According to reports, to reduce AI inference costs, Meta is replicating OpenRouter by developing a model routing tool called Switchboard. Its core logic involves evaluating task difficulty to route simple requests to cheaper, smaller models, avoiding wasted computational power on large models. This tool is not only for internal cost reduction but may also be released externally in the future, representing Meta's attempt to open up new revenue streams.
Under the pressure of continuously ballooning AI infrastructure expenditures, Meta is seeking ways to cut costs internally.
On July 21, according to a report by tech media The Information, Meta's internal AI incubator, AAI Labs, is developing an AI model routing tool named "Switchboard." Its core logic is highly similar to OpenRouter's Auto Router product—by scoring task difficulty, it routes simple requests to cheaper, smaller models, thereby reducing overall inference costs.
Switchboard is still in its early stages, and it is uncertain whether it will ultimately be launched. However, according to internal documents obtained by The Information, the Meta team has clearly outlined two potential paths: first, deploying it internally to compress costs; second, publicly releasing it to external organizations that run AI coding agents at scale.
Analysis points out that this means the tool is not just a cost-cutting measure but could also become Meta's attempt to carve out a new revenue source in the AI tools market.
It is also reported that the proposal of this project directly addresses a current pain point for the company: paying for top-tier model pricing for every coding request, including those simple tasks that could easily be handled by smaller models.
Targeting the Pain Point: Inference Cost is the Biggest Obstacle to Large-Scale Deployment
The report states that the rationale for the Switchboard project is expressed quite bluntly in internal documents: "We pay for top-tier model pricing for every coding request, including the simple ones."
The document further points out that most coding agent tasks can be fully handled by smaller models, with only a few tasks truly requiring the capabilities of cutting-edge large models. However, the current situation is that "all requests are sent to the same model, leading to excessive spending on simple tasks or underperformance on complex ones."
The document explicitly identifies inference cost as the "primary obstacle" to more widespread internal deployment of agents and bluntly states: "Cost is the key factor limiting our ability to run agents at scale."
This statement aligns with a series of recent actions by Meta to control AI spending. According to a previous report by The Information, in June of this year, Meta notified employees that it would begin setting caps on AI token usage just weeks after encouraging broader company-wide adoption of AI tools, while also building an internal platform to track AI spending and enforce token budgets.
It is worth noting that the Switchboard project belongs to Meta's AAI Labs, an internal incubator under the Meta Applied AI Engineering team, officially established in March of this year. According to internal documents reviewed by The Information, this mechanism allows employees to submit proposals for AI products and services. Once approved, small teams are responsible for building them, with the opportunity for public external release.
As of July this year, AAI Labs has approved approximately 200 projects, covering three main areas: consumer products, developer tools, and internal infrastructure. Switchboard is one of them.
This mechanism reflects the broader strategic vision of Meta CEO Mark Zuckerberg—leveraging AI to enable small teams to build products quickly. Zuckerberg told analysts in April this year that AI agents mean "small teams can make very rapid progress" and predicted that this technology would drive "a lot of innovation." He also stated that Meta could potentially build up to 50 new applications.
The Model Routing Arena: Beyond OpenRouter, Giants Are Entering the Fray
The model routing arena that Switchboard is targeting is attracting increasing attention.
OpenRouter has gained considerable popularity among developers by helping them access various AI models at lower costs. According to a report by The Information last week, OpenRouter has begun discussions with a larger tech company regarding a potential acquisition, a deal that could push its valuation to tens of billions of dollars—the company was valued at $1.3 billion in April of this year.
Wider attention to the model routing concept stemmed from the routing feature built into OpenAI's release of GPT-5—this feature automatically switches to cheaper models when user prompts are relatively simple. Since then, companies like Databricks and Palantir have also developed their own routing tools to manage costs and improve efficiency.
Meta's self-development of Switchboard is both an active response to its own cost pressures and a strategic choice to deploy its own capabilities in this rapidly heating arena.
Greater Ambition: AI Investment Seeks Diversified Monetization
Behind the Switchboard project lies Meta's overall desire to transform its massive AI investments into new tools, new businesses, and new revenue sources.
Meta previously estimated that its spending on AI infrastructure and other equipment and facilities this year could reach as high as $145 billion, more than double the level of 2025. At the same time, Meta is reorganizing its engineering teams to strengthen AI development capabilities.
Projects incubated by AAI Labs go beyond Switchboard. According to another internal document obtained by The Information, AAI Labs is also developing an AI tour guide application for drivers. It can run via Apple CarPlay and Android Auto, using AI to explain nearby landmarks and allowing drivers to ask questions.
The document positions this product as an extension of the Instagram map experience, potentially integrating location-based Reels content, travel recommendations, and even Meta Ray-Ban smart glasses in the future.
The report states that these projects collectively outline Meta's path: starting with employee ideas, rapidly prototyping AI products, and then selectively launching them to the external market—exploring incremental revenue space beyond its advertising business while controlling costs.
Comments