The September 2026 Ramp AI Index reveals US corporate AI spending is declining despite a 50% surge in usage since July. This divergence stems from aggressive price cuts and cheaper, more efficient models driven by intense competition between Anthropic and OpenAI. Meanwhile, massive infrastructure investments are straining tech hyperscalers, creating a near $1 trillion spending-to-revenue gap that is pushing investors to prioritize free cash flow over speculative AI growth.
Corporate AI spending and usage trends
- ▪Open-source models account for less than five percent of corporate artificial intelligence spending tracked by the Ramp AI Index.
- ▪The decline in US corporate artificial intelligence spending has lasted unusually long, with spending having peaked in July 2026.
- ▪According to the Ramp AI Index, corporate usage of artificial intelligence has climbed approximately 50 percent since spending peaked in July 2026, reaching a record high at the end of September 2026.
- ▪According to the Ramp AI Index, US companies are spending less on artificial intelligence due to price cuts on top models and cheaper, more efficient standard and lite models.
Competition between OpenAI and Anthropic
- ▪Ramp economist Ara Kharazian stated that the drop in corporate artificial intelligence spending comes almost entirely from competition between OpenAI and Anthropic.
- ▪During the final week of September 2026, Anthropic captured 51 percent of corporate token spending while OpenAI captured 44.5 percent, according to Ramp AI Index subsample data.
Government AI costs
- ▪A McKinsey report focusing on American government entities warned that the per-token price of artificial intelligence is collapsing while the total bill rises.
- ▪A McKinsey report and Senior Partner Tim Ward warned that near-zero artificial intelligence usage costs for US and many governments will end as pilot programs transition to agentic applications requiring significantly more compute.
Capital spending by technology hyperscalers
- ▪A Raymond James analysis found that by the second quarter of 2026, aggregate capital expenditures of technology hyperscalers began to exceed operating cash flow, pushing free cash flow into negative territory.
- ▪An analysis by Stanford Institute for Economic Policy Research economists Jared Bernstein and Ryan Cummings identified a nearly $1 trillion gap between capital spending by major tech hyperscalers and their artificial intelligence revenues since 2024.
Investor focus on free cash flow
- ▪The VictoryShares Free Cash Flow ETF (VFLO) experienced notable investor interest with monthly net inflows exceeding $900 million as of late September 2026.
- ▪Rising artificial intelligence capital expenditures are pressuring corporate cash flows, leading investors to focus heavily on free cash flow, return on equity, net debt leverage, and earnings consistency.
Debatable claims
- ▪Agentic applications will drive up total AI costs for organizations despite falling per-token prices
- ▪Tech hyperscalers' massive AI capital expenditures are justified by long-term returns
Story comments
Loading comments…