Today feels like fintech’s version of Fast & Furious: the plot doesn’t matter, the engine upgrades do. SoFi Technologies (SOFI) is stacking conviction with an insider buy and pushing stablecoin settlement into the mainstream rails. Affirm (AFRM) is teaming up with Stripe (private) so AI agents can “buy” without turning your card details into confetti. And LendingTree (TREE) missed EPS by a mile… then the stock moonwalked up anyway because markets love a good “but actually…” moment.
🚀 SoFi (SOFI) is basically the meme of “one guy doing everything at once”
SoFi CEO Anthony Noto just bought 56,000 shares in the open market for about $1.0M (weighted avg around $17.88). Translation: “I’m not just tweeting confidence, I’m wiring it.”
Then SoFi dropped the bigger, more structural headline: it’s expanding its partnership with Mastercard so SoFiUSD (SoFi’s fully reserved USD stablecoin) can be used as a settlement option across Mastercard’s global payments network, including SoFi Bank potentially settling its Mastercard-powered debit/credit transactions in SoFiUSD. Galileo (SoFi’s tech platform) is also expected to offer client banks the option to settle card transactions using SoFiUSD.
What’s the “why this matters” beyond the crypto vibes? Settlement is the boring back-office part of payments… and that’s where the money-speed advantage lives. If SoFiUSD becomes a legit “settlement currency” in more flows (cross-border, B2B money movement, payouts), you’re basically watching SoFi try to become a bank-grade, regulated onchain utility layer—not just a slick app with a high-yield savings tab. Mastercard also name-dropped its Multi-Token Network (MTN) as a place SoFiUSD is expected to be supported, which is a fancy way of saying: “we want stablecoins, tokenized deposits, and fiat to interoperate without everyone having a panic attack.”
Takeaway: SOFI is trying to turn “stablecoin” into “settlement feature,” and Noto buying shares is the confidence soundtrack.
🤖 Affirm (AFRM) walks into an AI checkout and Stripe quietly hands them the keys
The Affirm + Stripe headline sounds like sci-fi: Shared Payment Tokens that let AI initiate purchases for shoppers. But under the hood it’s more like: “make agentic commerce not insanely unsafe.”
Here’s the clean mental model:
In an AI shopping flow, the agent (think: your ChatGPT-but-for-buying-stuff) needs a way to pay without getting your raw card number or turning “card-on-file” into a free-for-all.
Stripe’s Shared Payment Tokens (SPTs) are designed so an agent can pass a payment credential without exposing the underlying credentials, and tokens can be scoped, time-limited, and amount-limited.
Stripe says SPTs are already used by merchants like Etsy and URBN brands, and now they’re expanding to support additional payment methods—including BNPL via Affirm—because sellers want more than “just cards” in agent-led checkouts.
With Affirm in the mix, the flow still includes Affirm’s real-time eligibility + plan selection, but it happens inside the AI experience while Stripe processes the payment in a way that doesn’t leak sensitive credentials.
Why would Affirm care? Because if AI shopping takes off and everything defaults to Visa/Mastercard card-on-file, BNPL risks getting “algorithm’d out” of checkout. This partnership is basically Affirm saying: “If the future is agents, we’re not letting the agent pay with someone else’s default button.”
Also: LendingTree is literally talking about AI reshaping product discovery and arguing why marketplaces won’t be easily disintermediated—so this is a nice live-action example of where the battle is actually happening: who owns discovery + checkout in an agent world.
Takeaway: AFRM + Stripe are building the tokenized permission slip that makes AI shopping possible without handing the agent your wallet.
🎢 LendingTree (TREE) misses EPS and still gets rewarded like it hit a buzzer beater
Ok, the LendingTree move looks unhinged until you remember what the market is actually pricing: the forward story.
Yes, LendingTree’s Q4 EPS print was ugly (-$0.39 vs +$0.71 expected), which is the kind of miss that usually gets you sent to the shadow realm.
But revenue was strong: $319.7M (about +22% YoY in the shareholder letter) and ahead of consensus.
And more importantly, the company guided Q1 revenue $317M–$325M and FY 2026 revenue $1.275B–$1.33B, both above the consensus numbers reported in the earnings coverage.
The shareholder letter also spotlights improving operating performance metrics like VMM ($92.0M, +6% YoY) and Adjusted EBITDA ($36.7M, +14% YoY), plus a very explicit “we’re using AI to drive efficiency and conversions” narrative.
So how do you get +~40% (ish) behavior on a day when EPS looks like it fell down the stairs? Easy:
Expectations were already in the basement,
Guidance + revenue momentum screamed “the core engine is fine,” and
The market loves a turnaround arc more than it loves GAAP purity.
Takeaway: TREE ripped because the forward guide and operating momentum outweighed the EPS faceplant.
Recap
SOFI is pushing stablecoin settlement into real payments rails, AFRM is making sure AI checkout doesn’t default to “cards only,” and TREE proved guidance can be the ultimate plot armor.
If this stack hit, subscribe to Fintech Stacks and send it to one friend who still thinks “stablecoin settlement” is a Kendrick lyric.
🎙 Podcast: Fintech Stacks Daily — 15-minute breakdowns, zero fluff.
📺 YouTube: Fintech Stacks — narrative-driven deep dives on the money internet.
Disclaimer
This content is for information and entertainment only and is not investment advice. I may or may not hold positions in some of the companies mentioned. Assume I at least own a fintech hoodie and a bunch of debit cards.