The AI development slowdown stocks rout that swept Asian markets on Monday offers a sharp reminder of how much speculative weight the sector is now carrying. South Korea’s KOSPI fell 3.7%, chipmaker SK Hynix slumped 5.75%, and Taiwan Semiconductor Manufacturing Company (TSMC) dropped 1.2%. SoftBank, though, was the session’s most arresting casualty: shares fell as much as 13% in Tokyo.
What Dario Amodei Actually Said
The catalyst was a public intervention by Anthropic chief executive Dario Amodei, who warned that ‘building [AI] too fast is reckless’ and suggested a swarm of AI agents could cause hundreds of billions of dollars of damage by ‘taking over the entire internet’ in future. Sam Altman of OpenAI and Elon Musk both backed the call. Some AI researchers have disputed the claims, but markets were not waiting for the academic consensus to settle.
Anthropic has its own internal framework for deciding when to pause development. Its published safety guidelines describe an ‘ASL-4’ threshold, triggered when AI systems achieve near-human autonomy or become a primary enabler of serious global security threats such as bioweapons. At that point, the company states it will require a detailed and precise understanding of AI risks before proceeding. Whether or not that threshold is imminent, the language from Amodei normalises the idea of a deliberate brake on progress, and investors are repricing accordingly.
SoftBank’s AI Development Slowdown Exposure Is Unlike Any Other
The scale of SoftBank’s bet on AI explains why its shares fell harder than almost any peer. SoftBank Group‘s own press release confirms the company has invested an aggregate of $34.6 billion in OpenAI through SoftBank Vision Fund 2 since September 2024, with a follow-on investment that would lift its ownership interest to approximately 13%.
The financing behind that position is aggressive. According to the Wall Street Journal, SoftBank used its OpenAI stake as collateral to borrow $10 billion from a group of banks, and is poised to pay a further $10 billion for OpenAI shares by October as the final tranche of a $30 billion investment announced earlier this year. To fund these commitments, Yahoo Finance reports, citing a SoftBank financial filing, that the company sold its entire Nvidia stake for $5.83 billion and made a $9.17 billion partial sale of its T-Mobile holdings as part of an asset monetisation strategy targeting a $40 billion OpenAI investment by year-end.
The concentration is extraordinary. CNBC reported that SoftBank’s CFO Yoshimitsu Goto said 60% of the company’s assets are now ‘ASI-oriented investments.’ The OpenAI stake produced a $17 billion gain in the April-to-December period alone, helping SoftBank book fiscal third-quarter net profit of 248.6 billion yen ($1.6 billion). That gain now looks far less assured if the pace of AI development slows materially.
CLSA has already moved. Investing.com reports that the bank lowered its SoftBank price target to JPY 28,062, noting that funding the OpenAI position required $16 billion of asset sales alongside probable debt financing. Leverage built on aggressive assumptions about AI demand looks different when the people building the models are the ones urging caution.
The Credit Risk Question
Ipek Ozkardeskaya, senior analyst at Swissquote, put the systemic concern plainly: ‘So if the AI race slows materially, the key question becomes: who pays for all that infrastructure? The leases, debt and power commitments remain even if expected compute demand and revenue growth slow. And that could bring credit risk increasingly into the AI story, particularly for highly leveraged data-centre operators and lenders exposed to projects built on aggressive assumptions about future AI demand.’
She adds that the slowdown impulse is coming from an unexpected source: ‘It comes from the actual people who develop these models.’ That is the part markets find hardest to price. Regulatory headwinds, rate cycles, and competitive threats are legible. A collective moral intervention by the industry’s leading figures is not.
OpenAI’s IPO is now off the table for this year, Altman confirmed, which removes one near-term catalyst that SoftBank investors had been watching. With borrowing costs still elevated and asset sales largely exhausted, SoftBank’s next move is constrained. The October deadline for that final $10 billion OpenAI payment is the number to watch.
