Kospi surges 8% on ceasefire hopes as SpaceX's Nasdaq debut ushers in the space-data-center era

The Kospi rallied more than 8% to break through the 8,400 level, closing around 8,426 points. The Kosdaq also climbed about 4.5% to reclaim the 1,000-point mark, ending at roughly 1,042. Foreign investors bought a net 1.4 trillion won worth of shares and institutions added another 1.8 trillion won, a marked improvement in market flows. By sector, program buying concentrated in large electronics manufacturers led by SK Hynix and Samsung Electronics.

U.S. President Trump had warned of a strike on Iran, then in the early morning hours announced he was calling off the attack after reaching an understanding with Islamic-bloc nations, cheering Wall Street and lifting sentiment in Seoul. Reports suggested a signing with Iran could come as soon as this weekend, with a pledge to immediately reopen the Strait of Hormuz once signed. Axios reported that a memorandum of understanding is being finalized under which Iran would give up nuclear weapons ambitions in exchange for resolving the uranium issue.

SK Hynix turned net-bought by foreigners the previous day for the first time in ten months. Provisional data showed foreign investors' top net buys as Samsung Electronics, SK Hynix, Samsung Electro-Mechanics, Samsung Electronics preferred shares, and Naver, in that order. Retail investors were net sellers, which commentators attributed largely to profit-taking on positions bought at lower cost rather than a bearish shift. The advice offered was to avoid selling into strength outright, to sell in tranches, and not to fully exit semiconductor positions.

Goldman Sachs argued the market is underestimating hyperscaler AI capex. This year's spending is projected at roughly $780 billion, but if investment scales toward 2-3% of GDP — comparable to past infrastructure build-outs like railroads and automobiles — annual spending could exceed $1 trillion by 2027, and reach as much as $1.4 trillion under a bull case.

Next week brings a cluster of events — the U.S. Fed's FOMC decision, Japan's rate decision, quad witching, a Bank of England policy meeting, and a U.S. market holiday on Friday — leading commentators to suggest Trump wants to wrap up ceasefire progress before the week is out.

OpenAI's Sam Altman is set to arrive in Seoul on Sunday, June 14 for a brief one-night visit, planning to tour Kakao's Pangyo campus and Samsung Electronics' DX division and meet with executives, as well as visit Naver to discuss cloud data-center cooperation. On his previous visit eight months ago, Altman urged Samsung and SK Hynix to expand advanced memory output, reportedly requesting up to 800,000 DRAM wafers per month — remarks that preceded sharp rallies in both stocks, so markets are watching closely for any similar signal this time. South Korea reportedly ranks second in the world in paid ChatGPT subscribers behind the U.S., and possibly first on a per-capita basis.

Hyundai Motor Group is pushing ahead with a Robotics Metaplant Center (RMC) in the U.S. — a Boston Dynamics robot training hub — and has reportedly begun hiring for it. Plans reportedly call for two joint-venture entities: a robot training center and a robot manufacturing plant. Samsung Securities flagged the RMC's shareholder structure as worth watching, suggesting Google DeepMind and Nvidia could invest in exchange for access to real-world robot action data collected at Hyundai's factories to advance their robot foundation models. Panelists also questioned why Korean chipmakers and battery makers haven't been brought into such investments.

Boston Dynamics' Spot security robots are being deployed at World Cup venues to patrol blind spots inaccessible to human security staff and monitor risks in real time; Atlas is also being showcased, seen as a branding push ahead of Boston Dynamics' own IPO.

LG is reportedly investing 400 billion won to build Korea's first robot training facility, sending LG-affiliated robotics stocks higher on the day.

Globally, robotics makers are said to be racing toward mass production by 2028, cycling through proof-of-concept iterations every three to six months. Tesla and Figure AI have unveiled large-scale manufacturing plants, and a Chinese company reportedly already shipped 10,000 robots annually last year, pushing the industry into a phase of competition over production capacity, quality, and unit cost.

SpaceX is going public on Nasdaq at an IPO price of $135 per share, valuing the company at $1.75 trillion, with $75 billion raised in the offering — more than double Saudi Aramco's previous record of $29.4 billion — making it the largest IPO in history. Underwriters reportedly slashed fees unusually low, a bet on winning future SpaceX business.

SpaceX's workhorse Falcon 9 effectively dominates the global commercial launch market, accounting for roughly 50% of all worldwide launches last year. Its reusability program drew skepticism early on from rivals like ULA, who argued there wasn't enough launch demand to justify it economically, but the Block 5 version stabilized reuse operations after 2018. The real bottleneck turned out to be demand — cheaper launches meant little without enough payloads to fill them — which is what drove SpaceX to build its own satellite internet business, Starlink.

About 75% of Falcon 9 launches last year carried Starlink satellites. Starlink revenue rose from roughly $1 billion in 2022 to surpass half of total revenue by 2023, reaching an estimated $10 billion last year — leading analysts to argue SpaceX should now be viewed as a telecom company, not a rocket company. Launches themselves are close to break-even: with a Falcon 9 designed for ten reuses, revenue from four external customer launches can cover the cost of all ten, making the remaining six essentially free Starlink launches.

SpaceX's IPO prospectus disclosed three core business pillars for the first time: AI, connectivity (Starlink), and space (launch). Notably, it revealed data-center leasing deals worth $15 billion a year with Anthropic and $11 billion with Google, leasing spare capacity from its Colossus data centers run by xAI (the Grok operator, now merged into SpaceX). Factoring in a rumored Cursor acquisition and full-year revenue guidance, the price-to-sales ratio falls from roughly 100x based on last year's $18.7 billion revenue to about 30x — still well above Tesla's roughly 15x six-year average, prompting some to call the valuation rich.

The next big frontier flagged was orbital data centers — satellites equipped with solar panels and compute hardware, sidestepping ground-based power and permitting constraints, built on the same launch-and-reuse infrastructure as Starlink. A recent Google data-center leasing deal implies roughly $50 billion per gigawatt, more than triple the roughly $15 billion consensus from just months earlier, underscoring how severe the compute shortage has become. The key variable is Starship's maturation — payload capacity needs to scale from Falcon 9's 20 tons to Starship V3's 100 tons and eventually V4's 200 tons for the economics to work. SpaceX targets 2028, though panelists suggested 2029 is more realistic.

Speculation also touched on a possible Cursor merger this year followed by a Tesla merger, the logic being that Optimus and self-driving both need AI and connectivity infrastructure that would benefit from Starlink-style vertical integration. There was also talk SpaceX could acquire telecom carriers in various countries to enable direct-to-cell satellite connectivity — it already spent over 20 trillion won acquiring EchoStar's spectrum last year. Panelists flagged political backlash and regulatory risk from a single company dominating launch, telecom, data centers, and AI models, plus growing concerns over satellite congestion and anti-satellite technology.

Host Lee Kwang-soo opened the show recounting a conversation with a humanoid robotics startup founder. The founder described the company as spanning everything from data to actuators, and voiced concern that if the gap between market hype and actual cash generation stretches too long, investors will grow fatigued.

The founder argued that failing to achieve mass production means the startup fails outright, and said that most rival founders in the space — himself included — are already preparing for mass production and actively fundraising. He said larger early-round investments are always welcome and pledged that any capital raised would go entirely toward building manufacturing capacity.

Many in the industry view 2028 as the real inflection point for commercialization, with the intervening years spent iterating through proof-of-concept cycles and revisions — a framing that echoes the Hyundai and LG robotics investments covered in the Stocks section.