I just noticed Wu Jihan's movement is quite aggressive. Bitdeer just sold all their Bitcoin—943 BTC liquidated in total, leaving their balance at zero. This is not a small decision for a mining company. But what’s more interesting is what they’re doing with the money: buying land and electricity worldwide for AI infrastructure.

So here’s the story. Bitdeer started as a mining sharing platform in 2018, now it’s one of the largest publicly listed mining companies. They have 63.2 EH/s of computing power—highest among public mining firms, roughly 6% of the total Bitcoin network hash rate. But Wu Jihan now says: mining has been around for a long time, AI infrastructure is the future.

Their total debt now is $1.3 billion. Not a typo—one point three billion dollars. Book debt by the end of 2025 is already over $1 million, and this February they issued another convertible bond worth $325 million. All of this for one goal: securing global land and power assets.

If you think this is crazy, wait, there’s more insane. Their total power capacity is now 3,002 MW. For context: giant data centers like Google and Microsoft usually operate at 100-300 MW. So Bitdeer is basically gathering enough electricity for 10-30 Google-scale data centers in one company. Anything that produces energy is considered one of the hardest assets to replicate in the AI industry, and Bitdeer is aggressively accumulating this.

Their main projects are three. First, Rockdale, Texas 563 MW—already operational, focused on mining, stable cash flow. Second, Clarington, Ohio 570 MW—this is the core of their entire AI transition plan. But there’s a problem: a steel plant in the same area, (American Heavy Plate Solutions), is suing them, claiming that the construction will disrupt shared infrastructure. If this lawsuit wins, the entire timeline is wrecked. Third, Tydal, Norway 175 MW—this is the most interesting, transforming a mine into an AI data center with hydro power, scheduled to finish by late 2026, with much lower transformation costs than building from scratch.

Now, their debt structure is actually more sophisticated than it looks. They have three series of convertible bonds maturing in 2029, 2031, and 2032. This is intentional—designed as a buffer. When the first batch matures in 2029, Tydal and Clarington should already be operational. When the second batch matures, AI revenue should be meaningful. By the third batch, the market will judge what this company really is.

But Wall Street isn’t very convinced. Keefe Bruyette lowered their target price from $26.50 to $14. The current stock price is around $8. The market is basically saying: “Show us the revenue, not the story.”

And the numbers are... concerning. AI/HPC cloud revenue in 2025 will be less than 1928374656574839.25T per year. That’s less than 2% of total revenue. Their GPU shipments jumped from 584 to 1,792 in three months—tripling. But utilization rate dropped from 87% to 41%. They installed machines too quickly, B200/GB200 are still in customer testing phases, not generating revenue yet. Power is already installed, machines are being set up, but revenue hasn’t caught up.

There are two floating revenue potential scenarios. Roth/MKM estimate, with full HPC capacity realization, potential annual revenue of $10 million. Management is more aggressive: allocating 200 MW for cloud AI, revenue could be over $850 milliard—three times the mining revenue in 2025. But both figures depend on three conditions: construction finishing on time, securing long-term hyperscaler contracts, and GPUs running at full capacity. Currently, none of these three are met.

Their annual interest—assuming an average 5% interest rate and $1.3 billion principal—is more than $2 million per year. Meanwhile, current AI revenue isn’t enough to even cover half of the six-month interest. All of this depends on continuous debt issuance to keep spinning. The pressure is real.

Oh, and one rarely discussed thing: SEALMINER. Bitdeer isn’t just building mining centers, but developing their own mining chips. The SEAL series is now in its third generation. SEAL03 has an energy efficiency of 9.7 joules per terahash. The A3 Pro, scheduled for mass production in September 2025, ranks among the top tier globally. SEAL04 aims for 5 joules per terahash—if achieved, it will surpass all existing mass-produced mining machines. Gross margins from in-house chip development are over 40%, much higher than just mining. This is basically Wu Jihan repeating what he did at Bitmain: from buying other people’s picks and shovels, to making his own picks and shovels.

So, timeline-wise, here’s what to watch. End of 2026, Tydal renovation completes, 164 MW hydro data center in Norway becomes operational, European customer contracts start coming in. 2027, hopefully Clarington wins in court $650 or lawsuit is resolved(, 570 MW Ohio construction officially begins, American customers follow. 2028-2029, two core assets fully operational, revenue heading into the billions, analysts reclassify Bitdeer from a mining company discount to an AI infrastructure premium. 2029, first debt batch matures, debt holders will look at the stock price and probably choose to convert to shares rather than take cash.

But there are many single points of failure. Clarington lawsuit could drag two years, halting construction. Tydal could delay, GPU utilization stay around 41%. The first debt matures in 2029, cash on the balance sheet isn’t enough, forced refinancing, shares keep diluting, and the conversion threshold becomes even harder to reach. The worst-case scenario is real.

And on the mining side, there’s no time to rest. February 2026, Bitcoin network difficulty spikes 14.7%—the biggest single increase since May 2021. With the same electricity, fewer coins are mined. Q4 gross margin drops from 7.4% a year ago to 4.7%. The mining branch is slowly becoming thinner.

But what’s most interesting about all this is the philosophy behind Wu Jihan’s decision. He sold all his Bitcoin—while MARA holds 53,250 BTC, Riot holds 18,000, Strategy holds 710,000. In the mining world, hodling is a statement of faith. Bitdeer now has zero.

But this isn’t about losing faith in Bitcoin. It’s about a fundamental evolution of mining logic. Ten years ago, mining bet on one thing: future coin prices higher than current electricity costs. Now, buying land bets on the same thing: future AI compute demand will be higher than current land/electricity costs. The object changes, but the time arbitrage logic remains the same.

What Wu Jihan is actually buying is a position: “Whoever wins the AI race, they will have to pay electricity to me.” He’s not guessing who will win. He’s just controlling the entry point into that race. Amazon doesn’t guess which internet company will succeed, just rents servers to everyone. AT&T doesn’t care what you say on the phone, only whether you call or not.

So this is basically Wu Jihan buying optionality with $1.3 billion. He’s waiting for AI money to catch up with the speed of his debt. The window is narrow—2026 to 2029 is a critical period. If he executes correctly, the timeline could work. If one thing fails—Clarington lawsuit, Tydal delay, or AI revenue doesn’t meet expectations—the dominoes could fall fast. This is a high-risk, high-reward play, and the market is still skeptical. But if Wu Jihan succeeds? This infrastructure play could become more valuable than mining itself.