Amazon Deepens AI Push Amid Market Concerns and Crypto Pivot

Amazon is intensifying its presence in the AI sector with the launch of Trainium 3, a new chip designed to rival Nvidia’s leading GPU hardware. The move comes amid growing concerns over heavy borrowing, sustainability, and potential shortfalls if AI demand slows.

Available through Amazon Web Services (AWS), the Trainium 3 chips deliver up to four times the training speed of their predecessor without increasing energy consumption. The chips will be deployed in Amazon’s new “UltraServers,” with each cluster capable of supporting up to 144 Trainium 3 units. The upgrade positions Amazon to compete directly with Google and Nvidia in the rapidly expanding AI infrastructure market, targeting large-scale language model training and other compute-intensive workloads.

Amazon’s AI expansion comes alongside Google’s continued dominance in model development, where it now holds an 87% likelihood of producing the leading AI model by year-end—a competitive edge that reportedly prompted OpenAI CEO Sam Altman to declare a “code red.”

However, scaling AI infrastructure introduces significant challenges. The high energy and space demands required for large AI clusters are difficult for most tech giants to address alone. Crypto mining firms, which already operate large-scale data centers, are increasingly repurposing their facilities to meet these needs and monetize the AI boom.

Following the 2024 Bitcoin halving, which halved block rewards, companies like Core Scientific, CleanSpark, and Bitfarms have pivoted from purely crypto-focused operations to AI-ready facilities. These firms are now seen less as speculative bitcoin bets and more as utility providers for hyperscale AI deployments.

IREN, a former bitcoin mining company now repositioned as a neocloud firm, saw its stock soar last month after securing a $9.7 billion AI cloud agreement with Microsoft. Similarly, TeraWulf partnered with Fluidstack on a $9.5 billion AI infrastructure venture backed by Google. These companies bring gigawatts of power capacity and infrastructure capable of supporting advanced cooling and stable grid requirements for AI clusters.

Market performance, however, has been mixed. Bitcoin has fallen more than 17% over the past 30 days, while the broader CoinDesk 20 (CD20) index dropped 19.3%. The tech-heavy NASDAQ 100 has been relatively resilient, down about 1.5% in the same period after recovering from a 7% drawdown.

Analysts caution that the current AI infrastructure surge could mirror past technology bubbles. OpenAI, for instance, has committed trillions to infrastructure spending but still needs to secure additional funding. Much of the capital fueling the AI arms race circulates among a small set of players offering AI chips or cloud services. Bain & Co. predicts that if AI demand decelerates, companies could face an $800 billion shortfall, requiring $2 trillion in combined annual revenue by 2030 to sustain projected compute needs.

A slowdown in AI compute demand could trigger liquidity issues reminiscent of the crypto crash in 2022, with repercussions likely spilling over into broader risk assets.

For now, crypto miners-turned-AI infrastructure providers are betting on a new digital gold rush—this time powered by GPUs rather than ASICs.