Bitdeer - Update (Better execution)

The stock has done nothing for a year.

Micron reported today. S&D remains tight in DRAM//NAND in 2027 and even 2028. Long term agreements cover 1/3 of sales through 2030. Presumably, some customers have visibility through 2030.

Feels like AI capX spending to remain robust in 2027. With this in mind, where does it leave Bitdeer ?

1. An important development has been bottoming of bitcoin price..... Now we are 18 months away from the next halving in April 2028..... Potentially, BTC price may pick up into the halving. Bitdeer is very much leveraged to a higher BTC price as it has now becoming the largest BTC miner in the world. Just over 100EH in mining capacity. It mined 1310 BTC in Aug 26 (yearly run rate of 15700BTC a year). If/when BTC gets back to $100K, we are looking at BTC revenue of $1.5bn or so. Its energy cost/coin is around $45-50k. We can work out its gross margin and EBITDA potential.

2. The next part of the story is AI. Let's be honest, even now, folks are skeptical. So far, Bitdeer has won 3 deals in AI. It has signed up Volta/Anthropic for its campus in Tydal, Norway (121MW = $4.7bn over 16 years => $290M in revenue on average and NOI margin of 90%). The lease will commence in Dec 26 and Mar 27.

3. Bitdeer has signed up another 2 deals in Malaysia. $800M over 5 years for 9.5MW + 15MW for $1.7bn over 5 years. So we are looking at $2.5bn over 5 years. CapX require would be between $735 and $980M ($30-40M/MW). It typically asks for 50% prepayment. Bitdeer needs to put in $370-490M in capX (1/3 equity + 2/3 GPU finance). Also as a rule of thumb, these neocloud deals will seek GPU capX payback in 2-3 years. Think about the FCF in Yr3-5 for Bitdeer.

4. And it has another 65MW to go live in Malaysia in Q3 27. Plus 43MW in Tydal in 2H 2027. Note it also has another 65MW coming online in late 2027 in USA.

5. Note I think its position outside of USA is interesting as my hunch is the AI build out has been slower outside of USA and there are fewer credible partners (vs. USA is a shit fight with everyone under the sun + hyperscalers building their own power supply).

6. If we are to apply pricing for its 15MW, the remaining 108M in Malaysia and Tydal could yield $12bn over 5 years.

7. What is the catch ? 108MW will cost $4.3bn in capX... Or Bitdeer needs to fund $2.1bn of it assuming its customers put up 50% of GPU capX. BUT as we can see, ROI/FCF of these neocloud deals can be fantastic.

8. And then we have Texas. It just paid $100M to buy land in Texas next to its site in Rockdale... So it is working on a colocation deal for as much as 500MW in IT capacity.... If Tydal is any guide, Rockdale is 4X larger.

9. Bottom line is if Micron is right, AI spending to remain more than healthy in 2027, Bitdeer can potentially sign up customers for its different sites. Even signing another 100MW in neocloud for Malaysia and Tydal could yield some big numbers (big ROI) for Bitdeer in time.

10. Now.... As you know I am skittish re AI capX...... My take is as long as Bitdeer can sign the offtakes in Malaysia and Tydal, it will be fine.. as they are really about the credit worthiness of the counter parties. In this case, why I am particularly intrigued by Malaysia is because... Credit worthiness of Alibaba or Bytedance or even Kuaishou is potentially better than OpenAI or Anthropic... ha..... So theoretically, even when AI capX slows, Bitdeer can still extract its ROI from its neocloud investments.

11. What else we needs to see ? 1) I am waiting to see how much does Bitdeer need to borrow and how much it has to pay.... Wouldn't surprise me to see if Bitdeer has a bond issuance of $1-2bn @ 8-10% in the next few weeks. 2) In the next 3-6 months, let's see if it can sign offtakes for more neocloud deals + Rockdale co-lo. 3) Will it be on time with building out Tydal for its co-lo deal by Dec 26 ?

MV = $2.9bn..... Just its Tydal co-lo deal will yield net profit of $250-270M (121MW) a year in 2027. Plus its BTC business will be profitable at current price ($100K BTC price = sales of $1.5bn... gross profit of call it a $750M). Its 25MW neocloud deals will yield FCF of as much $1bn in the next 5 years.... And then it has another 100MW of neocloud deals to sign + 500MW co-lo in Texas. You can eyeball the upside potential.

The BIG unknowns = AI cycle + it needs to raise money to build out its neoclouds/co-lo sites.....

One more thing I don't know... Building a neocloud business is hard and technically demanding.. Have a look at its job site... https://bitdeer.breezy.hr/

If Bitdeer can pull off this transformation, it should be worth much more than the likes of Riot/MARA/Cleanspark and closer in terms of technical capabilities to Nebius and CoreWeave etc etc...

What is it worth ? It is in the middle of converting its low value/volatile BTC power leases into supposedly high value recurring AI DC rentals/co-lo with more credible counter parties than the BTC days.....