Amazon is moving further into the artificial intelligence arms race with the release of Trainium 3, a chip designed to compete with Nvidia’s dominant GPU hardware.
The new chip, available through Amazon Web Services (AWS), trains four times faster than previous versions while maintaining the same energy footprint. The move would pit the tech giant against Google and Nvidia as the battle for infrastructure heats up.
Each cluster of Amazon’s new “UltraServers” can run up to 144 Trainium 3 chips, enabling them to handle large-scale language model training and other compute-intensive tasks. The launch is part of Amazon’s broader efforts to expand its artificial intelligence infrastructure and reduce its reliance on others.
Amazon’s push, coupled with Google’s dominance in the AI model competition, now has an 87% chance of Google getting the best model by the end of the year, with OpenAI’s Sam Altman reportedly announcing “Code Red.”
Artificial Intelligence and Cryptocurrency
However, building more AI servers poses a problem that few tech giants can solve alone: finding enough power and space. This is where crypto miners who already have large data centers running step in, using some of their hardware to enter the AI arms race and profit from it.
Amid the arms race, and in the wake of the 2024 Bitcoin halving (where block rewards are cut in half), several large mining companies have begun repurposing their energy-intensive operations into AI-ready facilities. Companies like Core Scientific, CleanSpark, and Bitfarms are now viewed less as Bitcoin bets and more as utility providers to hyperscale enterprises.
Last month, Bitcoin miner-turned-new cloud company IREN (IREN) soared after signing a $9.7 billion artificial intelligence cloud deal with Microsoft (MSFT). Likewise, TeraWulf (WULF) signed a $9.5 billion AI infrastructure joint venture with Fluidstack, backed by Google.
These companies control gigawatts of power capacity, and existing infrastructure is ready for AI clusters that require advanced cooling and stable grid connections.
Bubble risk?
Still, transformation comes with risks.
Miners are borrowing heavily to retrofit sites for artificial intelligence workloads, putting pressure on related risk assets such as tech stocks and cryptocurrencies as investors grow wary of the sheer speed and cost scale behind “AI trading.”
Bitcoin It is down more than 17% in the past 30 days, while the broader CoinDesk 20 (CD20) index has lost 19.3% in value over the same period. The tech-heavy Nasdaq 100 has fallen about 1.5% over the past month, having recently recovered from a more than 7% loss in that period.
Analysts warn that the AI infrastructure boom has similarities to past bubbles. OpenAI, for example, has committed trillions of dollars in infrastructure spending but still needs to raise capital.
Much of the money invested in the AI arms race is recycled through the same players, selling AI chips or cloud services. Bain & Company predicts that if demand for artificial intelligence slows, these companies will be short of as much as $800 billion, and by 2030, these companies will need $2 trillion in total annual revenue to fund the computing power required by projected demand.
If demand for AI computing slows, these hybrid businesses could face the same liquidity crunch that plagued the crypto industry in 2022. Such a hit could impact broader markets, causing risk assets to fall sharply.
For now, though, miners are betting the future of their businesses on a new gold rush powered by GPUs, not ASICs.
