Bitdeer Building an Energy Empire: $13 Billion Bet on Future Computing

The mining industry has always stood on a simple foundation: controlling energy. When Wu Jihan decided to shift Bitdeer from pure Bitcoin mining to AI infrastructure based on global energy resources, he was essentially applying the same logic but changing the scenario. In mid-February 2026, when Bitdeer announced it was clearing its Bitcoin positions with a final balance of zero, the company simultaneously completed a $325 million convertible bond issuance. These two decisions together marked the beginning of a $13 billion energy ecosystem.

From Skyscrapers to Energy Lords: Business Model Transformation

Founded in 2018 as a mining hardware rental platform, Bitdeer has now grown into one of the largest publicly listed mining companies, with a self-mining capacity of 63.2 EH/s—highest among public sector players, representing about 6% of the total Bitcoin network hash rate.

But Wu Jihan’s ambitions have far exceeded just selling computing power. Recent data shows that by early 2026, Bitdeer’s total global power capacity reached 3,002 MW—with 1,658 MW operational and 1,344 MW in construction or preparation. For context: giant data centers by Microsoft and Google typically operate at 100-300 MW per site. This means Bitdeer has accumulated energy needs equivalent to 10-30 Google facilities within a single entity.

The company’s book debt exceeded $1 billion by the end of 2025. Plus, the latest convertible loan in February 2026 of $325 million, bringing total liabilities over $1.3 billion. Every dollar of this massive debt is allocated to securing three critical assets: land, electricity, and computing capacity—what the AI industry calls the “most difficult-to-imitate assets.”

Funding Architecture: Playing with Time and Conversion

Bitdeer’s financing structure is designed with intricate strategic calculations. Since May 2024, when Tether made an initial $100 million investment and became the second-largest shareholder with an option for an additional $50 million, the company has conducted a series of coordinated funding rounds:

• June 2024: First convertible loan, $150 million at 8.5% annual interest
• November 2024: Second convertible loan, $360 million at reduced 5.25% interest
• November 2025: Combined offering of $548.4 million (400 million bonds + 148.4 million shares)
• February 2026: Re-raising round, $325 million in convertible bonds, extending the maturity of the initial bonds from 2029 to 2032

Each issuance follows the same pattern: stock price drops 10-17% upon announcement, yet the company always manages to raise funds. The core mechanism is the convertible bond with a strike price around $9.93—about a 25% premium over the simultaneous stock price of $7.94.

The logic of these convertible bonds is simple yet effective: as long as the stock price rises, bondholders will convert to equity rather than demand cash payments. This allows Bitdeer to effectively avoid short-term payment obligations, provided its transformation narrative is accepted by the market.

However, with an annual interest burden assuming an average rate of 5% on $1.3 billion principal, the company spends over $650 million annually. In 2025, revenue from its AI/HPC cloud services was less than $120 million—insufficient to even cover half a semester’s interest payments. This means the company currently relies entirely on ongoing debt issuance to sustain its financial cycle.

Geographic Shift: Three Strategic Locations Shaping the Future

Bitdeer’s energy investments are spread across three locations, each representing a different business philosophy:

Rockdale, Texas – A Solid Foundation
Capacity of 563 MW (including 179 MW expansion) focused on Bitcoin mining. This site generates stable cash flow and relatively consistent gross margins, serving as the main financial source for other expansions.

Clarington, Ohio – The Highest Stakes and Greatest Risks
A 30-year lease for 570 MW with signed power contracts. Expected to operate in Q2 2027 as a world-class HPC/AI hub. But now, this site faces serious legal hurdles: American Heavy Plate Solutions, a steel producer in the same industrial complex, has filed a lawsuit to halt construction, citing negative impacts on shared infrastructure. Clarington accounts for 42% of the capacity under development—if halted, the entire AI transformation schedule must be rewritten.

Tydal, Norway – The Lowest-Risk Option
Conversion from a mine to an AI data center with 175 MW capacity, scheduled for completion by late 2026, with effective IT of 164 MW. Local hydroelectric power offers competitive operating costs, and the transformation costs are far lower than building from scratch. This is the most advanced project in execution and the most controllable risk.

Technological Dimension: From User to Maker

One rarely highlighted but crucial dimension is Bitdeer’s proprietary mining chip development, the SEALMINER series. The third-generation SEAL has achieved energy efficiency of 9.7 joules per terahash. The A3 Pro, starting mass production in September 2025, has entered the top tier globally. SEAL04 aims for 5 joules per terahash—if achieved, it will surpass all mass-produced mining machines on the market.

Gross margins on proprietary chips exceed 40%, far higher than mining itself. This echoes Wu Jihan’s strategy at Bitmain: evolving from buyer (buying others’ shovels) to producer (making your own shovels).

GPU Growth and Utilization: Betting on Adoption

Bitdeer’s AI/HPC cloud business is growing exponentially, but capacity jumps have outpaced revenue growth. GPU units increased from 584 to 1,792 in three months—tripling. Yet, utilization rates fell from 87% to 41%, mainly because new machines’ installations outstripped actual demand.

The B200/GB200 GPUs are still in customer testing and have not yet generated significant revenue. Equipment is installed, machines are being set up, but revenue lags behind. How high is this growth ceiling?

Analyst Roth/MKM estimates that if HPC utilization reaches full capacity, annual revenue could hit $850 million. Management is more aggressive: allocating 200 MW for AI cloud, annual revenue could surpass $2 billion—three times the mining revenue of 2025.

But both projections depend on three conditions: infrastructure completion on time, long-term contracts with major hyperscalers, and full GPU utilization. Currently, none of these conditions are met.

Time as a Resource: Bond Maturity Structure

Bitdeer deliberately designed its bond maturity structure with three key dates: 2029, 2031, and 2032. This is no coincidence but a calculated buffer strategy:

• 2029: When the first bonds mature, the Tydal and Clarington projects should theoretically be operational with measurable revenue
• 2031: When the second batch matures, contributions from AI services should be significant
• 2032: When the third batch matures, the market will have enough data to assess whether the transformation succeeded

These three points create three opportunities for renegotiation—either with investors or the capital markets.

But Wall Street remains skeptical. Keefe Bruyette analyst lowered the target price from $26.50 to $14, while the current stock price hovers around $8. The market signals are clear: the transformation story must be backed by concrete revenue realization, not just projections.

Best-Case vs. Worst-Case Scenarios: Two Paths, Same Reality

Smooth Path:
By late 2026, Tydal renovations are complete, with 164 MW data center fully operational, European customer contracts underway. In 2027, Clarington wins in court, construction of 570 MW in Ohio officially begins, major US clients follow. By 2028-2029, both core assets operate at full capacity with revenues reaching billions, and analysts upgrade Bitdeer from “discounted mining company” to “premium AI infrastructure.” By 2029, first bondholders see stock appreciation and opt for equity conversion rather than demanding cash repayment.

Hindered Path:
Litigation in Clarington drags on for two years, construction halts. Tydal faces delays; GPUs remain at about 41% utilization. By 2029, the first bonds mature, cash on balance sheet is insufficient, forcing refinancing, stock continues to decline, and conversion thresholds become even harder to reach. Mining margins thin as Bitcoin network difficulty continues to surge (up 14.7% in February 2026 alone, the largest increase since May 2021).

Energy Control as an Eternal Moat

Bitdeer’s decision to sell all Bitcoin holdings (to zero balance) marks a fundamental philosophical shift. In mining, there’s a tradition of stacking coins as proof of faith in Bitcoin’s long-term value. Marathon holds 53,250 BTC, Riot 18,000, Core Scientific retains significant assets. Bitdeer now holds none.

But the arbitrage of time logic remains—the target has just shifted. Ten years ago, the industry bet on Bitcoin price appreciation. Now, buying land and electricity is a bet on explosive demand for computational power in the AI era.

What Wu Jihan truly bought is a position: “Whoever wins in the AI race, they will pay electricity to me.” Wu Jihan isn’t guessing the winner; he’s just controlling the entry point to energy for all potential winners.

Amazon doesn’t guess which internet company will dominate; it just rents servers to everyone. AT&T doesn’t care what’s discussed on the phone; it just charges per call. From product sales to service sales to rent collection—industry evolution always follows one path: control of fundamental infrastructure.

Wu Jihan has purchased access to this time window with billions of dollars. Now, he waits for AI revenues to accelerate his debt repayment, while ensuring that the energy he controls remains an unshakable asset in every future scenario.