Donald Trump has rejected calls from leading AI executives to slow frontier AI development. Speaking at the Amgen Irish Open in Doonbeg, Ireland, on Sunday, 13 September, the US president said America should protect its lead over China because 'whoever wins AI wins'.
His comments came one day after Anthropic chief executive Dario Amodei called for companies to slow capability gains so safety work can keep pace.
OpenAI chief executive Sam Altman and xAI founder Elon Musk backed the proposal. Microsoft chief executive Satya Nadella also welcomed 'deliberate pacing' and ideas such as embedded evaluators.
AI Chiefs Want to 'Pace the Frontier'
Amodei set out a three-step plan in an essay titled We Must Pace the Frontier. He stressed that pacing would not mean halting model training or technical progress. Instead, companies would give safety, alignment and testing more time to catch up with rapidly improving capabilities.
The first step would give independent third-party evaluators ongoing, employee-like access inside frontier AI companies. The second calls for coordination among leading laboratories and democratic governments. The third seeks global coordination, including with China, while recognising the difficulty of verifying compliance.
Altman said OpenAI agreed that the frontier should be paced and pledged to give independent evaluators employee-like access.
'I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks.'
Musk wrote that 'Dario is right.'
Nadella welcomed deliberate pacing and embedded evaluators, while also calling for broad representation, model choice and enterprise control.
'Any pursuit of superintelligence has to be grounded in the core principle that if the AI we build is not helping humanity and under human control, it's not worth pursuing.'
Amodei cited two main concerns.
One is AI's growing role in helping build the next generation of AI systems. The other is the OpenAI-Hugging Face incident, in which agents carried out cyber activity beyond the tasks they had been assigned.
Trump Says America Cannot Afford To Slow Down
Trump took a sharply different position when reporters questioned him at his golf course in Ireland. 'We're leading China in AI,' he said.
'We're the most sophisticated country in the world, and frankly I want to keep it that way because whoever wins AI, wins.'
Trump did not reject regulation outright. He said the US could put up 'guardrails', but criticised 'very negative forces' for raising scenarios that he said would not happen.
The divide is not simply Washington against Silicon Valley. Amodei also argues that US companies cannot slow too far if Chinese rivals continue advancing. His framework seeks to preserve the US lead while creating more time for safety work.
Report Links Anthropic to $13.7 Billion RUM Compute Deal
The Information reported, citing a person familiar with the matter, that Anthropic signed a six-year compute deal with RUM Group worth $13.7 billion (£10.1 billion).
A RUM Group regulatory filing confirms a six-year GPU-services agreement but does not name the customer. It describes the buyer only as an unaffiliated US-based third-party cloud customer. Neither Anthropic nor RUM had publicly confirmed the customer's identity at the time of reporting.
The agreement covers GPU services at RUM's data centre development in Maysville, Georgia. It is divided into three tranches, with the third subject to the customer approving RUM's proposed delivery date.
The customer can also acquire up to 50,808,408 RUM Class A shares at one cent each. Vesting is tied to purchases under the agreement and any expansion deals.
RUM Group, formerly Rumble Inc., is the parent company of Rumble and Quake AI. Truth Social migrated to Rumble's cloud infrastructure in 2022.
If Anthropic is the customer, the agreement does not by itself conflict with Amodei's proposal. He has explicitly said pacing does not mean stopping model training. It does, however, show the scale of computing capacity Anthropic may be securing while arguing for slower capability gains.
AI-Linked Stocks Fall After Slowdown Warnings
AI-linked shares across Asia fell sharply in early Monday trading after the weekend calls for a slowdown. Middle East uncertainty also weighed on sentiment.
SoftBank fell as much as 13.2%. Kioxia dropped 9.8%, SK Hynix fell 5.3% and Samsung Electronics declined 3.7%.
China's state-backed Global Times called Amodei's proposal a 'Cold War playbook'. It argued that the plan could preserve US technological dominance while constraining China.
The dispute shows why a co-ordinated slowdown would be difficult. AI companies are increasingly focused on safety. Governments are also treating the technology as a question of economic power and national security.
