I will use my judgment to determine which AIs are best at software engineering. Each AI will be from a different company. For example, this might be the top software engineering AIs from each of OpenAI, DeepMind, Anthropic, Meta, and DeepSeek.
Depending on progress the AIs may be better described as an "agent" or "assistant" or some new thing. If a general purpose AI agent is one of the best software engineering AIs then this also counts.
If OpenAI licenses an AI to Microsoft, and Microsoft makes it generally available via Azure or Github on a non-exclusive basis, that is not an "exclusive use" for the purpose of this market. The intention of this market is to predict whether a few large tech companies will attempt to monopolize AI models to prevent non-tech companies from creating their own software solutions. The question is whether AI usage is exclusive, not whether AI hosting is exclusive.
Unreleased models are not included. For example, if a model is not released because it risks causing human extinction, or because it is still being trained, or because it has a potty mouth, or because it cannot be secured against model extraction, or because it is undergoing recursive self-improvement, or because it is being used to generate synthetic data for another model, or any similar reason, that model is ignored for the purpose of this market.
However, if a model is ready for release, and is only not being released in order to monopolize its use in creating commercial software, then this counts as "exclusive use".
This does not include work to fine-tune a model or provide it additional tooling or otherwise make it more effective when working at a particular tech company or on a particular tech stack. The market is about general purpose software engineering models.
I do not intend to bet in this market due to possible subjective interpretation of its terms. Please ask for clarifications as needed.
I may need to extend this market a short distance into 2026 if a model is being used for software development as part of a "limited beta", and it's not clear if this is genuinely a pre-release beta, or a deceptive way of a company getting exclusive use to gain a competitive edge.
1,000My prediction: 20%. It's a short timeline, and "nothing ever happens". That said I cast the net wide by resolving YES if any of the top five labs do this. If it resolves YES I see this as most likely to occur via the OpenAI/Microsoft partnership, with Microsoft putting pressure on OpenAI not to release a "beta" model that is used internally at Microsoft and giving Microsoft a substantial edge over their rivals.