Amazon's New Chip Challenge: Speeds Up AI Training, Takes On Nvidia

Amazon is taking a major step in the artificial intelligence (AI) race with the launch of Trainium 3, a chip positioned as a direct competitor to Nvidia's dominant GPU hardware. These new chips, available through Amazon Web Services (AWS), promise a fourfold increase in training speed compared to the previous version while maintaining the same power consumption. This initiative places the tech giant in direct competition with Google and Nvidia as the battle for AI infrastructure rages on.

Each cluster of Amazon's new "UltraServers" can support as many as 144 Trainium 3 chips, enabling them to handle large-scale language model training and other computational tasks. This launch is part of Amazon's broader strategy to expand its AI infrastructure and reduce its reliance on other players. The combination of Amazon's ambitious plans and Google's dominance in the AI ​​model race—where Google has an 87% chance of having the best model by the end of the year—has created a sense of urgency among OpenAI, with Sam Altman even declaring a “code red”.

The cross-pollination of AI and crypto

However, building more AI servers presents a challenge that only a few tech giants can solve on their own: finding sufficient power and space. This is where crypto miners come in, as they already have extensive data centers. They are dedicating some of their hardware to participating in the AI ​​race and profiting financially from it.

In the wake of this battle, and following the 2024 Bitcoin halving—which halved block production rewards—several large mining companies have begun repurposing their more energy-intensive operations into AI-friendly facilities. Companies like Core Scientific, CleanSpark, and Bitfarms are now increasingly seen as essential service providers for hyperscalers rather than simply as Bitcoin investors.

A notable development is the rise in shares of IREN, a Bitcoin mining company that has transformed into a neocloud firm, recently closing a $9,7 billion AI cloud deal with Microsoft. TeraWulf has also established a $9,5 billion AI infrastructure joint venture with Fluidstack, with backing from Google. These companies have gigawatts of capacity, with infrastructure ready for AI clusters that require advanced cooling and stable grid connections.

Risk of a bubble?

Analysts have warned that the rise of AI infrastructure bears parallels to previous bubbles. OpenAI, for example, has committed billions to infrastructure spending, but it still needs to raise funds. Most of the capital invested in the AI ​​race is recycled by the same providers that sell AI chips or cloud services. Should demand for AI decline, Bain & Co. has predicted a shortfall of as much as $800 billion for these companies, which would need $2 trillion in combined annual revenue by 2030 to finance the computing capacity required by the expected demand.

If demand for AI processing declines, these hybrid operations could face the same liquidity problems that plagued the crypto sector in 2022. Such a loss would also impact the broader market and push risk assets sharply lower.

For now, the miners are betting their business's future on a new gold rush, fueled by GPUs, not ASICs.

Frequently Asked Questions

How are Amazon's actions impacting the crypto market?
Amazon's expansion into AI infrastructure could impact the crypto market, especially as many crypto miners are restructuring their operations to capitalize on AI. This transformation could lead to new business opportunities, but also to increased competition and price pressure within the sector.

What does the urge to cooperate mean between crypto and AI for investors?
Investors can expect potential synergies between crypto and AI, which could lead to innovative products and services. On the other hand, this collaboration also carries risks, especially if demand for AI dwindles and companies face financial pressure.

Are we witnessing a new AI bubble?
Current trends in AI infrastructure resemble past bubbles, where companies make huge investments without assurance of a return on investment. This raises questions about the sustainability of the current hype and the potential consequences for the broader market if demand declines.