Editor's note: The Global Morning Briefing summarizes global news reported by Seoul Economic Daily.
Tariff Truce, Rare Earths Swap Seen Likely; AI Deal May Stay at Level of Principle
Attention is building worldwide ahead of the U.S.-China summit at the White House on the 24th, as both governments are expected to bring a large contingent of Big Tech chief executives to discuss cooperation on artificial intelligence. Diplomatic observers say an extension of the tariff truce is likely to be traded against rare earth exports at the meeting, but that agreement on AI safety rules will probably go no further than broad principles.
According to major foreign media outlets on the 19th, Chinese President Xi Jinping will arrive at a base near Washington, D.C., on the 23rd and hold a summit with President Donald Trump on the 24th. AI and the tariff truce are set to be the central agenda items at the meeting, the third between the two leaders since Trump began his second term.
Big Tech figures from both countries are expected to turn out in force at the state dinner at the White House on the 24th. Invitations have gone to Amazon Executive Chairman Jeff Bezos, Alphabet Chief Executive Sundar Pichai, OpenAI Chief Executive Sam Altman and Nvidia Chief Executive Jensen Huang on the U.S. side. On the Chinese side, officials are considering including BYD Chairman Wang Chuanfu, CATL Chairman Zeng Yuqun and Xiaomi Chairman Lei Jun in the economic delegation.
AI safety rules have moved onto the agenda against a backdrop of Trump administration wariness over suspicions that Chinese companies have stolen U.S. technology. Beijing has pushed back, calling the concern an attempt to hold back its own technological development, but Wang Huning, chairman of the Chinese People's Political Consultative Conference, is said to have voiced agreement with U.S. figures last month on the need to discuss AI safeguards.
On the tariff truce, the United States wants an extension of six months to a year, while China hopes for one lasting through the remainder of Trump's term. Taiwan, Iran and North Korea are also expected to come up, and U.S. Treasury Secretary Scott Bessent and others began advance talks in New York on the 20th.
China's CXMT Starts Mass Production of Fifth-Generation DRAM at 11.95-Nanometer Half-Pitch
ChangXin Memory Technologies, China's largest memory chipmaker, said it has begun mass production of its G5 platform, a fifth-generation DRAM process. According to Reuters and other outlets, CXMT made the announcement on the 20th at the World Manufacturing Convention in Hefei, Anhui province.
CXMT Vice President Luo Xiaodong said the company's process technology is now close to the most advanced level in volume production in the industry. The company said it had reduced the active-area half-pitch of the memory array to 11.95 nanometers, adding that the result came from applying quadruple patterning, which repeats exposure and etching several times. Some analysts note, however, that the figure is a measurement for a specific structure and is difficult to compare directly with the "12-nanometer class" label used by Samsung Electronics (005930.KS), SK hynix (000660.KS) and Micron.
On density, CXMT said the potential number of dies per wafer, converted to an 8-gigabit basis, has risen at least 50% from the fourth generation. That figure is before screening for defects, however, and the company did not disclose actual yields or monthly output. CXMT also unveiled two 24-gigabit LPDDR5X products made with the new process, saying both have entered full-scale mass production.
CXMT ranked fourth in the world with a 9.5% share of global DRAM revenue in the second quarter of this year, according to TrendForce. The company recently announced mass production of LPDDR6, and its fifth-generation high-bandwidth memory, HBM3E, is said to be in trial production.
Against that backdrop, Acer Chairman Jason Chen of Taiwan said at a Taipei forum on the 19th that commodity memory for PCs is not in short supply thanks to expanded Chinese output. He said some LPDDR5 and DDR5 products are in oversupply, and that price increases could stop in the second half of next year once falling component prices are reflected. He also noted that supply remains tight for some items, including high-speed DDR5. Citi, meanwhile, forecast that DRAM demand growth will outpace supply growth in 2027 and 2028 on rising AI-driven demand.
Revenue Hits $65 Billion, but Anthropic's IPO Clock Slips
Anthropic, which is preparing an initial public offering this year, is drawing growing skepticism from investors despite steep revenue growth. With Chinese open-source AI models and OpenAI closing in and an industry debate over slowing down development, the listing, once expected in October, is now likely to be pushed back by about a month.
Anthropic's annualized revenue climbed to $65 billion as of July this year and is expected to top $120 billion by the end of the year, the Financial Times reported on the 19th. The FT described that pace as the fastest on record. Anthropic has been valued at $965 billion on the strength of its position in the enterprise AI market, making it the most valuable privately held AI company.
The pursuit by its biggest rival, OpenAI, has emerged as a variable. OpenAI released GPT-5.6 at aggressive prices in July and rapidly won over corporate customers, overtaking Anthropic in weekly customer spending for the first time in two and a half years on the AI model platform OpenRouter. OpenAI is pressing ahead with aggressive investment — its cumulative free cash flow deficit is projected to reach $278 billion by 2030 — and that spending is credited with lifting revenue by about 20% since the July model launch.
Low-cost open-source models from China's Moonshot AI and DeepSeek are also delivering performance close to Anthropic's closed models, bringing price competition into the open. Anthropic Chief Executive Dario Amodei has argued for slowing the pace of AI development to ensure safety, but critics say that stance does not help in competing with China and amounts to a pretext for cutting development costs. Amid that uncertainty, market expectations for Anthropic's post-listing valuation range widely, from $1.5 trillion to $4 trillion.
Anthropic, which is targeting a valuation of up to $2 trillion, now looks set to list in November rather than October, according to The Wall Street Journal. Reuters reported that Anthropic is considering a new model release of its own to counter attention drawn by OpenAI's new model, Astra.
An AI Force Like the Space Force: Trump Declares a Race for Speed
Trump said he will create a new "AI force" and fill the vacant post of AI czar, the government's top AI policy job. The announcement is seen as a restatement of his position that AI development will not be held back by regulation.
In a post on his social media platform Truth Social on the 19th, Trump said the government will not obstruct or suppress the growth of the AI industry in any way, while adding that problematic conduct will be addressed through the existing criminal and civil justice systems. The remarks stressed speed of development, on the judgment that the United States cannot fall behind in the advanced technology race with China, while also reflecting awareness of public concern over AI's risks.
Anthropic researcher Evan Hubinger said the same day that there is a greater than 10% chance AI will threaten humanity as a whole within the next decade. Trump called the tech industry's warnings that AI could destroy humanity a scam, countering that the only safeguard AI needs is a strong and smart president.
Citing the Space Force, created in 2019, as a successful example, Trump said the AI force would be run in a similar way and said he would name an AI czar soon. He went on to say AI will be a technology with an impact comparable to or greater than the next industrial revolution or the internet, and forecast that it could account for as much as 25% of U.S. gross domestic product. He also said the United States is ahead of China and other countries and that he intends to keep that lead.
Trump's anti-regulation stance appears to be hardening even as industry figures issue a string of warnings. Amodei, Altman, SpaceX Chief Executive Elon Musk and Google DeepMind co-founder and board chairman Demis Hassabis have all argued that the pace of development should be moderated as their own AI systems grow more powerful. The AI czar post is being refilled after the White House planned as recently as six months ago to leave it vacant; David Sacks, the Craft Ventures co-founder who previously stepped down from the role, retains influence as co-chair of the President's Council of Advisors on Science and Technology.
