Stripe in talks to acquire OpenRouter for ~$10B

OpenRouter's business is thinner than the price tag suggests: it takes roughly a 5 percent commission on the inference spend that flows through its routing layer, and doesn't own a single model itself ^[1]. Stripe already sits behind that revenue today, handling OpenRouter's invoicing, tax collection and fraud detection ^[2]. So what does an outright purchase add? Visibility into everything Stripe can't see as a mere payments vendor: which of the 400-plus models from more than 60 providers developers actually choose ^[3]. That data - not the routing code itself, which is straightforward to replicate - looks like the real asset. FourWeekMBA's read on the deal gets at the same idea from the strategy side: 'The models will keep competing; the rail that bills them is a different kind of asset entirely.' ^[4]It's also why developer skepticism about OpenRouter's technical moat has been loud online - critics note that swapping to a rival gateway is often just a config change - which may be beside the point if Stripe isn't buying lock-in but a live feed of the AI economy's spending patterns.

This isn't Stripe's first move here. In January 2026, it closed its acquisition of Metronome, a real-time usage-tracking and billing platform built for AI companies ^[2]. Metronome answers how AI usage gets measured and charged; OpenRouter would answer which model actually processes the request in the first place. Put together, they'd give Stripe a foothold on both sides of nearly every AI transaction that runs through third-party infrastructure ^[2]. Stripe has been explicit that it sees this as a natural extension of its original mission: 'As tokens become increasingly fungible with money, streaming payments in real time is an important part of Stripe's economic infrastructure for AI.' ^[5]Read that way, OpenRouter isn't a bolt-on AI feature - it's the second half of a billing-rail strategy Stripe started building roughly six months earlier.

The valuation move is the part straining credulity. OpenRouter raised a $113 million Series B in May 2026, led by CapitalG, at a $1.3 billion valuation ^[1]. Roughly ten weeks later, reported acquisition talks put the price near $10 billion ^[3]- a 7-to-8x markup on that prior valuation. Revenue tells a more modest story: annualized revenue had only reached about $50 million by March 2026, up from $19 million at the end of 2025 ^[2]. That's a company with roughly 48 employees and $164 million in total funding raised to date ^[1]being valued at close to 200 times current revenue. Whether that's justified depends entirely on whether the data-and-distribution thesis holds; if it doesn't, this looks like a land-grab price set by fear of missing the routing layer rather than by OpenRouter's own fundamentals. Either way, a deal at this level would reset the going rate for adjacent routing and billing-layer startups, effectively repricing tools like Portkey, Martian and Not Diamond as strategic infrastructure rather than developer utilities ^[4].

OpenRouter's whole pitch has been neutrality - a single API across providers so developers aren't locked into any one model. That's harder to sustain once a single company, with its own commercial incentives, owns the router. The competitive backdrop makes the timing pointed: Databricks had reportedly already approached OpenRouter about its own acquisition ^[6], and Cursor shipped its own model-routing product the same week the Stripe talks surfaced ^[5], on top of existing routing rivals and hyperscaler-bundled alternatives from AWS, Google and Microsoft. Stripe is also not the only company chasing the broader AI-agent payments layer - Circle, Google and a Coinbase-led consortium are all building competing protocols for how autonomous agents will eventually pay for things ^[7]. That race may be getting ahead of the actual market: on-chain AI-agent settlement volume totaled only about $73 million over the year to April 2026, against $46 trillion in total annual stablecoin settlement volume ^[8]- a reminder that the infrastructure land grab is running well ahead of proven usage. And none of it is locked in yet: reporting on the talks stresses the deal could still collapse or draw a competing bidder before anything is signed ^[3].