Samsung SDS plans to take its AI infrastructure from about 110 megawatts to more than 800 megawatts by 2031, Korean business daily edaily reported in July, and TheElec reported on 3 November that the company repeated the 800 megawatt goal on its 29 October earnings call. That is about 7.3 times the base, by our arithmetic. In the same week, LG CNS said it had completed a 30 megawatt AI data centre in Jakarta, the first phase of a site it now plans at 400 megawatts.
Lotte Innovate, the third of Korea’s big group IT-services companies, has taken on the operation of a data centre of about 40 megawatts in Ansan, MTN reported in October. The three companies are moving into AI data centres in three different roles: owner, builder and operator.
Samsung SDS reported third-quarter revenue of 3.7725 trillion won, up 11.2% from a year earlier, in its results release of 29 October. Cloud revenue rose 18% to 796.1 billion won, the release says. Within cloud, its CSP business, which rents out computing on its own cloudplatform and GPU as a service, grew 15%. Its MSP business, which runs cloud systems for clients, grew 25%.
The same release says Samsung SDS was named preferred bidder for a design-build-operate (DBO) data centre project of more than 200 megawatts for Pacific Asset Management. In a DBO deal the IT company designs, builds and runs a data centre that another party owns. The release adds that the company plans to grow colocation and other data centre work to raise profitability.
On 2 November, Samsung SDS said it had started building its own AI data centre in Gumi, with 60 megawatts of initial power capacity that can be expanded in stages to 120 megawatts. TheElec reported that completion is targeted for the end of 2028 and operations for March 2029. The board approved 427.3 billion won for the project in January, and TheElec wrote that the sum excludes GPU servers.
Announced capacity of four named AI data centre sites linked to Korea’s group IT-services firms, in megawatts: the Jakarta site LG CNS builds for KMG (30MW completed, 400MW after later phases), the Pacific Asset Management design-build-operate project where Samsung SDS is preferred bidder (more than 200MW), Samsung SDS’s own Gumi site (60MW initial, up to 120MW) and the Ansan Global Cloud Center Lotte Innovate will operate (about 40MW, as reported by MTN). The dashed line is the roughly 110MW of AI infrastructure edaily reported Samsung SDS running in July. Sites are not summed because the companies’ roles differ; the comparison with 110MW is our arithmetic.
Business Pill · Lead Time Explained
A short explainer on lead times: why big infrastructure takes years from plan to operation. It teaches the general idea only and says nothing about the companies in this piece.
The key insight: Samsung SDS plans to take its AI infrastructure from about 110 megawatts to more than 800 megawatts by 2031, about 7.3 times, by our arithmetic, while LG CNS has completed the first 30 megawatts of a Jakarta site planned at 400 megawatts and Lotte Innovate has taken on the operation of an Ansan site of about 40 megawatts, as reported by MTN.
What LG CNS Completed
LG CNS said on 2 November that it had finished building a 30 megawatt hyperscale AI data centre in Jakarta. The client is KMG, a joint venture of Indonesia’s Sinar Mas Group and Korea Investment Real Asset Management. LG CNS said customers had signed up for the full capacity before completion, and called it the first time a Korean company had built an AI data centre for an overseas client in that client’s country.
The release says LG CNS is designing the later phases, after which the site’s total power intake would reach 400 megawatts. TheElec reported that LG CNS first described a 220 megawatt long-term plan, so the new figure is about 82% higher, by our arithmetic. By the same arithmetic, the 30 megawatts built so far is 7.5% of the 400 megawatt plan. TheElec also reported the main construction contract at 100 billion won.
The power system supports racks of more than 300 kilowatts each, which LG CNS says is about 60 times what a typical data centre rack needs. In its own third-quarter release, LG CNS reported AI and cloudrevenue of 961.5 billion won, up 9.3%, and named colocation revenue at the AI data centres it runs for others as one driver.
What Lotte Innovate Took On
Lotte Innovate has won the contract to operate the Global Cloud Center in Ansan, a site of about 40 megawatts ordered by IGIS Asset Management, MTN reported on 16 October. The report says the centre is due for completion in 2027 and that Lotte Innovate will run it for five years. IT firm ITCEN CTS said on the same day that it had signed an agreement to support Lotte Innovate on the site, Byline Network reported.
MTN described the shift in Lotte Innovate’s business weight from building and managing IT systems toward data centre infrastructure, and reported a partnership with Vietnam’s CMC Group on 13 October to look for data centre projects there. The report noted that the overseas work is at an early stage.
Four Sites on One Axis
Put on a single axis in megawatts, the four named sites run from about 40 megawatts in Ansan to 400 megawatts in Jakarta. The chart shows site capacity only and does not add the sites together, because each company plays a different role: LG CNS builds for a client, Samsung SDS owns Gumi and would design, build and operate the Pacific project, and Lotte Innovate operates a site it does not own.
At its full 120 megawatts, the Gumi site alone would be about 109% of the roughly 110 megawatts that edaily reported Samsung SDS running in July, by our arithmetic. Its 60 megawatt first phase is about 55% of that base, by our arithmetic. As reported by edaily, Samsung SDS’s path runs from about 110 megawatts to 230 megawatts by 2029, about 2.1 times by our arithmetic, and then to more than 800 megawatts by 2031.
As we read the releases, the three companies are entering AI data centres through different doors. Samsung SDS owns Gumi and offers its own GPUs and colocation there; LG CNS builds for a client, KMG, and runs centres for others in Korea; Lotte Innovate is to operate a site ordered by IGIS Asset Management.
As we read it, the Gumi numbers show the scale of the step for Samsung SDS: at its full 120 megawatts, the one site would be about 109% of the roughly 110 megawatts that edaily reported the company running in July, by our arithmetic, and TheElec reported that the 427.3 billion won approved for it excludes GPU servers.
TheElec, 3 Nov 2026
“The approved investment covers construction of the data center building and installation of certain facilities. It excludes additional AI computing equipment, such as graphics processing unit (GPU) servers, meaning total investment could increase as the project expands”
Three Implications
FOR READING THE MEGAWATTS The four sites are not added together: LG CNS builds Jakarta for KMG, Samsung SDS owns Gumi and is preferred bidder on the Pacific design-build-operate project, and Lotte Innovate operates Ansan, so each figure measures a different role.
FOR THE CLOUD LINES Samsung SDS reported cloudrevenue of 796.1 billion won, up 18%, and LG CNS reported AI and cloudrevenue of 961.5 billion won, up 9.3%; the two companies define these segments differently.
FOR THE TIMETABLE TheElec reported Gumi completion for the end of 2028 and operations from March 2029; MTN reported Ansan completion in 2027; LG CNS said it is designing Jakarta’s later phases.
The framework reads: “the AI economy is not a list of companies racing each other. It is a layered industrial stack”.
As we read it, Korea’s IT-services firms are placing themselves in the physical layer of that stack, as owners, builders and operators of data centres, next to the cloud and software work they already sell. The framework was published in May 2026 and does not discuss these companies.
The 110 megawatt base and the 2029 and 2031 goals come from edaily’s July report and TheElec’s account of the call, not from a Samsung SDS release we read. The companies do not say how much revenue the data centre work brings in on its own, and their segment definitions differ, so the cloud figures are not a like-for-like comparison. The Pacific project is at preferred-bidder stage, not a signed contract, and its capacity is stated only as more than 200 megawatts. The Ansan capacity is as reported by MTN. We did not contact the companies.
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.
Korea’s three big group IT-services firms are each taking a role in AI data centres: Samsung SDS as owner and builder, LG CNS as builder for others, and Lotte Innovate as operator. Samsung SDS’s stated goal of more than 800 megawatts by 2031 is about 7.3 times the base edaily reported in July, by our arithmetic, and its Gumi site alone would exceed that base.
A note on sourcing. Samsung SDS and LG CNS figures are from their own releases of 29 October and 2 November 2026 and LG CNS’s third-quarter release, in Korean (our translation). Completion dates, the 800 megawatt call plan and the earlier 220 megawatt plan are as reported by TheElec on 3 November; the 110 megawatt base is as reported by edaily on 31 July; Lotte Innovate’s contract is as reported by MTN and Byline Network on 16 October. All ratios 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.
