A map, not a marketing document
Every February, a group of fintech founders, payments operators, and venture investors clear their afternoon for the same task: reading the Stripe annual letter cover to cover. It is the closest thing the industry has to a state of the union — not because Stripe runs the industry, but because the Collisons have spent fifteen years thinking harder about payments infrastructure than almost anyone else, and they tell you what they're seeing.
The 2025 letter is no different. Nine thousand words. Dense, specific, occasionally philosophical. It is not a marketing document. It is a map.
This is what matters in it if you're building in fintech, or thinking about it.
What the letter is and why it matters
The Collisons have written an annual letter since 2019. Over time they've evolved from product updates into something closer to a Berkshire Hathaway-style strategic narrative — the company's reading of the market, with specific data, candid opinions, and the occasional uncomfortable prediction.
Founders read these letters the way investors read Buffett's. Not for the facts alone, but for the worldview behind them. The 2025 letter tells you what Stripe believes about where money is going over the next decade. If you disagree, that disagreement should be informed. If you agree, you need to be acting on it.
1. Stablecoins reached escape velocity
The letter reports that stablecoin transaction volume crossed $1 trillion in 2024, growing more than 3x year-over-year. Stripe's own stablecoin product, launched in late 2024, processed billions in volume within months of going live.
What this means for you: the "stablecoins are crypto theater" argument is finished. Whatever your prior view, the data is now incompatible with dismissal. Stablecoins are a functioning payments rail with real volume, real businesses, and real use cases — cross-border B2B, remittances, payouts to contractors in countries with weak banking infrastructure, and increasingly USD-denominated transactions in regions where dollar access is constrained.
If you're building any product that touches international money movement, you need a stablecoin strategy. Not a launch plan necessarily, but a clear position on whether and when you'll integrate. The companies that arrive at this question in 2027 will be late.
2. AI agents are the new merchants
The letter spends meaningful real estate on what Stripe calls "agentic commerce" — the idea that AI agents will increasingly conduct transactions on behalf of humans, businesses, and other agents.
This is not science fiction in the letter's framing. Stripe is building infrastructure for it now. Tools that let agents authenticate, hold funds, make purchases, and respect spending controls set by their principals. The letter cites early examples: AI assistants booking travel, autonomous procurement systems, software agents that buy and sell services from other software agents.
What this means for you: payments infrastructure is being reframed. For a decade the central question was "how do we make checkout easier for humans?" The new question is "how do we make payments programmable for non-humans?" These are different problems with different solutions.
If you're building AI products, take payments seriously now — your customer of 2027 may not be a person at all.
The death of card networks is not near. But the architecture around them is changing in ways that won't be reversible.
3. The card networks aren't dying. But the architecture is changing.
The letter is careful here, but the signal is clear. Visa and Mastercard are not under existential threat in 2025. The card networks remain dominant, profitable, and deeply embedded in commerce infrastructure globally.
But the letter notes — without polemic — that more payment volume is moving onto rails that don't touch the card networks at all. Stablecoins, real-time payment systems like FedNow and UPI, account-to-account transfers, and increasingly the agentic commerce flows mentioned above. The card networks won't disappear. But their share of global payment volume is going to decline, slowly and then faster.
What this means for you: if your fintech business model depends on interchange revenue, model it under a scenario where that revenue shrinks meaningfully by 2030. If your business sits on the rails being built around the card networks, you may be in the right place.
The hidden thread: AI commerce is real, not hype
The most important thing in the letter isn't any single section — it's the through-line connecting them. Stablecoins, agentic commerce, programmable money: these are all the same trend viewed from different angles. The Collisons are telling you that the shape of commerce is changing in a way that requires new infrastructure, and they're building it.
Twelve months ago you could read a letter like this and treat AI commerce as 5-10 years out. The 2025 letter doesn't give you that out. The stablecoin numbers are real today. The agentic commerce examples are happening today. Whatever your timeline for "AI changes payments," compress it.
What to do about it
Three concrete actions for fintech founders:
Assume agentic transactions are 24 months out, not 60. If you're building payments products, your product needs to support AI agents as principals. Authentication flows, spending limits, audit trails — all of it gets re-architected for non-human actors.
Take payments infrastructure seriously if you're building AI. The AI companies that win commercially in the next cycle will not be the ones with the best models. They will be the ones whose AI products can actually transact — pay for services, sell their own services, manage funds. That requires payments primitives most AI teams haven't thought about yet.
Re-read the programmable money section. It's the densest part of the letter and the one most founders skim. If money becomes programmable in the way Stripe is describing, the implications cascade through dozens of fintech verticals — lending, treasury management, payroll, invoicing, accounting. New companies will be built on this primitive.
The takeaway
The Stripe annual letter has always been worth reading. The 2025 letter is the one that crystallizes what's been forming for two years: AI is not a feature being added to payments. Payments are being rebuilt around AI as a first-class actor. Stablecoins are not a niche. They are a rail. And the card networks are about to enter a long, gradual structural decline that smart founders will build around.
Whatever you're building, the next 24 months will reward people who took these signals seriously now and punish those who waited for unambiguous proof.
Want to understand the financial mechanics behind AI commerce and stablecoins? Try the Agentic Commerce and Stablecoins packs on Gargiulo — ten scenarios across both, grounded in real market data. Sterling has notes.
Sources: Stripe Annual Letter 2025