I was monitoring Bitdeer's numbers and noticed something quite interesting happening. At the beginning of February, the company liquidated all its Bitcoin reserves — 943 BTC sold at once, leaving the balance at zero. This drew attention because in the world of mining companies, accumulating coins is almost a religion. Marathon has 53,000 BTC stored, Riot has 18,000. But Wu Jihan did the opposite.

The official explanation is that they needed cash to buy land and expand AI infrastructure. It makes sense on the surface, but there's something much deeper going on here. Bitdeer accumulated $1.3 billion in debt over the past two years. That’s heavy. But if you understand the logic behind it, it’s less about easy money and more about a calculated bet on the future.

Look: from the beginning, mining is temporal arbitrage. You take electricity and machines today, betting that Bitcoin will be worth more tomorrow. Simple. Now Wu Jihan is changing the game. It’s no longer about the coin’s price; it’s about demand for computational power in the AI era. The structure of the bet remains the same, but the object has changed. Instead of waiting for Bitcoin to rise, he’s betting that demand for processing capacity will explode.

The numbers are impressive on paper. Bitdeer has 3,000 MW of planned global energy capacity. To compare, a large Google data center has between 100 and 300 MW. So we’re talking about 10 to 30 Google-scale data centers in a single company. Clarington in Ohio will be the core of all this — 570 MW focused on AI and HPC. Tydal in Norway, 175 MW with cheap hydroelectric power. Rockdale in Texas, 563 MW already operating.

But here’s the problem: AI revenue is still almost nothing. In 2025, it generated only $10 million per year, less than 2% of total revenue. They tripled the number of GPUs in three months, but utilization dropped from 87% to 41% because the machines were installed too quickly. The B200 and GB200 are still in testing phases. Power is connected, machines are arriving, but revenue isn’t keeping up.

Analysts estimate that when everything is ready, annual revenue could reach $850 million. Management is more aggressive and talks about $2 billion. But that depends on three things happening: construction on time, securing long-term contracts with hyperscalers, and GPUs running at maximum capacity. None of these three have happened yet.

The debt structure is well thought out, at least. Three series of convertible bonds maturing in 2029, 2031, and 2032. Each date is a potential renegotiation point. If all goes well, by 2029 Tydal will already be generating revenue, in 2027 Clarington will be out of the ground, and between 2028 and 2029, both main assets will be fully operational. Analysts then reclassify the company from a discounted mining operation to an AI infrastructure with a premium. Creditors, seeing the stock price rise, exchange bonds for shares instead of demanding cash.

But there’s a ticking time bomb: Clarington is being sued by a local steel company that rents in the same industrial park. They argue that the data center will interfere with electricity, roads, and shared railways. If this delays construction by two years, the entire timeline collapses. Clarington accounts for 42% of the pipeline under construction.

Meanwhile, mining is becoming tighter. In February, Bitcoin network difficulty jumped 14.7%, the biggest jump since May 2021. With the same electricity, you mine fewer coins. Gross margin fell from 7.4% to 4.7% in the last quarter.

What Wu Jihan really bought with these billions is a strategic position: being the one who controls the entry point, not who bets on which horse wins. Amazon didn’t bet on which internet startup would succeed; it just rented servers for everyone. AT&T doesn’t care what you say on the phone; it just charges for the call. The evolution is always the same: from selling products to selling services, then to charging rent.

The strategy is clear: mine Bitcoin now to support AI, while AI creates bubbles that may or may not materialize. If these bubbles turn into reality in the next two or three years, he controls the infrastructure everyone will need. If not, well, he’ll have a well-thought-out debt structure and three years to renegotiate.

It’s a colossal bet, but the logic is solid. Many people are talking about this in crypto communities, Telegram groups, and elsewhere. The question now is whether Wu Jihan can make money from this transformation before the debts become too heavy. The next two years will be decisive.