Amazon Advances in AI as Crypto Miners Pivot to Infrastructure
Amazon is ramping up its AI efforts with the launch of Trainium 3, a new chip aimed at challenging Nvidia’s GPU dominance. Available through Amazon Web Services (AWS), Trainium 3 delivers up to four times the training speed of its predecessor while keeping energy use constant.
The chips power Amazon’s new “UltraServers,” each capable of running 144 units simultaneously, enabling large-scale AI workloads such as language model training. This move puts Amazon in direct competition with Nvidia and Google, as the race for AI infrastructure intensifies.
Google currently leads in AI model development, with an 87% chance of producing the top model by year-end—a competitive edge that reportedly prompted OpenAI CEO Sam Altman to declare a “code red.”
Building massive AI infrastructure, however, is challenging. High energy and space requirements are obstacles most companies cannot solve alone. Crypto miners, with existing large-scale data centers, are stepping in to fill the gap. Firms like Core Scientific, CleanSpark, and Bitfarms are repurposing mining operations into AI-ready facilities, effectively acting as utility providers for hyperscale AI computing.
IREN, a former bitcoin miner turned neocloud firm, saw its stock surge after securing a $9.7 billion AI cloud contract with Microsoft. TeraWulf also signed a $9.5 billion AI infrastructure joint venture with Fluidstack, backed by Google. These companies control gigawatts of power capacity and data center infrastructure suitable for advanced AI workloads.
Market sentiment, however, is mixed. Bitcoin has fallen more than 17% in the past month, the CoinDesk 20 index dropped 19.3%, and the NASDAQ 100 is down roughly 1.5% after recovering from a deeper drawdown. Analysts warn that the AI infrastructure surge could echo previous tech bubbles. OpenAI’s multi-trillion-dollar infrastructure commitments depend on continued high demand, and Bain & Co. estimates that a slowdown could leave an $800 billion funding gap, requiring $2 trillion in annual revenue by 2030 to meet projected AI compute needs.
If AI demand slows, hybrid operations could face liquidity pressures reminiscent of the 2022 crypto crash, with potential ripple effects for broader risk assets.
For now, crypto miners-turned-AI providers are betting on a new digital gold rush—this time powered by GPUs rather than ASICs.
