On Monday, Decagon CEO Jesse Chang published an exciting new theory, titled “Everyone is wrong about open source AI in the enterprise.” The article grapples with one of the most interesting contradictions in today’s AI economy: More mature AI deployments are shifting to lighter models, he says, even at his own company. But overall spending on expensive newer models has barely budged.
It’s a new way to think about the relationship between frontier and open source models. According to Zhang, they are not competitors, and the success of open source models does not come at the expense of frontier labs. Instead, they are two stages of the same life cycle, with expensive parametric models used to prove use cases that can be passed on to cheaper open source alternatives as they mature.
As more mature use cases shift to lighter models, new use cases continue to emerge – and overall spending on the leading models is barely coming down.
Chang doesn’t provide much data to support this point, but the data isn’t hard to find. Vercel’s AI portal dashboard It shows that just last week, DeepSeek rose to the top spot in terms of token volumes, and is now processing just over a third of the tokens passing through the company’s infrastructure. Z.ai – the laboratory behind the popular GLM-5.2 model – jumped to a respectable fourth place during the same period.
But if you scroll down to total token spending, you’ll see that Anthropic still accounts for more than half of the total AI spending on the platform. Given that much of the recent change comes from higher Anthropic prices, the share has declined slightly over the past month, but not significantly.
OpenRouter It tells a similar story, capturing a much larger (but slightly less institutional) slice of the market. DeepSeek V4 Flash is the main winner in overall usage, processing 5.3 trillion tokens weekly. The most popular frontier model, Opus 4.8, handles just over $2 trillion. OpenRouter doesn’t rank models by total spend, but it records the average token cost for Opus 4.8 being about 23 times higher than V4 Flash ($1.37 per million tokens, compared to just 6 cents), meaning Opus still likely accounts for the lion’s share of spending.
These numbers don’t even reflect the latest arrival, Nvidia’s Nemotron, which is Ready to jump to the front of the pack Thanks to Nvidia’s powerful connections and the model’s superior adaptability.
These numbers don’t fully prove Chang’s point about AI life cycles, but they do show that leading labs like Anthropic Labs aren’t suffering much from the rise of open source — at least not yet. One explanation is that the market for tasks that can be handled by AI is growing so rapidly that top models have been able to maintain their position only by dominating early-stage deployments. As Zhang puts it: “Frontier labs will continue to own the discovery. Open source will increasingly own the production.” Another explanation may be that even as customers move to open source, many use cases are too difficult to completely replace with cheaper alternatives.
Either way, this two-tier model economy may become a relatively stable feature of the AI economy.
Last September, I was writing about the possibility that core labs might end up selling coffee beans to Starbucks — that is, serving as commodity inputs while the application layer reaps the benefits. Some parts of this prediction have come true: vertical AI plays have shifted to lighter models, for example, and the economics of GPT startups have remained mostly stable.
But we also see that, through a token, CSPs have been able to hold onto the most desirable part of the market – the token price. It doesn’t look like this will change any time soon.
When you buy through links in our articles, we may earn a small commission. This does not affect our editorial independence.
