AI token price collapse hits big tech valuations
The recent decline in AI token prices has sparked concerns over the sustainability of valuations for major technology companies, particularly those heavily invested in artificial intelligence infrastructure. As per-token costs have dropped significantly—by as much as 75% year-over-year—many organizations have not seen corresponding reductions in overall AI spending. Instead, the rise of agentic AI workflows has led to a surge in token consumption, with hundreds of thousands of tokens per task. This shift has created a "token cost illusion," where falling unit prices are offset by rising total usage, resulting in higher overall expenses for enterprises.
The financial impact is already evident in the earnings reports of major tech firms. Microsoft, for example, has seen its market value decline by approximately $613 billion year-to-date, with concerns over AI profitability and competition from models like Google’s Gemini and Anthropic’s Claude Cowork AI agent contributing to investor unease. Similarly, Amazon has lost $343 billion in market value, as it prepares for a more than 50% increase in capital spending this year. Nvidia, Apple, and Alphabet have also experienced significant declines in valuation, with investors shifting from long-term AI optimism to a demand for near-term earnings visibility.
The broader market has also shown signs of recalibration. The Nasdaq’s sharp decline in early June 2026 marked a potential turning point in investor sentiment, with traders questioning whether AI boom can sustain current valuation multiples without immediate returns. Analysts and industry leaders, including JPMorgan CEO Jamie Dimon and Bridgewater founder Ray Dalio, have warned of growing exuberance and the risks of a market correction. Dimon likened the current environment to past speculative bubbles, including the dotcom era, while Dalio noted that equity markets are approaching levels seen in 2000.
Meanwhile, the AI infrastructure boom continues to drive economic growth, particularly in the United States. Datacentre investment has become a critical component of GDP expansion, with one Harvard economist estimating that 92% of U.S. GDP growth in the first half of 2025 was attributable to AI-related infrastructure. However, the feasibility of sustaining this level of investment remains uncertain, as datacentre construction faces logistical and energy constraints.
As the AI market matures, companies are exploring alternative strategies to manage costs and maintain control over AI usage. Some are turning to on-premise solutions, such as NVIDIA’s NemoClaw, to reduce dependency on cloud-based AI services and gain greater visibility into token consumption. Others are adopting hybrid approaches balancing cloud flexibility with cost predictability and regulatory compliance.
The coming months will be critical in determining whether the AI boom can deliver the economic returns that justify its current valuations. For now, the market remains in a state of flux, with investors closely watching for signs of either a correction or a sustained shift toward profitability.
