-

Workweek Newsletter {beacon}

Alex Johnson

Aug 21st, 2026

{cta_url_read_in_browser = community_base_url + "/library/" + article_id + "?utm_source=newsletter&utm_medium=email&utm_campaign=" + edition_slug + "&utm_content=read_in_browser"}{cta_url_read_in_app = community_base_url + "/library/" + article_id + "?utm_source=newsletter&utm_medium=email&utm_campaign=" + edition_slug + "&utm_content=read_in_app"}{cta_url_join_conversation = community_base_url + "/library/" + article_id + "?utm_source=newsletter&utm_medium=email&utm_campaign=" + edition_slug + "&utm_content=join_conversation" + "#comments"} {if profile.vars.member_status == "lead" || profile.vars.member_status == "unfit"} Read in browser {else}{if profile.vars.member_status == "fit"} Read in Finity {else}{if profile.vars.member_status == "member"} Join the convo {else} Read in browser {/if}{/if}{/if}

Happy Friday, Fintech Takers!

We published A LOT of content this week:

A super depressing newsletter on Monday,
an article on cash flow intelligence on Tuesday,
a podcast on small business lending on Wednesday,
and a deep dive essay on personal lending, direct pay, and the value of certainty in consumer lending yesterday.

Whew!

I don’t want to overwhelm you — and my brain needs a rest — so the plan for today’s newsletter is simply to comment on one fintech news story from this week.

Easy peasy!

- Alex

Was this email forwarded to you?

Subscribe Here

DEEP DIVE

Five Reasons Stripe Might Have Bought OpenRouter

When the most highly valued private fintech company makes, by far, the largest acquisition in its history, I will take time out in my week to write about it!

What are the terms of the deal?

Stripe didn’t confirm the amount, but the New York Times pegs it at roughly $7.5 billion, with $1.5 billion allocated specifically to the founders and the remaining $6 billion going to investors and other equity holders.

Before this, Stripe’s largest acquisitions were Bridge ($1.1 billion) and Metronome ($1 billion), so this one is quite a step up!

What is OpenRouter? What does it do?

OpenRouter was founded in 2023. It raised a $12.5 million Seed round and a $40 million Series A in 2025, which valued the company at more than $500 million, and a $113 million Series B in May of this year at a $1.3 billion valuation. And now it is being acquired by Stripe for $7.5 billion, which is roughly 6x the company's last valuation.

OpenRouter is an AI model aggregation and orchestration platform. Instead of signing separate contracts with OpenAI, Anthropic, Google, DeepSeek, and dozens of others — and writing separate code to integrate with each — a developer writes to one API endpoint and OpenRouter passes the request along to whichever model they want. Today that's more than 400 models from over 80 providers. Because the connection is standardized, swapping models becomes a configuration choice rather than an engineering chore. Customers can route the cheap high-volume work to a cheap model, send the hard stuff to an expensive one, and fail over automatically when a provider goes down. The company doesn't mark up the tokens it resells. It makes money by charging 5.5% when you load credits into your account.

OpenRouter has more than 8 million users and moves roughly 25 trillion tokens a week, up from 5 trillion at the end of 2025 — a 5x increase in about eight months. Revenue went from around $19 million at the end of 2024 to roughly $140 million annualized by this July. Customers include Nvidia, Zoom, and Lovable. And in Stripe's own telling, token consumption on the platform has been compounding at 9% per week year-to-date.

Why did Stripe acquire OpenRouter?

I have five different possible reasons, which are, obviously, not mutually exclusive.

1.) Stripe isn’t a payments company. It’s a developer company.

The theory here is actually pretty simple. Stripe believes that great software developers are the key ingredient of the fastest growing and most successful companies in the world and if it can offer the best tools for developers to build with, it will attract the best developers and, thus, the fastest growing and most successful companies as Stripe customers.

The company says exactly this in its letter to investors about the OpenRouter acquisition:

Businesses on Stripe are growing significantly faster than the economy as a whole, owing to a combination of selection effects (innovative businesses are likelier to choose Stripe), as well as the cumulative impact of the thousands of small improvements we make each year to accelerate revenue growth for our customers.

Be the vendor that the most innovative companies choose and find as many ways as possible to help them grow.

Simple.

Stripe sees AI model aggregation and orchestration as one of the most important categories in developer infrastructure right now and it apparently felt it needed to immediately buy (rather than slowly build) its way into that category. Here’s the investor letter again:

We see capital and intelligence are becoming the two digital flows undergirding every business. Up until now, every developer has needed a straightforward and reliable way to manage their revenue pipeline, and serving this need gave rise to Stripe. Going forward, however, every developer will also need a straightforward and reliable way to manage their intelligence pipeline.
OpenRouter is exceptionally useful for any developer and Stripe is one of the world’s largest developer platforms. As such, we think that there will be many benefits and efficiencies in bringing these two core needs together.

Notice that in this part of the letter, Stripe is speaking about revenue (what it does today) and intelligence (what it wants to do tomorrow) as separate pipelines. Separate categories of infrastructure for developers that undergird every business.

Both are important if you want to remain the number one choice for the best developers in the world, so Stripe bought its way into the intelligence pipeline before a different developer-obsessed company could.

2.) AI routing is a cost problem.

Perhaps there’s more synergy between Stripe’s fintech roots and its AI ambitions than that last justification would suggest. Here’s the investor letter again:

Intelligence is special: it is expensive, heterogeneous, and constantly changing. As with financial capital, businesses must reason about cost and return of every unit in a deliberate and granular way. How valuable is this task? With which models can it be best handled? Who will pay, and when, and what is the time-value of that delay?

The most obvious application of this idea is expense management.

AI is becoming an incredibly costly category of spend within every company and the days of tokenmaxxing are long over. CFOs need to understand and control AI spend and, as OpenRouter itself advertises on its website, AI model orchestration can help companies “keep costs in check without sacrificing speed.”

However, if you think this use case for AI routing doesn’t sound very Stripe-y, I’d agree with you.

And if you think it sounds more like Stripe’s “WE’RE NEVER FUCKING GOING PUBLIC!” fintech peer, I’d direct you to this press release, which, coincidentally, came out the same day that Stripe announced the acquisition:

Ramp today launched Router.com, a single endpoint to every major AI model. Each request is routed to the lowest-cost model that meets the required performance level. Unlike a routing layer alone, Router connects model selection decisions to Ramp's broader AI spend visibility and controls, giving companies a clearer view of what AI work costs, who owns it, and where to optimize. Customers already using Router have reduced their inference costs by 40% on average. It is live now at router.com and free through 2026.

The most important detail in that announcement is the “free through 2026” part. AI routing is (for the moment) a customer acquisition play for Ramp; a hook to draw more companies into its platform and interchange-centric business model.

Stripe will use OpenRouter to help its customers save money on AI as well, but cost savings likely isn’t the primary motivator. You don’t pay $7.5 billion for that.

3.) AI routing is a credit problem.

So, what is the fintech play here, if it’s not expense management?

I think this quote, from slightly further down in the investor letter, is interesting:

We’re excited to extend Stripe’s financial capabilities to this new domain and to help businesses effectively allocate the new currency of intelligence capital.

“Allocate” is the key word.

You don’t allocate an expense. You allocate capital. More specifically, in the parlance of one of Stripe’s most successful business lines, you lend capital.

This is me theorizing a little bit, but stay with me.

Picture a company selling an agentic AI service — the kind of product where a customer kicks off a task, an agent chains together a few dozen model calls to complete it, and the customer gets billed for what they used. There are thousands of these companies now, and there will be tens of thousands soon.

This type of company is unusual in that the event (an AI agent calling a model) that generates its revenue is the same event that generates its costs. Stripe already sees the revenue side because it does billing and the payment processing, and with Metronome (the usage-based billing platform it bought in January for roughly $1 billion) it now sees it at the level of the individual metered event.

What Stripe hasn't been able to see is the cost side. Inference is the largest and most volatile line in these companies' cost of goods sold (COGS), and it's incurred per request, changing every time a model gets cheaper or a workflow gets more agentic.

OpenRouter is where that number lives.

Put the two together and, in theory, Stripe will be able to see the actual gross margin on a specific customer, on a specific workload, on the day it happens.

This is very different from how Stripe Capital works today. Stripe Capital advances a lump sum for a flat fee and repays itself by withholding a percentage of the sales flowing through a business’s Stripe account. It's a clever twist on a classic small business lending product (merchant cash advance), but its innovation is on the repayment side. Stripe sits in the payment flow, so it stands at the front of the line and takes its cut before the money reaches the merchant. The underwriting underneath it is essentially a revenue forecast: How much will flow through this account, and can I get paid back out of it?

Structurally speaking, the insights Stripe will soon have access to — at least for businesses generating revenue and costs from AI models — are more similar to the insights that traditional banks have used to underwrite businesses for a wide variety of capital products, far beyond merchant cash advances.

Except Stripe’s insights will be much higher fidelity.

Think about what banks get today when they underwrite commercial loans: Quarterly financial statements that are self-reported, backward-looking, and aggregated to the point of uselessness. A gross margin line that tells the bank nothing about whether profit is expanding or contracting, or which customers, specifically, are dragging it down. Bank underwriters have spent centuries building heuristics to compensate for the fact that they cannot really see inside the business.

Stripe will be able to see inside the business, continuously. It will be able to watch margin compress weeks before the company's own finance team closes the month. It will be able to tell which cohort of customers is unprofitable. It will be able to distinguish between a company burning cash because it's growing and a company burning cash because its unit economics don't work.

With this granular, real-time insight, Stripe could decline businesses that look fine on paper. It could also lend to companies that look uninvestable by every traditional measure — unprofitable, no collateral, eighteen months of operating history — because it can see that every marginal customer contributes gross profit.

4.) Stripe is buying a better take rate.

Credit to Matt Janiga for this point. It’s a simple one, but shouldn’t be overlooked:

To the extent Stripe is using cashflow from what is historically a low-multiple business (payments) to buy higher multiple revenue businesses (orchestration software), that feels like a good trade.

Indeed!

AI has been a massive tailwind for Stripe’s traditional payments business over the last couple of years. As the company itself noted in its investor letter, “today, 88% of the Forbes AI 50 (including OpenAI and Anthropic) are building on Stripe.”

However, Stripe's blended net take rate across all revenue is roughly 0.36% (according to my math and based on Sacra’s numbers), and that number has been trending down for years as competition has increased and Stripe has moved up market to serve more enterprise customers. OpenRouter, by comparison, charges 5.5% on the prepaid token credits it sells. Standard card processing runs 2.9% plus 30 cents, so what OpenRouter actually keeps is somewhere around 2.5% (ironically, today a big chunk of OpenRouter's gross take rate goes to Stripe, which it uses to process payments).

A 7x bump in take rate is a good trade, if it holds. But, of course, it may not. High take rates in commoditizing markets have a way of becoming low take rates, and Ramp isn’t the only company that has shipped a competitive AI router recently.

5.) Stripe needs its investors and employees to believe it's an AI company.

The weirdest part of Stripe’s investor letter, by far, is a section at the beginning, which starts off:

It’s a fuzzy and perhaps already overworked term, but we decided that January 1st marked the beginning of the singularity, and we have since been operating on that basis.

The company never explains what the singularity is or why it decided that January 1st was the date that it occurred. Nor, for that matter, does it attempt to wrestle honestly with the possibility that if the singularity has indeed occurred and humans have permanently lost control of technological growth, then maybe discussions about usage-based pricing models and agentic commerce protocols are trivial and meaningless.

However, at the end of the investor letter, Stripe does take time to comment on how well the business is doing, using distinctly public market-y framing:

We’re pursuing this expansion while paying close attention to shareholder returns. The profitability of Stripe’s core payments engine allows us to make acquisitions like these without significantly diluting existing stockholders. Even while undertaking significant organizational investment and M&A, Stripe’s share count is lower today than three years ago. Stripe’s share price has compounded at 31% since our Series D fundraise 10 years ago, versus 14% for the S&P 500 and 18% for the Nasdaq over that same period.

Before stating the obvious:

Stripe is, of course, a private company today.

And then invoking the singularity one more time to justify why it needs to stay that way:

We view this as a growing advantage as we venture into the vicissitudes of the singularity. The world is becoming harder to predict and we expect that deft helmsmanship will be required of every company.

Setting aside the value of OpenRouter as a higher take rate business and as a lure for developers and as a possible bridge to new and exciting fintech business lines, it does feel, just a little, like Stripe made the biggest acquisition in its history in order to demonstrate how serious it is about a technology that will render the world so uncertain and difficult to predict that it would be idiotic for Stripe’s investors and employees to ever insist on the company going public.

✈️ FinovateFall | September 9-11 | New York City

My can't miss fall conference! September in New York is glorious and the fintech conversations will be too.

✈️ Cash Flow Intelligence Summit | September 10 | New York City

Nova Credit has rebranded this from the "Cash Flow Underwriting Summit" to the "Cash Flow Intelligence Summit." Come find out why.

✈️ FDATA Global Open Finance Summit | September 17 | Toronto

This will be my first time at an FDATA event and my first time back to Toronto in a long time. If you work in open banking in Canada and want to yell at me for my bad takes in the past, this is your chance!