SK Hyper Seeks $350 Billion for South Korea AI Data Center Buildout as Gartner Flags Revenue Gap

SK Hyper Co., the data center unit spun out of SK Telecom within South Korea's SK Group, is in active talks with big tech firms and financial investors to raise capital for what its chief executive is pricing at $350 billion, and that figure covers only the first phase.

According to an interview reported by Bloomberg on October 23, CEO Chung Suk-geun told Bloomberg Television that <cite index="12-3,12-4">SK Group's fledgling data center arm has begun discussions with potential partners to help finance a plan to build 15 gigawatts of data center capacity in South Korea by 2035, and the first 5-gigawatt phase alone would require $350 billion including equipment.</cite> To put the equipment cost in context, <cite index="11-4,11-5,11-6">that figure is higher than what SK Telecom, SK Hyper's parent, published in July, when a release said a typical 1-gigawatt AI data center may require approximately 70 trillion won, or around $45.8 billion, with costs driven by computing infrastructure and rising memory prices.</cite>

The financing plan is staged. <cite index="13-9">Chung told Bloomberg the company aims to close its first round of external financing by the end of this year or early 2027, then keep raising over the following three to four years.</cite> On sharing the burden, <cite index="16-12">the CEO said SK plans to share the total cost with data center tenants and outside investors.</cite>

The demand thesis SK Hyper is selling to potential backers is geographic. <cite index="17-5">The plan is to draw global companies' AI infrastructure into South Korea and capture AI computing demand from elsewhere in Asia, including Japan and Taiwan, turning the country into a global AI infrastructure hub.</cite> Chung was specific about what he's chasing: <cite index="17-9,17-10">SK Hyper is in talks with global hyperscalers about bringing AI inference hubs to South Korea, and also aims to attract the large-scale computing infrastructure needed to train AI models.</cite>

Chung didn't pretend the path is clear. <cite index="17-7,17-8">"For the first 5GW project, we don't expect much difficulty securing power and sites," he said. "Securing actual demand in the global market and raising the capital needed to build the data centers will be the biggest challenges."</cite>

That candor lands against a backdrop that Gartner made considerably darker nine days earlier. In a press release published October 14 and reviewed by TechWire Asia and others, <cite index="19-3">Gartner warned that more than $3.5 trillion of AI infrastructure spending could be exposed to impairment by the end of 2030 if AI providers continue their current pace of capital spending.</cite> The math Gartner lays out is blunt: <cite index="20-8,20-9">to cover announced capital spending, financing costs, and a 12% return on invested capital, AI data center owners would need $8.5 trillion in cumulative revenue through 2030, but Gartner forecasts no more than $3.6 trillion in enterprise and consumer AI revenue attributable to those data centers over the same window.</cite>

The pain won't be distributed evenly. <cite index="20-13">Gartner expects more than 80% of neocloud providers, the GPU-as-a-service tier below the hyperscalers, to cease independent operations by 2030.</cite> <cite index="19-5">"The growth in spending on AI-optimized IaaS will be too low for most AI compute providers to recover their AI investments before the current generation of AI semiconductors is replaced," Gartner Senior Director Analyst Will Sommer said.</cite>

The two stories sit in direct tension. SK Hyper is betting that sovereign demand, Asian regional spillover, and hyperscaler appetite for inference capacity outside the U.S. will fill its buildings. Gartner's model says the industry as a whole is building far faster than revenue can follow. The distinction that matters is whether SK Hyper can pre-sell enough committed capacity to escape the spec-build trap that's swallowing the neocloud tier.

<cite index="16-13,16-14">The fundraising comes as investor sentiment has soured, with renewed debate over the profitability of AI and spreading concerns about credit risk tied to data centers. "Rather than pushing the entire plan at once, we will adjust the pace of investment according to actual demand," Chung said.</cite> That's the right thing to say to a nervous LP. Whether the demand materializes before the first capital call is due is a different question entirely.

Sources cited:
- Bloomberg (https://www.bloomberg.com/news/articles/2026-10-23/sk-data-center-arm-in-talks-to-fund-350-billion-ai-buildout)
- Seoul Economic Daily (https://en.sedaily.com/technology/2026/10/23/sk-hyper-seeks-350-billion-to-build-15gw-of-ai-data-centers)
- Gartner press release (https://www.gartner.com/en/newsroom/press-releases/2026-10-14-gartner-says-over-us-dollars-three-and-a-half-trillion-of-artificial-intelligence-infrastructure-spending-could-be-exposed-to-impairment-by-the-end-of-2030)
- TechWire Asia (https://techwireasia.com/2026/10/gartner-ai-infrastructure-spending-risk/)