OPENING & CAST
Welcome to AI Global Investment Journey of BITDEER, a case study drama inspired by Bitdeer’s real evolution. Meet Helen, our investor; Donna, our chair; Akash, the CEO; Mason, Chief Strategy Officer; and me, Jaxon, your narrator. We will make three decisions: where to enter AI, how to transform from bitdeer’s business, and how to develop across different global markets.
ACT I — SPA CENTER
. At a luxury spa, two old friends step away from their meetings—but not from investing.
Donna, everyone’s talking about AI these days—robots, ChatGPT, chips, cloud… I’ve got way too much cash just sitting around. I want in.
First, AI is not one industry. Think of six connected layers. At the top are applications—AI agents, robotics and tools used by businesses. Below them are foundation models, the intelligence behind those applications. Both depend on cloud platforms and software that organize computing workloads.
Oh, so it’s not just about building another chatbot. Okay—so what’s under the cloud?
Then come the chips and servers that generate computing power—GPUs and other accelerators. They live inside AI data centers, which provide electricity, cooling, networks and reliable operations. At the very foundation are energy, land and grid access. Every layer creates value, but demands different capabilities.
ACT I — AI VALUE CHAIN & INVESTMENT QUESTION
As their conversation continues, Donna puts the AI industry into perspective.
So which one would you pick? Apps scale fast, models are the exciting bit, and chips—well, chips are basically printing money.
We should not confuse market size with our right to win. Applications need products and customers. Models need research talent and enormous compute. Chips require specialist design and supply chains. We already know power sourcing, construction and operating large facilities. That points us toward AI data centers—but we still need to prove the business case.
Got it. So the real question isn’t “What’s the hottest part of AI?” It’s “Where are we already good enough to win?”
Exactly. Tomorrow at headquarters, we ask the board what can transfer from crypto mining, what new capabilities we need, and who will pay us for AI computing.
ACT II — BOARD MEETING
Donna Called her waged slaves for a boarding meeting the next morning.
Akash, explain our current crypto mining economics. Mason, show us the opportunity and the risks. Helen, challenge our assumptions.
CRYPTO MINING vs AI DC — COMPUTE ECONOMICS
Our existing business is self-crypto mining. Electricity powers ASIC servers, the servers contribute computing work to the Bitcoin network, and successful block production earns BTC. We sell BTC at prefered market price. The network, not a contracted customer, determines the reward mechanics.
In an AI data center, electricity also becomes computing capacity. But the output is useful AI training or inference. An offtaker must pay for it. Customers include foundation-model labs training large models, hyperscalers leasing long-term capacity, GPU-cloud platforms reselling compute, and enterprises purchasing inference services.
Let me make sure I’ve got this. Bitcoin is basically a commodity—the market sets the price, and we just take it. AI is different. We have to find real customers, agree a price with them, and keep them happy with capacity and uptime.
Correct. And not every mining building can satisfy AI cooling, network, uptime and security needs.
AIDC BUSINESS MODELS
There are four investment models. Powered sites monetize grid-ready land and electricity access. Colocation rents reliable power, cooling and space under contract. Managed hosting operates customer-owned GPU systems. GPU cloud owns the expensive accelerators and charges for their use.
Hang on—why don’t we just buy the GPUs ourselves? Isn’t that where the real money is?
It can also be where capital intensity, obsolescence and idle-capacity risk are greatest.
Then start with what we can execute and sell. Do not confuse a fast-growing industry with a profitable investment.
CAPABILITIES & INVESTMENT GATES
From mining we bring energy procurement, construction, site operations and experience with round-the-clock infrastructure. We still need high-density cooling, GPU networking, stronger security, enterprise-grade SLAs, and commercial sales.
The right sequence is customer evidence, site feasibility, partner and compliance checks, then ROIC and NPV approval. We should fund information before we fund a massive GPU purchase.
Fine. So when do we get to go big?
Only when a credible offtaker, a viable site, and a risk-adjusted return clear the board’s gates. Approved: staged AIDC entry.
ACT III — CEO BUSINESS PLAN MEETING
Three months later. The board has approved exploration,. The CEO now presents global market strategies.
Our AIDC business plan Going China and Going West. Shared engineering strengths—but different customers, technologies, regulations and business models.
GOING CHINA vs GOING WEST
In China, we would first validate enterprise inference and industrial AI demand. We would partner with domestic accelerator vendors and software integrators, offering compatible and compliant managed infrastructure.
In Western markets, we would initially pursue power-ready colocation and managed facilities for cloud platforms, AI labs and large enterprise tenants. Long-term offtake contracts and dependable power availability are essential.
Why not just copy and paste the same setup in both markets? Surely that’s cheaper?
Because the ecosystem, customer preferences, chip access and security requirements are not identical. Standardize the repeatable capabilities; localize the offer.
GLOBAL EXECUTION & TRUST
Going global is now about navigating institutional rules, not only winning customers. Export controls can restrict specific chips, while investment reviews, data rules and security expectations affect operations. The details depend on the transaction, and laws can change.
We can share energy engineering and project governance, while keeping local hardware decisions, data boundaries, permits and operations compliant.
Professor Chen’s point is that digital trust should be verifiable. Do not market your way around a regulatory review. Build a system clients and regulators can inspect.
BITDEER REAL-WORLD CHECK
Now back to reality. Bitdeer’s publicly announced projects show that this transformation is already being pursued—but with different business models and financial metrics.
Tydal in Norway represents 121 IT megawatts tied to a 16-year lease with roughly 4.7 billion US dollars in initial contract value. Rockdale in Texas has 563 megawatts of existing site electrical capacity;.
In September 2026 Bitdeer reported a global AI Cloud pipeline of 333.5 IT megawatts, roughly 2.9 billion dollars of AI Cloud backlog and approximately 86 million dollars of annualized recurring revenue.
GEOPOLITICAL FRAGMENTATION — INDUSTRY OUTLOOK
Professor Chen’s framework brings our story together: 天时, 地利, 人和 — Timing, Strategy and People.
Opportunity opens the window. Strategy chooses the path. People create lasting value. Thank you.