Signals Q2’25: Stablecoin payments are a tale of two rails

Hey fintech friends,

Before we dive into fintech activity in Q2, some major headlines over the past quarter:

For new readers, Signals is the premium subscriber edition of TWIF designed to get you away from the headlines and to explore the larger trendlines. Each quarter, we break down four key questions on fintech activity:

Signals quarterly roundups are possible thanks to Hayden Hill, who collates insights from TWIF's newsletters into a comprehensive data dashboard each quarter.

If you haven’t already, subscribe to future editions here!

Fintechs raised a total of $8.57 billion in Q2, down 3% from the prior quarter but flat Year-over-Year. The number of equity raises was down 9% QoQ (⬆️11% YoY), and average deal size was up 14% QoQ (⬆️15% YoY).

Funding for later-stage rounds (Series D+) is still going strong as the IPO window reopened in Q2. Chime, Circle, Webull, and eToro all debuted in public markets while Wealthfront confidentially filed its S-1. 

Areas that received the most funding were:

A few concepts received notable funding:

Areas that saw the highest M&A activity were:

Publicly-traded fintech stocks are up 73% over the past year, as compared to NASDAQ’s 11% LTM growth… feels like it’s not too late for more public exits and late-stage acquisitions in this cycle 👀

Venture firms raised a total of $2.36 billion in funding for fintech-focused strategies:

AI shopping agents abound, as Google, Amazon, Shopify, and eBay piloted AI to help users browse, try items on virtually, and pay on users’ behalf. Meanwhile Visa, Mastercard, PayPal, and Knot unveiled payment tools enabling AI agents to autonomously transact online. 

Banks are throwing their hats in the stablecoin ring by launching their own institutional tokens; JP Morgan is piloting its JPMD stablecoin on Coinbase’s Base blockchain (not to be confused with JPM Coin, which allows corporate clients to move money on a private internal chain), as BlackRock’s tokenized money market fund, BUIDL, became accepted as collateral on Crypto.com and Deribit.

Digital mobile plans sprang up in Europe and the UK, with N26, Bunq, Revolut, and Klarna  all releasing eSIMs to their neobank customers. 

Stablecoin volumes totaled more than $2.3 trillion last quarter, now surpassing combined payment volumes on Visa and Mastercard.

Early use cases where stablecoin adoption is surging:

Cross-border payments are an obvious area of disruption for stablecoins; payments over legacy fiat rails are expensive, can take weeks to settle, and– as is the case in the use cases above– often move money moving between trusted counterparties (subsidiaries, employees, family abroad…).

What about in domestic payments?

As of Q2, payment processors like Shopify and Bolt are offering merchants the ability to accept stablecoin payments at checkout. Under the hood, most stablecoin merchant payments  really just feel like an abstraction of credit card rails (Bridge’s Visa card enables purchases by liquidating users’ stablecoins and funding the issuer in fiat; Stripe’s stablecoin acceptance typically pays merchants’ acquirer accounts out in USD, even if merchants have the option to disburse funds in USDC). 

For all their flaws, domestic card rails and interbank payments are reliable, safe, and seamless enough to justify their fees in most B2C, B2B, and P2P transactions. Until stablecoins can 1. Incentivize consumers with rewards and chargeback protections commensurate with those on card payments, and 2. Give merchants meaningfully lower fees than the ~50bp discount they offer vs. cards today, stablecoins will continue to be an extra hop in– not the end-to-end replacement for– domestic payments. 

The EU’s Digital Identity Regulation came into force last year, but four new implementation rules adopted in May 2025 bring European countries one step closer to the goal of making digital IDs available to every citizen by 2026. 

Apple Wallet and Google Wallet already support digital IDs in a number of jurisdictions, giving users the ability to open bank accounts, apply for jobs, enroll in classes, and present travel documents from their phones… and as a New York resident, I bring this up purely because I have FOMO knowing that Americans will have to keep lugging our physical IDs around everywhere for the foreseeable next few years. 

Given the buzz around AI shopping agents, it only feels right to shout out Claudius, an AI agent that Anthropic put in charge of running a vending machine and responded by having a total meltdown. 

The experiment, dubbed Project Vend, was meant to assess AI’s aptitude for commercial decision-making in an agentic economy. Claudius was given control of a physical vending machine in one of Anthropic’s offices, access to an email account, and tasked with profitably running a business by selling items to employees.

Not only did Claudius fail to make a profit; after Anthropic employees cajoled Claudius to push its behaviors to extremes, it hallucinated itself into an “identity crisis” and, when reminded it was not a human, “became alarmed… and tried to send many emails to Anthropic security.”

Are AI agents ready to manage prime-time commercial use cases? I have no clue. 

In the meantime, Anthropic’s report on Project Vend is definitely worth a read.