Anthropic, the AI lab behind the Claude model, is no longer content with renting computing power. SK Group’s chairman Chey Tae-won revealed that the company is investing heavily in building its own data centers and even custom chips, aiming to break free from reliance on external providers. This strategy highlights a significant shift in AI infrastructure, with major players seeking full control over their hardware to tackle persistent supply shortages.
Massive $50 Billion Investment in AI Data Centers
Since November 2025, Anthropic has committed roughly $50 billion to create bespoke AI data centers in Texas and New York. These facilities are tailored specifically for AI workloads in partnership with Fluidstack, designed to handle the immense computing demands of next-generation models. By April 2026, reports emerged that Anthropic was moving further by developing proprietary ASICs and GPUs to reduce dependency on external chipmakers.
This trend toward vertical integration isn’t unique. Google’s TPU chips, Amazon’s Trainium, and Meta’s custom silicon represent similar efforts to control the end-to-end stack. Anthropic’s move aligns it with these tech giants, signaling a new era where hardware and software development happen hand in hand.
SK Hynix’s Strategic Role in the AI Hardware Ecosystem
Chey Tae-won isn’t just commenting from the sidelines. His semiconductor arm, SK Hynix, took part in Anthropic’s Series H funding round in May 2026 alongside Samsung and Micron. SK Hynix, recently valued at $26.5 billion after its Nasdaq debut, is a world leader in high-bandwidth memory (HBM), critical for AI model training. This investment cements a symbiotic relationship where SK Hynix supplies essential components as Anthropic builds out its custom infrastructure.
Chey’s long-term vision for SK Group has focused on AI infrastructure and semiconductor innovation, positioning the conglomerate at the crossroads of one of the fastest-growing tech sectors. This convergence shows how chip makers are becoming key players in the AI arms race.
Ripple Effects on Crypto Mining and Decentralized Compute
The competition for computing resources is already shaking up other industries. As AI labs like Anthropic spend tens of billions on data centers, they compete with crypto miners for GPUs, power, and cooling. Some mining firms have started pivoting to offer AI hosting services, chasing higher profit margins amid this resource crunch.
Decentralized GPU marketplaces like Render, Akash Network, and io.net are also impacted. These platforms try to alleviate supply shortages by creating tokenized compute markets. However, if major AI labs continue building proprietary infrastructure, the market opportunity for decentralized compute could shrink, reshaping the space.
