Just saw an update from Bitdeer last week— they sold all their Bitcoin holdings, leaving the balance at zero. It seems crazy from the perspective of an ordinary Bitcoin trader, but this is actually a very strategic move by Wu Jihan.

So here’s the story. Bitdeer is no longer just an ordinary Bitcoin mining company. They are heavily investing in AI infrastructure, which requires a lot of money. By the end of 2025, their debt has reached 1 billion dollars. Then in February, they issued another 325 million in new debt, bringing the total to 1.3 billion dollars. That’s no small figure.

If you notice their pattern, this isn’t about 24-hour volatility or daily movements. It’s about a massive long-term bet. They are buying land and power assets in various locations: Rockdale Texas (563 MW), Clarington Ohio (570 MW), and Tydal Norway (175 MW). These three strategic sites will become their AI data centers. With a total capacity of 3002 MW, this is equivalent to the power needs of 10 to 30 Google data centers in one company.

Their debt structure uses convertible bonds maturing in 2029, 2031, and 2032. This is not accidental—it's calculated. Wu Jihan is giving a buffer period for each asset to start generating revenue. If everything goes according to plan, by 2029, their income should be able to speak for itself, and bondholders will prefer to convert into shares rather than take cash.

But this also carries significant risk. Clarington—representing 42% of the pipeline under construction—is facing a lawsuit from local steel producers. If this drags on, the entire schedule could be disrupted. Meanwhile, on the mining side, Bitcoin network difficulty just increased by 14.7% in February, and their gross margin dropped from 7.4% to 4.7%. Their mining operations are starting to thin out.

What’s interesting is the logic behind all this. Bitcoin mining for 12 years has always played the arbitrage game: use electricity and machines today, exchange for Bitcoin tomorrow. Now Wu Jihan is shifting the target. It’s no longer about Bitcoin price, but about the long-term demand for computational power in the AI era. He is buying this window with billions of dollars. He’s waiting for AI-generated revenue to pay down his debt faster.

Their AI revenue last year was only 10 million dollars, less than 2% of total revenue. Their GPUs over the past three months increased from 584 to 1,792, but utilization rate dropped from 87% to 41%. The machines are still in testing, not yet generating revenue. But analysts estimate: if HPC is fully realized, it could reach 850 million dollars in annual revenue. Management is more aggressive, claiming it could hit 2 billion.

Three prerequisites for all this to materialize: construction completed on time, long-term contracts with hyperscalers, and GPUs running at full capacity. Currently, none of these have been achieved. This is the battle Wu Jihan is playing—mining to support AI, AI is painting a big dream, and execution over the next two to three years will determine everything. If he hits the timing right, Bitdeer will no longer be just a mining company, but an infrastructure play in the AI era. If not, well, there’s a steel plant in Ohio already waiting.