MARA Holdings, Inc. Q4 FY2024 earnings call

• Transformed to a vertically integrated energy and technology solutions provider, securing 300% more energy capacity to 1.7 gigawatts and deploying owned power generating assets to lower energy costs.
• Focused on developing solutions to optimize power consumption, storage, and distribution for data centers, AI operators, and energy markets.
• Intends to deploy 30 megawatts of inference AI using two-pick liquid cooling technology in 2025, with a focus on inference AI as the real profit area in AI.
• Aiming for near-zero cost of energy by owning energy assets and being in control of energy sources.

In Q4 2024, revenue increased 37% to $214.4 million from $156.8 million in Q4 2023. For 2024, revenues grew 69% to $656.4 million from $387.5 million in 2023. Net income in Q4 2024 was $528.3 million (up 248% from Q4 2023's $151.8 million), and full-year net income grew 107% to $541 million from $261.2 million in 2023. Adjusted EBITDA in Q4 2024 was $794.4 million (up from Q4 2023's $259 million), and full-year adjusted EBITDA was $1.2 billion vs $417.1 million in 2023. Direct energy cost per bitcoin in 2024 was $28,801, cost per kilowatt-hour for own sites was $0.039. Cost of revenue per petahash per day improved 5% in Q4 and 17% for the full year.

• Expect to accelerate vertical integration in 2025, with 50% of capacity from international markets by 2028.
• Focus on opportunistic growth in hash rate, prioritizing capital efficiency over just scale.
• Plan to grow internationally by partnering with energy companies in centrally controlled markets and domestically by acquiring energy assets to address rising grid energy costs.

• Geopolitical risks such as potential tariffs on Chinese-manufactured miners affecting cost competitiveness.
• Permitting and transmission challenges in the US limiting energy capacity and increasing costs.
• Cyclical nature of Bitcoin, which could impact profitability and hash rate growth.

Q: Brett Knoblauch asks about AI plans and co-location model.

A: Fred Thiel discusses focus on inference AI, stating the real profit in AI is in inference, and avoiding the commodity space of large data center build-outs.

Q: Brett asks about energy acquisition regions.

A: Fred Thiel explains international markets are centrally controlled, partnering with energy companies there, while domestically focusing on acquiring energy assets to get near-zero cost energy.

Q: Brett asks about Bitcoin holdings and capital allocation.

A: Fred Thiel explains using converts to buy Bitcoin as an appreciating asset, better than investing in depreciating miners, and Salman Khan adds on capital allocation diversification away from heavy reliance on ATMs.