SK Telecom spins off AI data centers as KKR-backed SK Horizon

The AI build-out kept extending past the models themselves today: SK Telecom carved its data centers into a dedicated investment vehicle backed by KKR, while Kioxia moved to add a third Japanese fabrication plant for AI-driven memory demand. Two bets on the physical infrastructure underneath the boom.

SK Telecom announced it will spin off its data-center business into a new company called SK Horizon, bringing in a combined 3.08 trillion won (about $2.3 billion) of equity from KKR and a consortium led by South Korea's IMM Investment. Upon completion of the deal, KKR will hold a 29% stake and the IMM Investment–Stonebridge Capital consortium 20%, leaving SK Telecom with 51%, management control, and a clear separation between an AI-infrastructure unit and the rest of the telecom group. SK Horizon starts with 318 MW of operating capacity across eight existing sites in Seoul and its suburbs, plus the Ulsan and Guro facilities now under construction, and it also owns the submarine-cable network supplying overseas AI workloads.

The restructuring is the clearest sign yet that South Korea is treating AI data centers as strategic national infrastructure rather than a side business. SK Telecom is stacking SK Horizon (build and run) alongside SK Hyper (net-new projects), targeting 5 GW of phased capacity by 2029 and 15 GW by 2035, with the capital — and the valuation signal — coming squarely from global infrastructure money like KKR, which has more than $70 billion invested across digital and power assets. South Korea lacks the cheap domestic power and land that the US and parts of Asia enjoy, so cheap money from institutions betting on the AI build-out is what underwrites the ambition. The spin-off is expected to complete in the first quarter of next year, pending shareholder and regulatory approval.

Kioxia Holdings said it will build a new chipmaking plant in northern Japan, its third facility in Iwate Prefecture, in partnership with SanDisk to meet surging AI demand for memory storage. The move is a direct answer to the global NAND crunch that has been pushing up the cost of electronics, as AI training and inference require ever-larger pools of high-density flash. Kioxia is also racing to close the gap with rivals — the aggressive capacity push follows a period in which it lost ground to YMTC in NAND — and the new fab signals that memory makers see the AI storage shortage as durable enough to justify multi-year chip-plant construction. For anyone watching the AI supply chain, this is the demand for compute spilling into the storage layer: frontiers labs are hungry for more than just GPUs.

Do you think state-supported AI data centers will become the next stage of the global race — and how much is too much to spend? Tell us in the comments.

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day

Perplexity open-sources its agent sandbox after 118 failed escapes

Perplexity handed its agent containment layer to the open-source world, and Washington's AI task force stopped reading briefings and started working. Perplexity open-sourced SPACE on Thursday — the sandbox it built to keep its own AI agents from getting out — and says nine frontier models failed to break out of it across 118 attempts. Each agent runs inside a small virtual machine with its own operating system, with credentials and network access managed outside the sandbox so an agent can re

Read more

Brain organoids classified Braille at 83% — in a trio, not alone

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Amazon's new Bee devices listen all day — for $11.99 a month

Perplexity open-sources its agent sandbox after 118 failed escapes