The decline in AI token prices is reshaping the economics of the industry. According to Silicon Data's LLM Token Expenditure Index, the price per million tokens fell to $0.9665 on August 31, down from a May peak of $2.0651. Analyst Damir Tokic suggests this nearly 60% drop could be a warning sign for hyperscaler spending.
Hyperscalers like Alphabet, Amazon, Meta, and Microsoft are investing hundreds of billions of dollars in GPUs, data centers, and energy infrastructure. Reuters estimates they are on track to spend roughly $600 billion in 2026, with some estimates placing total capital spending closer to $650 billion. However, a July Reuters analysis indicates that combined capital expenditures for Microsoft, Alphabet, Amazon, Meta, and Oracle could exceed their combined free cash flow by 2027.
The price drop is driven by the shift toward smaller, open-weight models. On one major routing platform, the share of open-model tokens rose from 34% in January to 65% in June. This trend is further accelerated by model routers like OpenRouter, which Stripe acquired last month for more than $7 billion. OpenRouter serves 8 million users and provides access to over 400 AI models, processing more than 10 trillion tokens per day.
The industry is facing additional pressure from rising borrowing costs. U.S. tech giants have issued approximately €40 billion in euro-denominated debt to fund data center construction. Overall, AI-related capital expenditure is estimated to be between $5 trillion and $7 trillion through 2030. The success of the AI boom depends on whether token consumption can grow faster than prices fall.
