Stripe added $500bn of TPV in the past 12 months. Valued at $159bn. Rumored to be acquiring PayPal. And possibly powering Meta's stablecoin comeback.

Each part of that sentence is mindblowing.

But what it speaks to is the growth dividend. Stripe has consistently indexed itself towards new, smaller companies with hypergrowth potential. In a market where PayPal stock was crushed for not growing, that indexing is now the difference between being the acquirer and being the target.

There are now two types of payments companies.

Those who are growing fast.

And M&A targets.

Before the week's news landed, the market had already panicked.

On Sunday, Citrini Research published a speculative scenario paper imagining 2028. AI agents had begun routing around card networks. Settling via stablecoins. Targeting the 2-3% interchange fee as an obvious inefficiency. Settlement near-instant. Transaction cost measured in fractions of a penny.

By Monday morning, Visa fell nearly 5%. Mastercard dropped almost 6%. Amex plunged over 7%.

Let me say this clearly: this is a wild overcorrection. Agents using stablecoins does not equal card networks die. Cards still offer consumer protections, rewards, credit, and a trust layer that stablecoins haven't replicated. Even for agent-to-agent payments, cards and other rails will be just fine.

But the Citrini paper raises a more subtle question. Not will cards survive — of course they will. But where does the next wave of growth come from, and are you indexed towards it?

Any company not catching the AI and stablecoin tailwinds is being hammered. That's the lens through which this entire week makes sense.

(I cover the market reaction in much more detail in the Things to Know section below)

Into this anxiety, Stripe dropped its annual letter. The timing could not have been more pointed.

The key message was the "sorting machine." The economy is bifurcating. You're either indexed towards computing, software, and AI — and therefore growing — or you're not.

Nearly half (46%) of US GDP growth is driven by computing and software. Stripe is first-in-line for any new company in that segment. Their 2025 cohort is growing 50% faster than 2024. More new companies joined Stripe than any year prior.

This is the growth flywheel.

Patrick Collison described AI and stablecoins as the two "gale force winds" impacting the industry. While many payments companies spent Monday watching share prices crater, Stripe is winning in both.

On AI. Agentic payments are still early. Salesforce said 1 in 6 Black Friday / Cyber Monday purchases were assisted by AI — but almost none were made directly by agents. The whole industry is betting that changes fast. PayPal, Google, Stripe, OpenAI, Adyen, Checkout, Visa, Mastercard. Everyone is building protocols for it.

Why, if there's no volume?

Because AI has an annoying habit of being useless until it's suddenly not.

Stripe launched the Agentic Commerce Protocol with OpenAI, only to have Adyen, Shopify, Checkout, and Google launch the competing Universal Commerce Protocol a month later. This is what happens when you confuse something being technically open with the optics of being open. If other payment companies aren't there at launch, they see it as a Stripe thing, not an industry thing.

Stripe can still quickly support whatever gets traction. Not a mortal blow. Just a classic example of the engineering-led mindset and its limits. And remember: Stripe has most of the AI companies as clients. That matters more than protocols right now.

Then there’s machine-to-machine payments. What happens when AI agents need to buy from each other? We don’t know when that will happen or what format it will take. But Stripe is already building for that future.

There's a generally accepted orthodoxy that agents will eventually make an order of magnitude more transactions than humans do today. Even if there's a 10% chance of that outcome, you want to be indexed towards it.

On stablecoins. Crypto is entering its winter. Stablecoin summer continues.

And Tempo, my employer, is building stablecoin settlement infrastructure for exactly this moment — instant, compliant, built for the reliability bar that regulated finance demands. (Disclaimers below.)

The next era of stablecoins is about moving from cross-border long-tail use cases to something backwards-compatible with highly regulated TradFi and forwards-compatible with machine payments.

Stablecoins are the natural money for AI. If agents prefer stablecoins because they're better, faster, and cheaper, you want at least the option to support them as a rail. Ideally, you want to be a main character in that story

Now combine the two. Stablecoins and AI. Finkle and Einhorn.

Being indexed to growth companies, and leading from the front on AI and stablecoins, and you get a double flywheel effect.

Meta and "coin" in the same sentence. You immediately think Libra. The ill-fated 2019 attempt to build a new currency for billions of users. Cue central bank panicking about sovereignty. 

"That thing's dead," Zuckerberg reportedly told Stripe's John Collison.

The 2026 version could be wildly different.

Meta using stablecoins is meaningfully different from Meta building its own currency. 

At a minimum, Meta could use stablecoins to improve its corporate treasury and cross-border flows to the hundreds of countries where correspondent banking is slow or expensive.

But think about AI. Meta is spending tens of billions in 2026 capex, mostly on AI. Building agents that shop and transact autonomously. Agentic commerce.

In that world, stablecoins aren't the product.

They're the settlement layer for AI-driven payments.

When an AI agent buys something for you on Instagram, it needs to settle instantly, cheaply, across borders. Traditional rails can't do that at Meta's scale or speed. Stablecoin rails across Facebook, Instagram, and WhatsApp would give 3 billion+ users access to faster, cheaper value transfer.

Being indexed towards growth means being closer to the things that drive tomorrow's growth. AI and stablecoins are the two strongest candidates.

The Bloomberg reports that Stripe is "considering" acquiring PayPal may or may not materialize (it’s likely just bankers testing the market). But it reveals one thing: PayPal is an M&A target, not a growth company.

$1.79trn of TPV in 2025. Just 7% growth. Stripe delivered 34%. Checkout.com delivered 64%. Once valued at ~$360bn in 2021, PayPal now sits around $44bn. It shed a third of its value in 2025 alone. And $7bn of the current market cap was added Tuesday just from the Stripe rumor.

This follows the removal of Alex Chriss. The board said the "pace of change and execution did not meet expectations." Branded checkout growth decelerated to 1% in Q4. PayPal has become a cautionary tale of what happens when you buy your way into growth but never properly integrate.

On paper, the "synergies" sort of write themselves. (God, I hate that word.)

(Each of those bullets could be an entire essay.)

But payments mega-M&A has a near-perfect failure rate. FIS bought Worldpay for $43B in 2019, then divested it at $18.5B enterprise value four years later. The "cross-selling" never materialized. Bank of America walked immediately.

Migrating merchants is insanely hard. Each one is a full project — data mapping, systems change, existing contracts and guarantees on the merchant side.

Stripe has been exceptional at M&A historically. They absorb founders, ship fast, make things work. Talk to Bridge, Privy, or Orum. Happy founders. Doing some of the best work of their careers.

But PayPal isn't a 50-person startup. It's 400M accounts, a brand new CEO, and a decade of cultural debt.

That's a completely different integration problem.

Adyen has always been the king of enterprise. Stripe, by its own admission, is still learning that motion. The market dinged Adyen for guidance of growth being lower in the future, but it's to my mind still the only company that can go pound for pound with Stripe in almost every dimension.

The comparison in TPV is striking. Both left 2024 with roughly $1.4trn. By end of 2025, Adyen was at $1.65trn — adding $250bn. Stripe doubled that pace. 

Why?

The AI companies. The AI companies are adding all the TPV.

Still, enterprise companies that want feature-rich, in-store and online, ultra-efficient payments — Adyen is a force. As AI companies scale, they'll go multi-processor. Everyone does eventually. Where do you go when you want a second processor alongside Stripe? Adyen. Just as Shopify has.

Checkout published its own annual letter this week. $300bn of total processed volume, up 64% YoY, profitable full year. 99.999% uptime — five nines — a sign they're gunning hard for enterprise.

They now count Spotify, Uber, and Netflix as major clients. And they've rapidly jumped at supporting the UCP for agentic commerce. Why is that interesting? It's the protocol every PSP except Stripe is vocally supporting.

Sometimes there's power in being in the everyone-else category.

Checkout and Adyen — and the chasing pack of "Gen 3" processors — are well positioned. If Stripe gets distracted by a PayPal integration, these companies could be the biggest beneficiaries.

A few years back, I wrote "Stripe's Difficult Teenage Phase" on the back of a downturn and job cuts. 2023 was not kind. Products looked like "me too" offerings. Stripe was stuck refactoring old code. M&A wasn’t driving growth.

Since then, three things:

The blind spot is still enterprise. The "we'll build it, and everyone will use it because the code is beautiful" mindset. That works for indexing towards the next big thing. But sometimes you get caught offside launching protocols that are intended to be "open" and you're the only one that supports them.

Yet they're outgrowing their nearest rivals by 2:1.

And they're still unlikely to IPO. John Collison told CNBC this week that going public isn't in his top twenty priorities. They don't need to. And their best growth days may yet lie ahead.

If the AI and agentic commerce boom is anything like the e-commerce boom?

Stripe could be the world's first trillion-dollar fintech company.

ST.

Disclosures:

To state the obvious, Tempo is my employer, and Stripe is a sponsor of the Tokenized podcast. These views are very clearly my own, and while I gave the comms teams a heads up, they remain my own and 100% unedited.

1. Complyance - Compliance AI for Enterprises

Complyance provides AI agents that risk-score vendors, identify emerging risks before they escalate, and create a hub for drafting, managing, and auditing policies. They have ready-made frameworks for SOC2, PCI-DSS, and ISO 9001. It provides 24/7 monitoring of controls and connects to existing enterprise tools like Confluence, Azure, Cloudflare, Datadog, github and crucially, Office 365

🧠 GRC (Group Risk and Compliance) is a hidden iceberg of complexity in most enterprises. If you’ve never lived it, its filled with countless word documents, sharepoints and spreadsheets with audits. They’re all managed in a central hub, but updating and analyzing it is extremely manual and slow. This is a ready-made use case for AI Agents and a platform. 

2. Levi - Stablecoin payment infrastructure

Levi provides named bank accounts for GBP, USD, and EUR with tier 1 banks to provide currency conversion and cross-border payments in 70 markets. It combines crypto payment acceptance with card acceptance for Fintech companies, PSPs, and remittance companies.

🧠 I’d never heard of these guys, but they have some fairly major partners like Terrapay and Nala (likely providing last-mile payouts and pay in support). There are so many of these, but Levi is based in Switzerland, which may give it some structural advantages and a unique client base. 

3. Smart Bricks - The Real Estate Investment AI

Smart Bricks helps investors find, buy, and manage higher return, lower risk investment properties.  Their team (and AI) identify top performing property investments globally and handle 99% of the investment process. They have a self-learning AI algo that is continually helping them identify new opportunities to invest.

🧠Managing a real estate portfolio is work. Making that 100x simpler and feel almost turnkey is no easy feat. This is the kind of thing that could make real estate investing much more accessible. 

4. Birch Hill - Institutional Grade Credit Underwriting for Onchain Finance.

Birch Hill will source, structure and underwrite onchain investment opportunities such as DeFi vaults like Morpho. For investors who want to allocate to that asset class, Birch Hill helps design risk thresholds and allocation across protocols. 

🧠This isn’t really a start-up in the technology sense, but it's still fascinating. These boutiques can scale massively in the age of AI, and if all finance goes onchain, today’s smaller shop could be tomorrows big name 

1. OCC Notice of Proposed Rule Making - Opens 60 days for comments

The GENIUS Act, passed July 18, 2025, prohibits any non-GENIUS issuer from issuing stablecoins in the United States. It includes licensing and regulatory requirements and grants the OCC sole federal jurisdiction over the rule (except for state-chartered community banks that wish to issue). In addition, the OCC will have regulatory authority over foreign payment stablecoin issuers.

🧠 Yield looks fairly hard to pass through. It’s prohibited if the issuer has any contract or agreement with a 3rd party. It’s also prohibited if the 3rd party or issuer pays the holder of the stablecoin for holding, using, or retaining the stablecoin.

🧠 What about white-label stablecoins (e.g., Klarna USD)? The OCC appears to be saying, you can’t pass that yield to holders. It reads to me like Bridge could pay yield to Klarna, but Klarna couldn’t pay that to consumers or businesses. “The prohibition is also not intended to prevent a permitted payment stablecoin issuer from sharing in the profits derived from the payment stablecoin with a nonaffiliate partner in a white-label arrangement.”

🧠 Tom Noyes made a couple of great points on X.

🧠 So if you’re passing yield, the OCC assumes you’re evading them. Unless you’re a whitelabel arrangement, and merchants can independently offer discounts.

🧠 So let’s play this out. Walmart could work with Anchorage to issue a white-label Walmart USD. Paxos could pass that yield to Walmart, and that’s fine. Walmart could independently reward users for having stablecoins. BUT, you couldn’t use the yield from the stablecoin and pass that to the holders. Clear as mud.

🧠 The OCC is asking the market how this rule can be clarified. There’s 60 days for comments. I wonder if the lobbying battle will now move from the CLARITY act to this?

2. The Citrini whitepaper wipes billions from payments company stocks.

A speculative research letter imagining AI agents routing around card networks toward stablecoins wiped billions off V (-4.4%), MA (-8%) and AXP (-8%) on Monday. The market reaction was brutal.

🧠 AI fear is now a tradeable event. Anthropic announced Claude can streamline COBOL, IBM dropped 10%. Cybersecurity stocks tanked after Anthropic released security audits. Citrini landed at exactly the right moment.

🧠 Stablecoins already use cards so this isn’t an “or” it’s an “and”. The second fastest-growing stablecoin payment category is stablecoin-linked cards. Visa cited ~$5bn annualized volume, up from $1bn six months prior (likely helping them outperform peers). Companies like Kast, Dolar, and Altitude let consumers hold stablecoins and spend anywhere. More stablecoin adoption = more card volume, not less.

🧠 Stablecoins improve settlement. They don't replace authorization at 200m+ merchant locations. The letter conflates authorization with settlement. Authorization is the 250ms "can they pay?" check that lets you walk out of the store. No money moves. Settlement is the boring 1-30 day process of actually moving funds.

🧠 Stablecoins are not cheaper; they're better. Previous efforts to compare costs confused network fees with merchant fees. They're only cheaper in some flows, some of the time. Overselling this hurts the cause.

🧠 Agents will use cards AND stablecoins. Agents buying stuff for a consumer will use the rewards card that already works everywhere. Agents buying API keys from other agents will optimise for cost and speed. Today, every AI company accepts cards. No world exists where agents exclusively use stablecoins.

Remember, if you're enjoying this content, please do tell all your fintech friends to check it out and hit the subscribe button :)

Want more? I also run the Tokenized podcast and newsletter.

(1) All content and views expressed here are the authors' personal opinions and do not reflect the views of any of their employers or employees.

(2) All companies or assets mentioned by the author in which the author has a personal and/or financial interest are denoted with a *. None of the above constitutes investment advice, and you should seek independent advice before making any investment decisions.

(3) Any companies mentioned are top of mind and used for illustrative purposes only.

(4) A team of researchers has not rigorously fact-checked this. Please don't take it as gospel—strong opinions weakly held

(5) Citations may be missing, and I’ve done my best to cite, but I will always aim to update and correct the live version where possible. If I cited you and got the referencing wrong, please reach out