Bitcoin Mining/Data Center Development: 1Q26 Earnings Updates for WULF, CIFR, and MARA

Research evidence excerpt

Bitcoin Mining/Data Center Development: 1Q26 Earnings Updates for WULF, CIFR, and MARA

FoundationMtransactions.

• Risk of data center overbuild due to a number of potential drivers,

including slower AI adoption rates, more efficient AI models, and quantum

computing breakthroughs that reduce the volume of compute needed from

classical computing. We expect to see continued innovation in LLM

architecture that results in more efficient usage of compute, but expect

these efficiency benefits to continue to be outweighed by a combination of

(i) a proliferation of AI users and use cases and (ii) more complex usage of AI

tools (the average duration of an Agentic AI task — the length of time an AI

tool can independently perform a complex, multi-step task — has been

doubling every 7 months). If data centers end up as an overbuilt asset, the

cash flow multiple applied to this asset could be significantly lower than

20x, we think a legacy telecom multiple of 10x would be likely in this

scenario. On the other hand, if AI spend accelerates further, the US has a

significant power deficiency, which could result in a higher strategic value to

data center assets.

• Crypto market cyclicality, and the upcoming Bitcoin "halving":

Cryptocurrencies remain a volatile asset class as measured by both price and

trading volume, which tend to be correlated. If the crypto market enters a

downswing, the price of Bitcoin could fall faster than the reduction in the

cost to mine Bitcoin; as a result, Bitcoin mining assets (which are the primary

driver of MARA's valuation) would experience lower margins. In addition, the

upcoming Bitcoin "halving," which will next occur in 1H28, could negatively

impact margins for Bitcoin mining companies. MARA CEO Fred Thiel