Bitdeer Mined 1,355 BTC in September. The Coin Count Isn't the Point.

I've read a lot of miner monthly updates. Most of them are theater. A company posts a big BTC number, the number goes up, the stock pops, and nobody asks what it cost to produce those coins. Bitdeer's September report deserves more attention than that, just not for the reason the headline suggests.

The company mined 1,355 BTC last month. A year earlier, the same month produced 452 BTC. That's roughly a 3x jump. Impressive on its face. But production growth is the least useful number in any mining report, and here's why.

The Math Behind 1,355 Coins

Start with the hash rate split, because most people get this wrong. Bitdeer reported 82.3 EH/s of self-mining hash rate and 28.7 EH/s of co-mining capacity. Those are two different businesses. Self-mining means the company owns the machines and keeps the coins. Co-mining usually means partner-owned hardware running in Bitdeer facilities, with a hosting fee or a revenue split attached. You can't add 82.3 and 28.7, call it 111 EH/s, and treat the whole thing as Bitdeer's earning power.

Divide the production number by the self-mining fleet and it works out to about 16.5 coins per EH/s for the month. That's the figure I'd track month over month. If it holds steady, the fleet is running as expected. If it slides while hash rate grows, difficulty is eating the gains and the company is running harder just to stand still.

And that's before cost. Rising BTC output can coexist with shrinking margins. If every competitor is plugging in new machines, the network absorbs the extra hash and per-machine yield falls. Add a hot summer of curtailment or a spike in power prices and a 3x production jump can land with a worse cash cost per coin than the year before.

Nobody in the release tells you capex. Nobody tells you the all-in cost per BTC. Those two numbers decide whether those 1,355 coins are profit or just inventory funded by investors.

So is Bitdeer actually making money on this? That depends on a number the company didn't lead with. At a $60,000 average September price, 1,355 coins would gross about $81 million. At $40,000 it's $54 million. Same coins, wildly different quarter.

AI Capacity Isn't Revenue

Sitting in the same update is the part the market actually cares about. Bitdeer keeps building out AI cloud and data center capacity, including agreements in Malaysia with delivery windows stretching into 2027 and 2028.

Here's what matters: a signed capacity agreement isn't revenue. It's a promise about a future date. Between now and 2027, the company has to finish the shell, secure the power, install cooling, and land a tenant willing to pay for it. Every one of those steps costs money up front and carries schedule risk.

That's not a knock on Bitdeer specifically. It's a feature of the pivot. Mining sites and AI data centers both need megawatts and cooling, so the overlap looks obvious from the outside. Inside, the workloads are different, the networking requirements are different, and the hardware refresh cycles are different. Retrofitting a mining hall for GPU racks isn't a swap. It's a rebuild with a new budget.

What the street is missing: these AI deals are credit trades dressed up as tech trades. The value hinges on who's signing the contract and whether that counterparty can pay in 2028. A 100 megawatt commitment from a shaky startup is worth far less than 30 megawatts from a hyperscaler. Investors are pricing the megawatts and ignoring the names on the paper.

Here's my hot take. Most of these miners aren't AI companies. They're power landlords with a mining hobby, and they'll be valued as landlords the moment the market tires of the word AI. That's not a bad business. It's just a different multiple than the one being applied today.

How I'd Read the Next Report

Two things I want from October's numbers. First, self-mining hash rate growth paired with that coins-per-EH figure. Growth without yield is just spending. Second, capex and any fresh financing. Miners fund data center builds with converts, equity, and equipment debt. Share count creep is the quiet tax on this whole sector, and Bitdeer is asking investors to fund two capital-hungry businesses at once.

The bull case is real. A miner with operating cash flow from hashing plus a contracted data center pipeline is a more durable business than a pure miner, especially as halving cycles compress margins. Frankly, that's the only version of this industry that survives the next difficulty adjustment with its balance sheet intact.

The bear case is simpler. The AI pipeline keeps sliding into 2027 and 2028 while the mining fleet keeps consuming capital today.

On risk, this is a show-me story. The mining side is showing. The data center side is talking. Until delivery windows turn into invoices, I'd weight the 1,355 BTC seriously and the Malaysia pipeline lightly.

My honest take: September was a genuinely good operational month, and the market's response should hinge on cost per coin, not coin count. Ask the boring question. What did it cost to mine those coins, and what did the company spend to keep hash rate climbing? If the answers aren't in the next update, that silence is information too.