Matt Brown the Go-To Guy at Nvidia’s Spring Event
At this year’s Nvidia-hosted spring technology conference, Matt Brown, a top executive at Core Scientific Inc., has quickly become one of the most sought-after attendees. Brown’s growing prominence is no coincidence; the AI boom has caused a clamor for data center space and the GPUs that power AI computations, making his insights invaluable.
From employees of Nvidia to representatives of Silicon Valley giants like Google and Amazon, and even to barely-known AI startups, it seems everyone wants to engage with the 46-year-old executive. Their primary question? Whether Core Scientific’s data centers have extra capacity to spare.
“I’ve been receiving unexpected calls and texts from industry peers searching for capacity, as they’re realizing there just isn’t any available,” Brown shared in an interview last week. “Everyone is scrambling to figure out where they can secure capacity in the next 12 months, and it’s becoming increasingly difficult to find.”
It’s not hard to understand why. The AI industry is experiencing unprecedented growth, leading to a severe shortage of data center space and GPU chips used for AI, as well as a struggle to secure sufficient power to run these operations. New facilities under construction are quickly snapped up. A March report from commercial real estate company CBRE GPUs Group highlighted that approximately 83% of data center capacity being built is already pre-leased, largely driven by demand from AI companies and cloud service providers.
AI Boom Fuels Data Center Growth and GPU Shortages
It’s an unfamiliar scene: data centers, traditionally thought of as the backbone of the internet, are now contending with an unprecedented surge in demand due to the AI boom. GPUs, which are essential for AI computations, are in tight supply. Matt Brown, the 46-year-old executive from Core Scientific Inc. has become a crucial figure at Nvidia’s spring technology conference. Representatives from tech giants like Google and Amazon, as well as smaller AI startups, are all keen to inquire about potential extra capacity in Core Scientific’s data centers.
“You wouldn’t believe the number of spontaneous calls and messages I get. Everyone’s scrambling to secure capacity for the next 12 months. It’s becoming increasingly hard to find,” Brown explained in an interview last week.
This surge in demand has led to a significant shortage of both data center space and the GPUs necessary for AI applications. According to a March report from CBRE Group, around 83% of data centers under construction have already been pre-leased, driven by demand from AI companies and cloud service providers. New facilities are swiftly reaching their capacity limits, intensifying the scramble for any available space.
The result is that many in the AI industry are now turning to cryptocurrency mining companies, which already possess the necessary resources and are seeking more profitable ventures. For instance, Core Scientific recently announced a partnership with AI startup CoreWeave, expected to generate about $3.5 billion in revenue over 12 years. This development underscores the growing trend of AI companies leveraging existing data center infrastructures designed initially for other purposes.
Although a Bloomberg report suggested that CoreWeave made an offer to acquire Core Scientific for approximately $1 billion, this claim was quickly overshadowed. However, the implications remain significant. “We see an opportunity to convert our infrastructure to host large GPU arrays, becoming the primary providers for AI clients,” said CEO Adam Sullivan before the acquisition news surfaced. Sullivan declined to comment on the acquisition report released on Tuesday.
In this environment, it isn’t surprising that companies like Core Scientific are garnering acquisition interest. Such developments indicate that the market recognizes the strategic value of converting existing data center capacity for AI use. As companies like TeraWulf Inc. see their stock prices surge, it’s clear that investors are optimistic about the prospects of this pivot.
Core Scientific’s New Path Moving from Crypto Mining to AI Dominance
Core Scientific’s pivot comes at a critical juncture. Traditional crypto mining, while still lucrative, faces increasing difficulty due to factors like currency volatility and regulatory changes. By shifting focus to AI, the company taps into a rapidly growing market with seemingly insatiable demand for computational power.
Adam Sullivan, Core Scientific’s CEO, underscores this strategic move by emphasizing the seamless adaptability of their existing infrastructure. “Our facilities are built to support high-density mining operations, which isn’t too different from the requirements for large-scale AI computation,” he explains. This transition essentially repurposes the same robust data centers, bolstered by sophisticated cooling and power systems, to meet AI clients’ specific needs.
The collaboration with CoreWeave exemplifies how Core Scientific plans to harness this infrastructure. This partnership aims to generate an estimated $3.5 billion over 12 years, providing significant financial stability and growth potential. Bloomberg’s report on CoreWeave’s interest in acquiring Core Scientific for around $1 billion further suggests the viable and lucrative path down which the AI transformation could lead.
Moreover, Core Scientific has mapped out around 500 megawatts of data center capacity for potential conversion. This magnitude of infrastructure is projected to become one of the world’s largest AI-dedicated GPU installations. To put this in perspective, that amount of electricity could power hundreds of thousands of households for a year, a testament to the scale and ambition of this venture.
The broader market response has been telling. As news of these developments surfaced, industry stocks reacted positively. For instance, TeraWulf Inc.’s shares jumped 22% in one day—a clear indication of investor confidence in the emerging synergy between crypto mining capabilities and AI-driven applications.
While the transformation promises great rewards, it also comes with its set of challenges, primarily related to securing adequate energy resources. Building the necessary infrastructure, including constructing substations to channel power from the grid, is an intricate and lengthy process, often taking several years to complete. Expediting these timelines can incur substantial costs. According to Morgan Stanley, data center developers could pay a premium of 101% on electricity prices to outpace competitors by two years, assuming a six-year GPU economic lifespan. Even with a ten-year lifespan, the premium remains significant at 61%.
This high demand for quick turnarounds makes the strategy of repurposing existing crypto mining facilities particularly attractive. These sites are already energy-intensive and outfitted with the requisite power capabilities, making the switch to AI usage both logical and economically viable.
As similar crypto mining companies navigate this AI boom, many are adopting diversified strategies. Some are opting to lease their existing data center capacities to AI companies, while others purchase specialized hardware to directly sell computational power. Companies like Northern Data AG, Bit Digital Inc., and Hive Digital Technologies Ltd. have already invested heavily in high-demand Nvidia chips to cater to AI’s computational needs.
Brown wraps it up succinctly, “This evolution in our business not only aligns with market demands but also positions us at the forefront of a technological revolution. It’s an exciting time to be in the data infrastructure industry.” As Core Scientific, along with its peers, adapts to these seismic shifts, one thing becomes clear: AI is not a just future prospect but a present reality reshaping industries far and wide.
