The AI market should be measured in dollars, not tokens: analysis by Dragonfly analyst
Haseeb Qureshi, managing partner at venture firm Dragonfly, has sharply criticized the common approach to evaluating the AI model market. In his view, relying on raw token consumption is a fundamentally flawed method that distorts the market picture. The only adequate metric is real cash flows—that is, the dollars spent by users.
Why Tokens Are an Unreliable Indicator
Qureshi highlights four key problems that make the "token share" metric practically useless for comparing models.
The first problem is subsidies. Chinese labs regularly launch new models with aggressive discounts or even free access. This artificially inflates token consumption: users migrate from one free model to another, creating an illusion of high demand, while actual monetary costs are zero.
The second problem is varying model sizes. Small models, such as Qwen 3.5-27B, cost roughly a hundred times less per token than flagship models like Claude Opus. An increase in Qwen usage on a chart might look like a sharp jump in the share of open models, but in economic terms, it is completely insignificant. According to Qureshi, the market should be analyzed strictly within weight categories—by model size.
The third problem is multi-agent systems. You can spend the same amount on a complex architecture based on DeepSeek or GLM 5.2 as on a single query to a cutting-edge model like Opus or GPT-5.5 Pro, with comparable performance. However, a multi-agent configuration "burns" far more tokens for the same money. Qureshi provides a clear example: if 5% of Opus usage migrates to such a system with a fourfold token consumption, the chart would show about an 18% loss in Opus's share, while actual spending shifts by only 5%. "Such charts exaggerate the importance of low-value tokens," the expert concludes.
The fourth problem is the choice of the OpenRouter platform. If a company has settled on one leading lab, it is more profitable for it to go directly to Anthropic or OpenAI rather than through OpenRouter with its markup. On the chart, this looks like a decline in the US share, even though the tokens simply move outside the platform. Qureshi's conclusion: OpenRouter is useful for assessing the share within open models but is completely unsuitable for comparing open and closed ones.
The Future Belongs to Cheap Models
A similar idea is developed by Trevor Noren, founder of SageRoad Research, linking it to pricing pressure on the industry. He cites a JPMorgan estimate: many tokens in the future may be consumed not by cutting-edge models but by small open models that are sufficient for specific tasks. Amazon already offers about fifty open models at a fraction of the cost of advanced ones, while Nvidia, together with Dell, Lenovo, and HP, is creating computers for AI agents.
The bank notes that their own small models, Claude Haiku and GPT-5.4-mini, are currently uncompetitive on the "efficiency frontier," which is now dominated by Chinese developers—DeepSeek, MiniMax, Xiaomi, and Alibaba.
The cost example provided by JPMorgan is particularly illustrative: running the Artificial Analysis Intelligence Index benchmark on Claude Opus 4.8 costs $3,700 with a result of 56 points, while DeepSeek V4 Pro scores 44 points for just $186—roughly 20 times cheaper. The conclusion is obvious: cutting-edge intelligence is not needed for everything, only where it is necessary. Meanwhile, GLM 5.2 from Z.ai appears comparable to the top models from Anthropic and OpenAI.
Noren believes that the commoditization of models will come not only from competition among leading labs but also from companies seeking cost control through cheaper, specialized models. In his assessment, corporate spending remains the most viable path for cloud giants to recoup their AI investments, but companies will spend as little as possible.
Both positions converge on one point: the AI market should be measured by money, not tokens, and under pricing pressure, the advantage is increasingly shifting toward cheap models.
My professional opinion: The arguments of Qureshi and Noren are not just an academic discussion. For investors in the crypto and AI sectors, this is a signal of a fundamental shift in metrics. If the market begins to reassess models based on monetary revenue rather than "raw" consumption, we will see a redistribution of capital in favor of projects that demonstrate real monetization, rather than just generating noise in tokens. This is especially critical for evaluating tokens associated with decentralized computing networks.