Samsung SDS, the Samsung group’s IT services company, is in early-stage talks with Anthropic and OpenAI to potentially sign them on as data centre tenants, the Financial Times reported on 15 November.
Analysts expect Samsung SDS to generate annual data centre revenue of Won6.6tn ($4.9bn) by 2031, “roughly six times current levels”, the FT says. By our arithmetic, that implies current annual data centre revenue of about Won1.1tn.
Business Pill · Vertical Integration Explained
An evergreen Business Engineer explainer on vertical integration, when one company controls several steps of a supply chain. It teaches the general idea only and does not discuss this story.
The key insight: As we read the FT report, the new element is the customer list: Samsung SDS, which mostly runs data centres for its own group, is in early-stage talks to bring Anthropic and OpenAI in as tenants, the FT says.
Samsung SDS began as an in-house IT services provider and “operates data centres mostly to support the group”, the FT says. The group hopes subsidiaries that still count Samsung as their primary customer can win outside clients as AI demand grows.
The FT describes a group plan to build data centres through construction arm Samsung C&T, operate them through Samsung SDS and supply energy storage through battery maker Samsung SDI, alongside Samsung Electronics’ memory chips. Samsung Heavy Industries is listed for potential floating data centres.
In the past year, the FT says, the group has invested $1bn in a KKR- and Nvidia-backed data centre operator, acquired a European cooling systems maker, which the FT names as FläktGroup, and signed a partnership with OpenAI. We covered the $1 billion Helix investment in September; separately, we reported talks to buy cooler maker Modine on 13 November.
Samsung SDS AI capacity plan (edaily and TheElec, via FourWeekMBA) against analysts’ expected data centre revenue rise to 2031 (Financial Times, 15 Nov 2026); multiples are our arithmetic.
The Numbers Against Samsung SDS’s Own Plan
Samsung SDS plans to take its AI infrastructure from about 110 megawatts to more than 800 megawatts by 2031, edaily and TheElec have reported, as we covered on 3 November. By our arithmetic, that is about 7.3 times, close to the roughly sixfold revenue rise analysts expect.
Combining the two, $4.9bn of annual revenue across 800 megawatts would be about $6.1 million per megawatt a year, by our arithmetic. The FT figure is an analyst estimate, and the capacity plan comes from Korean press reports, not a Samsung SDS filing.
Samsung SDS began building an AI-only data centre in Gumi on 2 November with 60 megawatts of initial capacity, expandable to 120 megawatts, the company said.
What Analysts Told the FT
Morgan Stanley said in a recent report that Samsung SDS was well positioned to benefit from a surge in data centre construction, the FT says. Citi analyst Peter Lee told the FT that Samsung “has been relatively weak at bringing them together”, referring to the group’s capabilities.
Sanjeev Rana, head of Korea research at CLSA, told the FT the group’s more immediate opportunity was designing systems inside data centres. “Owning or operating data centres, I am not sure actually that will play to their strengths”, he said.
The Structural Read
As we read the numbers, analysts expect revenue to rise about sixfold by 2031, the FT says, close to the roughly 7.3x capacity build Korean press has reported for the same date, by our arithmetic.
As we read the analyst comments, the FT quotes both sides: Morgan Stanley sees Samsung SDS well positioned, while CLSA says operating data centres may not play to the group’s strengths.
Samsung, statement after its $1bn Helix investment, September 2026, as quoted by the FT
“Moving forward, Samsung is poised to expand its role from a traditional hardware component supplier to an active architect of the global AI infrastructure ecosystem”
Three Implications
FOR SAMSUNG SDS Analysts expect Won6.6tn ($4.9bn) of annual data centre revenue by 2031, the FT says.
FOR ANTHROPIC AND OPENAI The talks are early-stage, and the FT reports no terms or capacity.
FOR THE GROUP C&T would build, SDS operate and SDI supply batteries, in the plan the FT describes.
The framework says: “The economy is only one layer of the story.”
As we read it, a chip maker moving into building and running data centres is one place where the economic layer shows up; the framework does not discuss this story.
What Is Not Established
The FT does not say how much capacity Anthropic or OpenAI might take, when talks began or whether either has agreed to terms. We found no statement from Samsung SDS, Anthropic or OpenAI on the talks. The FT does not name the analysts behind the Won6.6tn estimate.
This Week In The AI Supercycle
AI Lost Its Central Banker, the Lab Oligopoly, and the Future of Open Source
This week’s issue: the Nvidia leadership change, read as the AI build-out losing its central banker: who now finances the GPU landlords, backstops the data centres and sponsors open models, and why two labs are becoming the bank.
Samsung SDS is in early-stage talks with Anthropic and OpenAI to potentially sign them on as tenants, the FT reported on 15 November, as analysts expect its data centre revenue to reach Won6.6tn ($4.9bn) by 2031, roughly six times current levels.
A note on sourcing. The talks and the analyst revenue estimate come from the Financial Times report of 15 November 2026, read in full that day. The capacity plan comes from edaily and TheElec as cited in our 3 November report. Multiples and per-megawatt 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.
