Slower AI Race Could Reward Investors With Capital Discipline

The frontier AI race has required escalating investment in model training, chips and data centers by Anthropic, OpenAI and major cloud providers, underpinning demand for suppliers such as Nvidia. The Bank for International Settlements estimates AI infrastructure investment at $1 trillion across 2025 and 2026. A slower training cycle could curb capital expenditure, reduce the risk of inadequate returns and lift free cash flow, although chipmakers and other infrastructure vendors would face weaker demand.

Anthropic Chief Executive Dario Amodei called for a slower pace of capability development on Sept. 12, warning that rogue AI agents could threaten the internet within six to 12 months. OpenAI’s Sam Altman and xAI’s Elon Musk subsequently backed the case for restraint. The Nasdaq closed 0.6% lower on Sept. 14 as investors weighed near-term pressure on AI-linked suppliers against the potential benefit of lower model-training costs and more disciplined spending by developers.