South Korea's AI Chip Stock Pullback | Rewire News Morning Update

Bitsfull2026/06/08 09:369992

概要:

South Korean AI Chip Stock Plunge Triggers Supply Chain Valuation Reassessment, While Nvidia Moves Forward with Collaboration with SK Hynix.

South Korean AI Chip Stocks See Significant Pullback on Monday. Industry collaboration continues, as the market begins to reevaluate the valuation pace of the AI supply chain.


1|South Korean AI Chip Stocks Retreat as Market Reassesses Supply Chain Premium


The South Korean stock market witnessed a sharp decline on the morning of June 8th. Livemint, citing a Bloomberg report, stated that the KOSPI fell by 8.8% at one point, retreating approximately 15% from its recent high and nearing a technical correction. Samsung Electronics saw an 11% decline, while SK Hynix dropped by 10%. This was not a typical sector rotation but a simultaneous sell-off of core assets in the South Korean AI trade. Samsung and SK Hynix are global hubs for HBM and DRAM supply and have been the main drivers of the South Korean stock market's rise over the past few months.


Making things more intriguing is that positive news was not absent. Reuters reported on the same day that NVIDIA announced partnerships with SK Hynix, Naver, and Doosan to establish AI data centers. NVIDIA Newsroom also confirmed that NVIDIA has signed a multi-year technical collaboration with SK Hynix to advance next-generation memory for global AI facilities. While the industry logic is strengthening, the stocks plummeted. The market reaction indicates that the certainty of the AI supply chain alone can no longer support valuations, as funds are starting to address crowded trades and lofty expectations first.


(Source: Livemint / Bloomberg / Reuters / NVIDIA Newsroom)


2|OpenAI Aims to Transform ChatGPT from a Chat Window to an Operating Gateway


TechCrunch reported that OpenAI is still advancing its ChatGPT super app plan, with insiders directly stating, "Chat is dead." This statement does not actually mean the disappearance of chat functions but that ChatGPT can no longer just be a question-and-answer window. It needs to integrate coding tools, agents, document processing, app connections, and commercialization capabilities into a single gateway. 36Kr also mentioned in its morning news report on the same day that ChatGPT is about to undergo the most significant upgrade since its launch.


On the surface, this signal appears to be a product iteration, but at its core, it is a gateway competition. Over the past two years, users have treated the model as a smarter search box. Now OpenAI wants it to become the default operational layer for workflows. The questions have also changed: whoever owns user context can determine tool invocation, payment methods, and the placement of third-party apps. If ChatGPT becomes a super app, the model's capability will only be the foundation, with the true value lying in the gateway hierarchy.


(Source: TechCrunch / 36Kr)


3|NVIDIA Expands AI Factory Partnership from HBM to Power and Industrial Equipment


NVIDIA and SK hynix have announced a multi-year technology collaboration focused on advancing next-generation memory to serve the global AI factory buildout. Reuters and Yonhap report that this collaboration revolves around the next-generation memory required for AI data centers. On the same day, 36Kr quoted Huang Renxun as stating that NVIDIA has been purchasing billions of dollars' worth of chips from SK hynix each year, with this year's procurement set to increase substantially, extending the cooperation period to over two years.


Another development comes from Doosan. NVIDIA Blog states that NVIDIA is partnering with Doosan Group to cover physical AI, robotics, and AI factory infrastructure, where Doosan Heavy Industries will explore supporting AI factories with gas turbines, steam turbines, small modular reactors, and hydrogen fuel cells. AI factories are transitioning from chip procurement to system-level integration. Whoever can simultaneously secure HBM, power equipment, industrial automation, and materials is closer to controlling the next-generation compute infrastructure.


(Source: NVIDIA Newsroom / NVIDIA Blog / Reuters / 36Kr)


4|Trump Urges Israel to Restrain as Iran Ceasefire Returns to Negotiation Table


Iran launched missiles at Israel on June 7, the first direct attack since the ceasefire on April 8. Axios and AP both point to the same chain of events: after Israel's strike on Beirut, Iran responded with missiles, putting renewed pressure on the already fragile ceasefire. Al Jazeera's live coverage also shows Iran and Israel continuing to threaten each other with retaliation and counter-retaliation.


A critical turn of events occurred in Washington. Axios reports that Trump urged Netanyahu not to retaliate against Iran's missile attacks to avoid disrupting the U.S.-Iran negotiations. The U.S. is now playing a role not just as an ally but in managing risks between Israeli military actions and Iranian negotiation windows. The variables of the Middle East conflict are shifting from the battlefield to the negotiation table, and the market is starting to reprice the risks of an Iran war, AI valuations, and upward interest rates in the same risk bucket.


(Source: Axios / AP / Al Jazeera / Fortune)


5|AI Security Failure Leads to Litigation, Product Claims Under Court Scrutiny


Ars Technica reports that a survivor of a Nashville high school shooting is suing AI firearm recognition system company Omnilert. The plaintiff alleges that the system failed to identify the weapon before the shooting, while the company had previously marketed it as capable of detecting firearms before firing. Local media WSMV's report further mentions that the litigation dispute includes camera placement, distance, angle, lighting, and weapon visibility limitations.


The significance of this news lies not in a single system failure, but in AI security products entering the realm of liability boundary testing. In the past, AI companies often marketed their products based on accuracy, real-time detection, and automatic alert capabilities. However, when a real incident occurs, the court will inquire about the conditions under which the system is effective, whether any limitations were disclosed at the time of sale, and whether the school created a false sense of security as a result. As AI transitions from the laboratory to public safety scenarios, mistakes are no longer just a matter of model metrics but also encompass product liability, procurement reviews, and public institutional trust issues.


(Source: Ars Technica / WSMV)


Also worth noting ↓


Agentic AI has accelerated coding, yet the bottleneck of software engineering has shifted. VentureBeat reports that with the increase in code generation speed, requirements definition, system integration, testing, and deployment have become new bottlenecks. AI is not eliminating software engineering but rather pushing software engineering back towards system governance.

(Source: VentureBeat / TechRadar / arXiv)


White House AI advisor Sriram Krishnan will step down at the end of June. According to reports from The Washington Post and TechCrunch, he plans to leave his government position and establish an external organization to continue influencing tech policy. The institutionalization pace of U.S. AI policy still lags behind the industry's expansion.

(Source: Washington Post / TechCrunch)


After Notion restored Anthropic access, application layer dependency issues were brought back into focus. A single interruption in a model vendor visit can impact a large number of user workflows. The risks of using AI are shifting from pricing and efficacy to supply chain resilience.

(Source: TechCrunch)


JPMorgan suggests that the Strategy may need to rebuild its dollar reserves. The Block reports that the market's focus has shifted from how much Bitcoin Strategy holds to whether it has enough cash cushion to pay dividends and maintain investor confidence. Bitcoin treasury companies are being re-evaluated as credit structures.

(Source: The Block / JPMorgan)


Adyen wins the Gov.UK Pay contract, replacing Stripe. The vendor switch in the government payment system signifies that payment infrastructure is not a set-it-and-forget-it situation. Public sectors are re-evaluating who will take over the national-level payment interface.

(Source: Finextra)


Huachao Shenkong completes million-dollar angel round financing, betting on AI-based ultrasound brain-machine interface. The capital narrative of brain-machine interfaces is shifting from invasive chips to medical imaging, ultrasound, and AI diagnostic pathways, with early-stage projects seeking easier clinical adoption. (Source: 36Kr)


AI content creators are becoming increasingly hard to distinguish. The Verge reports that AI virtual creators are transitioning from obvious experimental subjects to more indistinguishable platform personalities. What content platforms will have to deal with next is not just generating content but also the authenticity of identity itself. (Source: The Verge)


Qubit claims that domestically developed open-source AI long video framework has entered the global top tier. The focus of Chinese AI video tools is no longer just generating a few seconds of video but the engineering capabilities of long video consistency, latency, super-resolution, and more. (Source: Qubit)


Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia