AI Slowdown Fight Wipes $150B From Nikkei

Highlights

Nikkei 225 shed roughly 2%, wiping out about ¥23.6 trillion ($150 billion) in market value
Kioxia, SoftBank, Advantest and Tokyo Electron led an AI-linked chip-stock rout
Trump dismissed a joint slowdown call from Anthropic, OpenAI and xAI's CEOs, citing China
Anthropic, OpenAI and Google are separately discussing a shared AI standards body

Japan's Nikkei 225 shed close to 2% this week, wiping out roughly ¥23.6 trillion, or $150 billion, in market value, as a rare joint call from three of the AI industry's most powerful chief executives to slow the pace of development rattled chip and technology names across Asia. The selloff hit the same stocks that have carried Japan's market to records over the past year, turning a story about AI safety into one about AI valuations.

Kioxia Holdings led the decline with a 7% drop, followed by SoftBank Group down 4%, Advantest off 6.5%, and Tokyo Electron sliding 2.6%, as traders repriced the AI infrastructure trade that has underpinned much of this year's chip-sector rally. The move echoed a similar, larger rout in July, when more than $1 trillion was erased from global chip stocks over doubts about the AI buildout's near-term payoff.

The immediate trigger was a bipartisan-style appeal from the chief executives of Anthropic, OpenAI and xAI urging the industry to slow down and coordinate on safety before releasing ever more powerful models. Anthropic CEO Dario Amodei's push for shared technical testing and auditing standards has, according to The Information's reporting, already moved past talk: Anthropic, OpenAI and Google have been quietly discussing forming a joint AI standards body, with OpenAI CEO Sam Altman backing the idea of an industry-run testing and auditing function that the major labs would set up themselves rather than wait for Washington to act.

The White House was not receptive. President Trump told reporters on Sunday that the US cannot afford to cede ground to China by slowing down, saying, as the Washington Post reported, that “whoever wins with AI wins.”

Whoever wins with AI wins.

David Sacks, the White House's AI and crypto czar, went further in a lengthy post on X, accusing Anthropic and OpenAI of holding a “duopoly on frontier intelligence” and warning that a self-appointed standards body could functionally freeze out smaller labs and open-source developers under the banner of safety.

What such a body would actually enforce remains unsettled. Proposals floated around the talks include mandatory third-party audits before a new model ships, public disclosure of safety-testing results, and agreed capability thresholds that would trigger extra scrutiny once a model crosses them. None of that has been formally adopted, and the three companies have not said when, or whether, they intend to announce anything concrete.

House Speaker Mike Johnson staked out a middle position, proposing that tech executives meet at the White House to discuss safety guardrails without halting research outright. Johnson argued that unilateral restraint would simply hand China the lead, framing the dispute as a balance between national security and the immediate risk of releasing insufficiently tested models.

For markets, the episode is less about who wins the policy argument and more about how quickly investors will treat AI-safety rhetoric as a catalyst for repricing chip and infrastructure names. Japan's chipmakers and equipment suppliers have been among the most direct beneficiaries of the AI capital-expenditure boom, which means they are also the most exposed to any perception that the pace of deployment could slow. Whether this week's drop marks a genuine shift in sentiment or a one-day reaction to a political spat in Washington will likely depend on what, if anything, Anthropic, OpenAI and Google actually agree to in the coming weeks.