Skip to main content

From Ethereum Mining to AI Hyperscaler: How CoreWeave Became the Backbone of the AI Revolution

· 8 min read
Dora Noda
Dora Noda
Software Engineer

In 2017, three Wall Street commodities traders pooled their resources to mine Ethereum in New Jersey. Today, that same company—CoreWeave—just received a $2 billion investment from Nvidia and operates AI infrastructure worth $55.6 billion in contracted revenue. The transformation from crypto mining operation to AI hyperscaler isn't just a corporate pivot story. It's a roadmap for how crypto-native infrastructure is becoming the backbone of the AI economy.

From Ethereum Mining to AI's Backbone​

CoreWeave's origin story reads like a crypto industry fable. Founded in 2017 as Atlantic Crypto by Michael Intrator, Brian Venturo, and Brannin McBee—all former commodities traders—the company initially focused on mining Ethereum using GPUs. The 2018 crypto crash, which saw ETH plummet from $1,400 to under $100, forced a existential reckoning.

Rather than doubling down on mining or exiting entirely, the founders recognized something crucial: their massive GPU inventory could serve markets beyond cryptocurrency. In 2019, they rebranded as CoreWeave and began renting compute to visual effects studios and 3D rendering companies.

The timing proved prescient. When OpenAI released ChatGPT in late 2022 and generative AI exploded into mainstream consciousness, CoreWeave was uniquely positioned. Unlike Amazon Web Services (AWS) or Microsoft Azure, which had to retrofit general-purpose data centers for AI workloads, CoreWeave had built high-density GPU clusters from the ground up—facilities designed specifically for the thermal and power demands of massive AI training.

The results speak for themselves:

  • 2022 revenue: $16 million
  • 2023 revenue: $440 million (2,650% increase)
  • 2024 revenue: $1.92 billion (337% increase)
  • 2026 projected revenue: $11.6 billion

Nvidia's $2 Billion Vote of Confidence​

On January 26, 2026, Nvidia announced a $2 billion investment in CoreWeave at $87.20 per share, bringing its stake to 11% and making it the second-largest shareholder. This wasn't Nvidia's first bet on CoreWeave—the GPU giant had previously invested $100 million in April 2023—but the scale signals something profound about AI infrastructure dynamics.

Jensen Huang, Nvidia's CEO, framed the investment in near-civilizational terms: "High-performance computing data centers are the foundation of the AI industrial revolution. AI is entering its next frontier and driving the largest infrastructure buildout in human history."

The deal includes more than capital. CoreWeave will be among the first to deploy forthcoming Nvidia products, including new storage systems and CPUs. Nvidia has also committed to purchasing over $6 billion in services from CoreWeave through 2032.

Why would Nvidia, which could partner with any hyperscaler, invest so heavily in CoreWeave? The answer reveals the strategic chess game underlying AI infrastructure.

The Shadow Cloud Strategy​

Amazon, Google, and Microsoft control the vast majority of cloud computing. But each is also developing proprietary AI chips—Amazon's Trainium, Google's TPU, Microsoft's Maia. When these hyperscalers push their internal silicon over Nvidia's GPUs, Nvidia needs alternatives.

CoreWeave represents Nvidia's "shadow cloud"—a GPU-first infrastructure provider that will always prioritize Nvidia hardware. By investing in CoreWeave, Nvidia ensures it has a massive, growing customer that won't substitute its chips for proprietary alternatives.

The strategy is working. CoreWeave's contracted backlog reached $55.6 billion by Q3 2025, nearly double the previous quarter. Major contracts include:

  • OpenAI: $22.4 billion total contract value for training next-generation models
  • Meta: $14.2 billion for AI cloud infrastructure
  • Microsoft: 62% of 2024 revenue came from Microsoft Azure overflow

Even the hyperscalers CoreWeave ostensibly competes with use its services as a "pressure valve" when their own capacity falls short.

The Power Bottleneck Crisis​

CoreWeave plans to use Nvidia's investment to "speed up land acquisition and power access for new data centers across the United States," targeting over 5 gigawatts of AI data center capacity by 2030.

That focus on power isn't incidental—it's existential. The AI industry faces a crisis that no amount of GPU production can solve: there isn't enough electricity.

Microsoft CEO Satya Nadella captured the problem bluntly: "The biggest problem we face right now is not a shortage of compute, but a shortage of power. My issue today isn't chip supply—it's that I don't have facilities with sufficient power and cooling to deploy those chips."

The numbers are staggering:

  • 2024 US AI data center power: 4 gigawatts
  • 2035 projected demand: 123 gigawatts (30x increase)
  • Time to build a data center: Less than 3 years
  • Time to build power generation: 3-6 years for solar/wind, ~6 years for gas turbines

By 2026, approximately 70% of the US power grid is approaching end-of-life, having been built between the 1950s and 1970s. The infrastructure that powered the industrial age cannot support the AI age without massive upgrades.

This is where crypto miners' unique assets become valuable.

Bitcoin Miners: The Unexpected AI Infrastructure Play​

CoreWeave isn't the only crypto company pivoting to AI. The April 2024 Bitcoin halving reduced block rewards to 3.125 BTC while network difficulty hit 126.98 trillion in May 2025, pushing mining costs to an estimated $70,000 per BTC. Miners with strong energy portfolios had a choice: compete in an increasingly brutal mining market or leverage their infrastructure for AI.

Several have chosen AI:

Core Scientific rejected CoreWeave's $9 billion acquisition offer in 2025, betting that independent operation would yield higher returns. The company converted mining facilities to AI hosting and is now projecting billions in new contracts for 2026.

Hive Digital Technologies (formerly Hive Blockchain Technologies) announced plans in 2023 to redirect computing resources toward AI workloads.

TeraWulf struck a deal with Core42 in December 2024 to repurpose infrastructure for high-performance computing.

The thesis is straightforward: Bitcoin miners already solved the hardest problems in AI infrastructure. They have:

  • Relationships with utilities and access to cheap power
  • Experience operating high-density compute facilities
  • Land and permitting for energy-intensive operations
  • Cooling systems designed for continuous high-load operation

However, Nvidia's deepening partnership with CoreWeave creates competitive dynamics. When Nvidia invested $2 billion in CoreWeave on January 26, 2026, most Bitcoin miner stocks fell. Analysts suggested Nvidia's preferential relationship could "divert GPU access and funding away from independent miners trying to pivot into AI."

The $500 Billion Infrastructure Buildout​

The scale of AI infrastructure investment defies historical precedent. Amazon, Microsoft, Google, and Meta collectively poured an estimated $320 billion into AI infrastructure in 2025 alone. The Stargate Project—a $500 billion, multi-year initiative involving OpenAI, Nvidia, Microsoft, and CoreWeave—aims to deliver up to 10 gigawatts of AI-ready power, equivalent to the combined consumption of New York City and San Diego.

CoreWeave's capex is expected to exceed $30 billion in 2026, more than double 2025 levels. The company will help build AI factories with up to 120,000 GPUs operating in the UK alone.

Global AI infrastructure spending is projected to reach $375 billion in 2025 and $500 billion by 2026. By end of 2026, around 10 gigawatts of new IT load will serve generative AI workloads, requiring 13-15 million GPUs deployed globally.

What This Means for Crypto​

CoreWeave's trajectory reveals several dynamics relevant to the broader crypto ecosystem:

Infrastructure arbitrage works. The same GPUs that mined Ethereum in 2017 now train frontier AI models. Crypto miners who built robust power and compute infrastructure have optionality that pure software plays lack.

Power is the new hashrate. Just as Bitcoin mining gravitated toward cheap electricity, AI training follows power. Regions and companies with energy advantages will capture disproportionate value.

Crypto-native talent translates. The skills required to operate large-scale distributed computing—thermal management, power optimization, hardware lifecycle management—apply directly to AI infrastructure.

The hyperscaler oligopoly has cracks. CoreWeave's rise proves that specialized providers can compete with AWS, Azure, and GCP when they offer differentiated capabilities. The same may prove true for decentralized compute networks.

The Road Ahead​

CoreWeave's stock has surged 146% since its March 2025 IPO, though it remains 46% below its all-time high from June. Deutsche Bank upgraded the company to "buy" following Nvidia's investment, setting a $140 price target—42% above current levels.

The company faces real risks. Two clients (Microsoft and OpenAI) account for 77% of revenue. Microsoft may eventually bring AI workloads in-house as Azure capacity expands. The GPU shortage that boosted CoreWeave could ease as Nvidia and competitors ramp production.

But the broader trend is unmistakable: crypto infrastructure is merging with AI infrastructure, and the companies that built for one revolution are now building for another. CoreWeave started as three traders mining Ethereum in New Jersey. Eight years later, they're at the center of what Jensen Huang calls "the largest infrastructure buildout in human history."

The lesson for crypto builders is clear: infrastructure advantages compound. The power deals, hardware expertise, and operational knowledge accumulated during crypto's growth phases have value far beyond mining or blockchain protocols. In an AI-hungry world desperate for compute, those assets may prove more valuable than anyone imagined.

BlockEden.xyz provides enterprise-grade blockchain infrastructure for developers building the next generation of decentralized applications. As compute becomes the critical resource across both crypto and AI, reliable infrastructure partnerships become essential. Explore our API marketplace to build on foundations designed for scale.

Share on Twitter