Samsung SDS said on 2 November that it has started building an AI-only data centre in Gumi, North Gyeongsang province, with 60 megawatts of initial power capacity that can be expanded in stages to as much as 120 megawatts. The company called it the first visible result of Samsung’s plan, announced in July, to invest about 60 trillion won in the Yeongnam region.
Samsung SDS disclosed in January that it would invest 427.3 billion won in the new Gumi facility. By our arithmetic, that is about 0.7% of the 60 trillion won pledge.
The key insight: Samsung SDS said on 2 November it has started building a Gumi AI data centre with 60 megawatts of initial capacity, expandable to 120 megawatts, which it calls the first visible result of Samsung’s roughly 60 trillion won Yeongnam plan; the 427.3 billion won disclosed for it is about 0.7% of that pledge by our arithmetic.
The Gumi AI data centre sits on a site of about 14,000 pyeong in Gongdan-dong, Gumi, according to the Samsung SDS release. Gumi approved construction at the end of August 2026, work is under way, and completion is scheduled for the end of 2028, the company said.
Samsung SDS said the building uses hybrid cooling: air cooling for lower-heat areas such as network and storage, and liquid cooling, which circulates water, for dense GPU server areas. Server rooms are designed for racks of different power densities, and the company said it plans to support ultra-high-power racks later.
The design is not tied to a particular GPU or service, Samsung SDS said. It will run its own GPUs there and also offer colocation, in which customers install and use their own GPUs, to serve outside customers. The release also says the centre will support Samsung affiliates, including AI training for precision-work robots in their manufacturing robotics transformation (RX) projects.
Samsung’s roughly 60 trillion won Yeongnam physical-AI pledge (July 2026), its Q3 2026 capital expenditure of 29 trillion won, and the 427.3 billion won Samsung SDS disclosed for the Gumi AI data centre. Sources: Samsung SDS newsroom, 2 November 2026; Samsung Q3 2026 results as reported by FourWeekMBA. The 0.7% is our arithmetic; the scopes differ and are not additive.
The 60 Trillion Won Plan Behind It
Samsung announced the roughly 60 trillion won plan in July at a regional advanced-industry event at Gyeongsang National University in Jinju, with the aim of making Yeongnam a global physical AI innovation cluster, according to the release. The release lists four focus areas: humanoid robots and an AI data centre in Gumi, all-solid-state batteries in Ulsan, package substrates for AI servers and MLCCs (multilayer ceramic capacitors) in Busan, and high-value ships in Geoje.
Samsung SDS said the project is aligned with the government’s three mega-projects. It also signed a memorandum of understanding with North Gyeongsang province and Gumi city for the data centre, the release says.
Business Pill · Sovereign AI Explained
A short explainer on sovereign AI, countries building their own AI capacity. It teaches the general idea only and says nothing about any company in this story.
How the Pledge Compares
Samsung reported third-quarter capital expenditure of 29 trillion won, as we covered in Altman meets Samsung chiefs as capex hits 29 trillion won. By our arithmetic, the 60 trillion won Yeongnam pledge is about 2.1 times that single quarter. The two are different scopes: one is a multi-year regional pledge, the other one quarter’s spending.
As we read the release, the first piece of the 60 trillion won plan to break ground is compute, not robots or batteries: the data centre is the item Samsung SDS says will train robot AI for affiliates’ factories.
As we read it, the colocation offer matters as much as the size: Samsung SDS says customers can bring and run their own GPUs there, which makes the site a landlord for outside AI capacity as well as a host for its own.
Lee Jun-hee, CEO of Samsung SDS, in the release (Korean original; in English: AI is changing how companies work and how industries compete)
FOR SAMSUNG’S FACTORIES The release says the centre will support affiliates’ manufacturing robotics transformation (RX) work, including AI training for precision-work robots.
FOR OUTSIDE AI CUSTOMERS Samsung SDS says it will run its own GPUs and also offer colocation, in which customers install and use their own GPUs.
FOR THE REST OF THE PLEDGE The release lists all-solid-state batteries in Ulsan, AI-server package substrates and MLCCs in Busan, and high-value ships in Geoje as the other focus areas of the 60 trillion won plan.
The framework reads: “We are entering a phase where four scaling laws are simultaneously shaping the AI ecosystem, while several converging forces are radically transforming the physical layers underpinning AI itself.”
As we read it, Samsung’s Yeongnam plan is a bet on those physical layers: the release ties a data centre, robot training, capacitors and substrates to one regional pledge. The framework is the owner’s argument and does not discuss Samsung’s plan.
The release does not say how the rest of the 60 trillion won is split across the four areas, or over how many years. It does not name a GPU supplier, any colocation customer, or the cost of expanding from 60 to 120 megawatts.
The release gives a completion date of the end of 2028 but no date for when the first GPUs go into service.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Samsung SDS has started building a 60-megawatt AI data centre in Gumi, designed to reach 120 megawatts, and Samsung presents it as the first step of its 60 trillion won Yeongnam plan. The disclosed 427.3 billion won for the building is about 0.7% of that pledge, by our arithmetic.
A note on sourcing. This piece is based on the Korean-language Samsung SDS newsroom release of 2 November 2026, translated by us, and on Samsung’s third-quarter capex as we reported it on 31 October. The 0.7% and 2.1-times figures are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
