Open AI models present an intriguing economic puzzle. They can often deliver performance approaching that of leading proprietary models at a fraction of the cost. Yet enterprises continue to overwhelmingly favor closed, proprietary models — and by some measures, their share of enterprise AI spending is actually increasing. Why is that the case?
The answer may have less to do with the relative merits of open versus proprietary AI than with where enterprises currently find themselves on the AI adoption curve.
“Artificial intelligence is reshaping economic systems at a pace we have rarely seen in modern technological history,” wrote Frank Nagle, — research scientist at MIT’s Initiative on the Digital Economy and newly appointed chief economist at Microsoft’s AI Economy Institute, — in a November 2025 blog, “Revealing the Hidden Economics of Open Models in the AI Era.”
“Yet amid the excitement, a crucial part of the story has been missing,” he added. “Specifically, understanding the role that open models play in the AI economy, and how much value is being left on the table when organizations overlook open alternatives, are two topics requiring a closer look.”
In “The Latent Role of Open Models in the AI Economy,” Nagle and Georgia Tech professor Daniel Yue probed these questions by analyzing a comprehensive dataset of AI model usage, prices, and performance. Their research uncovered a striking economic puzzle.
Closed models account for roughly 80% of model usage and 96% of revenue, even though they cost, on average, six times more than competing open models. Meanwhile, open models routinely achieve 90% or more of the performance of proprietary models and typically close much of the performance gap within a few months after the release of a new closed model.
“The findings surprised even us,” Nagle wrote.”
So why are enterprises continuing to spend so much more on proprietary models?
A recent article, “Everyone Is Wrong About Open Source AI in the Enterprise,” by Decagon co-founder and CEO Jesse Zhang, offers a very interesting explanation.
“The prevailing story right now is that open source is eating the enterprise,” Zhang wrote. “The capability gap between the best closed and open models has shrunk to low single digits. A third of the Fortune 500 has verified accounts on Hugging Face.”
“And yet enterprise spend as a whole is moving the opposite direction. Open source models just fell to 11% of enterprise LLM spend, down from 19% a year ago. The trend is actually moving the other way compared to the popular narrative. Why is this and what does it mean for the future?”
Frontier Models for Discovery; Open Models for Production
Zhang offers a compelling resolution of this apparent paradox based on his own company’s experience. Decagon develops conversational AI agents for customer service, and roughly 90% of its workloads run on open models. “It wasn’t cost, and it wasn’t because our customers demanded it (though they don’t mind it),” he wrote. “It was because we had no other option.”
The reason has to do with where an AI application is in its development lifecycle.
“When a use case is new, you want the smartest general-purpose model you can get,” Zhang explained. “You don’t know the shape of the problem yet, so you pay a premium for intelligence you may not end up needing. That’s the right trade at that stage.”
But the economics change as the application matures.
“Once the use case is fully built out, when you know the distribution of inputs, the behaviors you need, and the failure modes to guard against, the trade flips. Now general intelligence is overhead, and you want the smallest, fastest model fine-tuned to do your specific thing extremely well.”
In other words, the AI deployment lifecycle might be summarized as:
Experiment → Learn → Specialize → Scale
During experimentation, the most capable frontier models offer significant advantages. Enterprises are still discovering what AI can do, which applications create real value, and how their workflows need to change. At this stage, paying a premium for the flexibility and broad intelligence of a proprietary frontier model can make considerable sense.
But once an application is well understood and operating at scale, the requirements change. Speed, latency, customization, control, and cost become increasingly important. At that point, smaller, specialized open models may become much more attractive.
Customer service is a particularly good example.
“Customer service happens to be one of the most obvious AI use cases in the industry,” Zhang wrote. It involves “well-understood workflows, enormous conversation volume, tight quality bars.”
Companies like Decagon may simply be further along the AI maturity curve than the average enterprise deployment. “When you’re running AI agents in production for customer service, latency makes or breaks the product,” Zhang explained. “A conversation where every turn takes 8 seconds is not a product anyone will use.”
That creates a need for smaller, faster models. But small models generally don’t meet demanding enterprise quality standards without significant customization. “They only get there through heavy fine-tuning on the exact task.”
And that, according to Zhang, is where open models have an important advantage.
“The frontier labs don’t really sell this combination. You can’t fine-tune their best models the way we need to, and their small models aren’t ours to shape. Small + fine-tuned means open weights.”
“The cost savings are real but secondary,” he added, “and enterprise comfort with self-hosted models is a nice side effect, not the reason.”
This suggests that the continued dominance of proprietary models doesn’t necessarily mean that open models are losing the competition. It may simply mean, as Zhang argues, that “enterprise AI as a whole is at the very beginning of the maturity curve.”
Enterprise AI Moves at Enterprise Speed
This argument is consistent with a theme I’ve discussed in several previous blogs: AI technology may be advancing at extraordinary speed, but enterprise adoption is moving much more slowly.
In a September 2025 blog, “The AI Revolution Will Happen in Enterprise Time,” I referenced a Goldman Sachs article, “The Outlook for AI Adoption as Advancements in the Technology Accelerate,” which noted that despite high expectations among executives, “AI is changing business much more slowly than expected.” While technical achievements are racing ahead, enterprise adoption still has to overcome corporate inertia and demonstrate a convincing return on the enormous investments being made.
An article in The Economist similarly asked, “Why is AI so slow to spread?” “Talk to executives and before long they will rhapsodise about all the wonderful ways in which their business is using artificial intelligence.” But announcing AI projects and fundamentally transforming business processes are two very different things.
I also discussed this issue in a subsequent blog about “AI as a Normal Technology,” a research paper by Princeton computer science professor Arvind Narayanan and PhD candidate Sayash Kapoor.
They argue that, as with earlier general-purpose technologies, the economic and societal impact of AI will unfold gradually because “diffusion is limited by the speed of human, organizational, and institutional change.”
The ultimate impact of AI won’t be determined simply by how rapidly model capabilities improve. It will depend on how quickly those capabilities are translated into useful applications and diffused throughout organizations and the broader economy.
This helps explain why proprietary models may dominate enterprise AI today even if open models eventually become much more important.
Most enterprises are still experimenting. They are trying different models, identifying promising use cases, redesigning workflows, and figuring out where AI can deliver measurable value. They often haven’t yet accumulated the experience, data, infrastructure, or specialized expertise required to optimize and fine-tune smaller open models.
At this stage of the maturity curve, it’s simply easier to plug into a highly capable proprietary frontier model. Organizations don’t have to manage the underlying infrastructure, and they retain the flexibility to experiment and iterate as their understanding of the application evolves.
What Happens as Enterprise AI Matures?
Zhang believes the balance will eventually shift.
“If that’s right, then every use case being prototyped on a frontier model today is a future open source migration,” he wrote. “As deployments mature, companies will do what we did: distill, fine-tune, specialize. The frontier labs will keep owning discovery. Open source will increasingly own production.”
It’s a compelling hypothesis, but the outcome isn’t necessarily inevitable.
Proprietary AI providers aren’t standing still. They can lower prices, offer smaller and faster models, expand fine-tuning and customization capabilities, and provide managed infrastructure that many enterprises may prefer to operating their own open AI systems.
And, as Zhang himself acknowledges, fine-tuning takes considerable effort, and many organizations don’t yet have the necessary resources or expertise. A use case generally needs to be sufficiently valuable, stable, and deployed at enough scale for the investment to make sense. Organizations also need enough high-quality data to ensure that a smaller specialized model can perform as well as a frontier model on the particular task.
Until those conditions are met, using a proprietary frontier model may remain the simpler and more economically sensible choice. Zhang therefore expects open models’ share of enterprise LLM spending eventually to rise — but not necessarily anytime soon.
The future may not belong exclusively to one or the other.
Enterprises may increasingly use different models for different stages and purposes — powerful frontier models to explore new applications and specialized open models where workloads are stable, scale is large, and customization, latency, control, or economics justify the additional investment.
The result could be a heterogeneous enterprise AI environment in which proprietary and open models coexist, with organizations selecting the right model for the particular problem they are trying to solve, as is the case with open source software (OSS), where despite major advances over the past three decades, OSS coexists with proprietary software.
Today’s overwhelming share of enterprise spending going to proprietary models may tell us less about the ultimate competitive position of open AI than about the still-early maturity of enterprise AI itself.
The history of enterprise technology suggests that the transition from experimentation to large-scale production takes time. As AI follows that same path, the balance between proprietary and open models will likely change with it.