the most recent February 2026 tranche locked at comparable rates.

The financial structure reveals implicit assumptions about stock performance. Current bonds mature in 2029, 2031, and 2032 respectively, with conversion prices set at approximately $9.93 per share—a 25% premium to contemporary equity offering prices around $7.94. The bond design operates on a fundamental premise: Wu Jihan needs the stock price to rise sufficiently for bondholders to choose equity conversion over cash repayment. If conversion fails, BitDeer must refinance—effectively requiring another equity raise and additional dilution.

The market has expressed skepticism through consistent action: each convertible bond announcement triggers a 10-17% stock price decline. BitDeer’s shares currently trade around $8, below the conversion price, leaving substantial room for recovery. Keefe Bruyette, the investment research firm, slashed its price target from $26.50 to $14, signaling that Wall Street demands revenue generation to justify the AI infrastructure narrative.

The Clarington Lawsuit: The Unpredictable X-Factor

In the same Ohio industrial park where BitDeer planned its 570 MW facility, American Heavy Plate Solutions—a steel manufacturer operating under a 2018 lease covering 9.9 acres—filed suit. Their argument centers on shared infrastructure: power distribution, roads, railroads, and communications lines. An AI data center would exceed permitted usage thresholds, they contend, violating restrictive covenants.

American Heavy Plate seeks a permanent injunction preventing construction. Clarington represents 42% of BitDeer’s stated under-construction pipeline. Should the court grant the injunction, the entire execution timeline unravels.

This lawsuit represents the singular biggest risk to Wu Jihan’s strategy—not debt dynamics, not stock price volatility, but a Pennsylvania steel company with legal standing to halt progress. Should construction remain blocked through 2028-2029, GPU deployment stalls around 41% utilization, revenue projections evaporate, and the 2029 bond maturity arrives without sufficient operational improvements to justify conversion pricing. Refinancing becomes the only option, triggering further dilution and widening the gap between conversion price and trading price.

The Narrowing Window: Execution or Extinction

Wu Jihan constructed a deliberate maturity schedule. By 2029, Tydal should be operational, generating European client contracts. By 2027, Clarington wins litigation and breaks ground, attracting major US customers. By 2028-2029, both assets operate at scale, revenue approaches $1 billion, and analysts reclassify BitDeer from discounted miner to premium AI infrastructure provider.

Three milestones, three opportunities for renegotiation. The first batch of bonds matures in 2029—precisely when, theoretically, BitDeer’s value proposition crystallizes. Bondholders watching rising stock prices would likely accept equity conversion rather than demanding cash payment.

But this scenario requires three conditions to materialize: construction completes on schedule, hyperscaler-level long-term contracts are secured, and GPU utilization reaches full capacity. None have been achieved yet. Roth/MKM estimates full-capacity annualized revenue potential at $850 million. Management targets $2 billion, triple 2025’s mining revenue. Both projections depend on simultaneous execution across multiple fronts.

Meanwhile, Bitcoin mining—the industry’s traditional core—is thinning. February 2026 witnessed the largest network difficulty surge since May 2021: a 14.7% jump in a single adjustment. With constant electricity costs, miners extract fewer coins per unit of energy. BitDeer’s Q4 gross margin compressed from 7.4% to 4.7% year-over-year. The mining cash flow that once seemed reliable is eroding.

The Philosophical Evolution: From Coins to Rent

BitDeer stands alone in emptying its Bitcoin treasury. Marathon Digital accumulated 53,250 BTC; Riot holds 18,000; Strategy holds 710,000. In mining culture, hoarding demonstrates faith—a public endorsement of Bitcoin’s long-term value. Wu Jihan abandoned this tradition entirely.

Yet the underlying logic remains unchanged. Bitcoin mining represented temporal arbitrage: deploy today’s electricity and machinery to capture tomorrow’s price appreciation. Now BitDeer deploys today’s capital and land to capture tomorrow’s computing power scarcity rent.

The shift from product sales to services to infrastructure rent is not unique to Wu Jihan. Amazon didn’t predict which internet company would dominate; it simply rented servers to all competitors. AT&T didn’t care what was discussed over phone lines; it collected revenue from every call. Wu Jihan bought a position where “regardless of which AI model or platform wins, they must pay my electricity bill.”

This is the true elegance of the strategy: Wu Jihan isn’t betting on any single AI outcome. He’s betting on the universality of computing power demand and his ability to supply it profitably.

The question reduces to timing. His $1.3 billion borrowed capital must sustain operations through the revenue inflection point. The Clarington lawsuit, if unresolved, collapses this timeline. The stock price must recover sufficiently to trigger bond conversion. GPU customers must commit to long-term contracts rather than short-term spot capacity purchases.

Wu Jihan bought this window of opportunity for over a billion dollars. He’s now waiting for the AI funding cycle to accelerate rapidly enough to catch up with debt service obligations due in 2029. The outcome remains genuinely uncertain, but the bet itself—transitioning from cryptocurrency accumulation to utility rent collection—mirrors infrastructure transformations that have enriched industries for decades.