Samsung and SK Hynix just refused to prepay $18.7B of AI power bills

Share
Samsung and SK Hynix just refused to prepay $18.7B of AI power bills

The pacing debate has been argued in Washington and Tokyo for three days. On Monday the most useful verdict came from neither: it came from two chipmakers looking at a five-year electricity bill and saying no.

Samsung Electronics and SK Hynix both rejected Korea Electric Power Corp.'s request to prepay roughly 25 trillion won (about $18.7 billion) of electricity, citing doubt over whether the AI chip boom holds. Industry sources told Reuters that KEPCO had asked Samsung for 20 trillion won and SK Hynix for 5 trillion won — around five years of bills at last year's payment levels — as a way to front-load financing for the semiconductor clusters under construction at Yongin, south of Seoul, and in the southwestern Honam region. Both companies completed internal reviews and notified the utility they could not accept, reasoning that committing that much cash upfront is hard to defend while long-term demand for AI silicon is genuinely unknown. The context makes the refusal sharper than it looks: KEPCO ended June with 210.7 trillion won of debt and was paying 11.5 billion won a day in interest, which is why it wanted prepayment at all. The read is that the caution is no longer sitting with the safety people. The two firms selling the memory that every AI accelerator needs have looked at the demand curve five years out and declined to underwrite it — a much harder signal than an op-ed, because it costs them something to say no.


SoftBank closed a two-year loan facility for its OpenAI stake, and the banks oversubscribed it. Bloomberg reports that SoftBank secured $11.87 billion in commitments from around 20 lenders against a $10 billion target for the bridge financing that supports its OpenAI investment. The number worth dwelling on is not the size but the tenor: two years. Capex on this scale is normally financed against a decade of assumed demand, and the market is currently offering roughly one-fifth of that horizon to the single most exposed name in AI. SoftBank Group was among the heaviest drags when the Nikkei broke below 63,000 on Monday morning — we tracked that session here — Nikkei breaks 63,000 as the AI slowdown debate hits the market. Put the two Monday stories side by side and the shape of the risk is clear: lenders will fund an AI bet on a two-year clock, and the chipmakers supplying it will not prepay for five. Nobody in that arrangement is arguing about safety. Everyone is arguing about the shelf life of the demand.

What to watch: whether KEPCO's Yongin power plan slips its construction schedule now that the prepayment is off the table, and whether SoftBank returns to the market for a longer tenor at a worse price.

The labs can't agree on how fast to build models, but Korea's chipmakers and Japan's banks have already agreed on how long the demand lasts. Which of those two signals do you trust? Tell us in the comments.

Read more

Akhetonics says its all-optical CPU reaches a customer in 2026

Akhetonics says its all-optical CPU reaches a customer in 2026

Light-based computing keeps promising more than it delivers — but one Munich startup has just put a date on its bet, and the interview laying it out is doing the rounds on Hacker News this week. Akhetonics says it will deploy its first commercial machine with a major customer by the end of 2026, with several more planned for 2027. The company, founded by Michael Kissner and Leonardo Del Bino, is building a computer where data enters as light, is switched as light, and circulates through memory

The Week in AI — October 5–11, 2026

The Week in AI — October 5–11, 2026

Every big claim this week turned out to rest on fine print more interesting than the headline: revenue only the company reporting it can define, safety tests sandboxed while the product keeps the web, and a Pentagon phase-out nobody would confirm until reporters kept asking. The week's top 5 1. OpenAI's revenue was $20 billion below the numbers everyone quoted — and the gap was definitional. The Financial Times reported Thursday that OpenAI's annualized revenue runs roughly $20 billion unde

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Russia's largest tech company is learning what the AI era's infrastructure war looks like from the receiving end — three data centers in four days, and with them much of the cloud layer Russian businesses run on. A Ukrainian drone strike knocked out Yandex's data center in Vladimir early Sunday morning, the third of the company's facilities hit since October 8. The site — reported at roughly 50 MW and designed for about 2,880 server racks — stopped operating completely after the attack, Yandex

Agent teams cost up to 5x more, barely score higher

Agent teams cost up to 5x more, barely score higher

The multi-agent hype train hit a benchmark this weekend — and the grid and the trucking regulators had quiet weeks of their own. Vals AI put agent teams head-to-head with single agents on its Vibe Code Bench, and the teams cost between 1.8 and 5.1 times more for almost no extra quality. The evals company ran GPT-6 Sol and Claude Opus 5.5 solo and in teams across 50 apps at two reasoning efforts; out of four comparisons, only one was statistically significant — Sol at medium effort, where the t