6 min read · 13 sources

Klarna just pulled off the comeback it has been chasing for over a decade. The BNPL firm is now powering Apple’s new US device leasing program, including the $1,999 foldable iPhone Duo and the $1,199 iPhone 18 Pro. That is the same Apple that passed on Klarna back in 2013 - and now Klarna is the rails under Apple’s push to turn hardware into a subscription.

The deal gives Klarna massive distribution, but the bigger story is what it signals: Apple is commoditizing its own hardware margins for predictable revenue. For engineers, this means Klarna is now running leasing logic - upgrades, buyouts, returns - at Apple scale, which is a very different operational beast than plain BNPL checkout. If the leasing program takes off, expect Klarna’s infrastructure to be the template every other device maker copies.

Klarna is financing Apple’s $1,999 foldable iPhone Duo on 24-month terms, a distribution win a decade after Apple first rejected it.

Robinhood Gives Stock Tokens Actual Teeth After AMC's "Vile" Backlash

Source: unchainedcrypto.com ↗

Robinhood is adding one-for-one in-kind redemptions and voting rights to its Stock Tokens, per Unchained Crypto. The move directly answers AMC CEO Adam Aron, who called the tokens “vile” because they carried economic exposure without actual shareholder rights. Coinbase and other tokenization platforms are watching closely as Robinhood builds out redemption and governance mechanics that more closely mirror traditional equity.

This is the hard part of tokenized equities: the legal and technical design of what happens when you want out, and whether you get a vote. Robinhood’s one-to-one redemption means the token is now backed by something real, not just a promise. For anyone running settlement systems, this is the moment tokenized equities stop being a workaround and start becoming a parallel market with real governance overhead.

Nasdaq's $100M Bet on Blockchain Settlement

Source: lex.substack.com ↗

Nasdaq has invested $100 million into a blockchain-based equity settlement play, per Fintech Blueprint’s Lex Sokolin. The thesis is simple: the current T+2 settlement cycle is a relic, and blockchain-based settlement can compress it toward real-time. Nasdaq is betting that the infrastructure shift is coming, and it wants to own the rails rather than be disrupted by them.

For engineers, this means the plumbing of the public markets is finally getting a backend rewrite. The shift toward blockchain-based equity settlement has huge implications for how clearing houses, custodians, and brokerages reconcile positions. If Nasdaq gets this right, the traditional DTC model - which still runs on batch processes from the 1970s - starts looking very expensive.

American Express Builds a Small-Business Banking Stack

Source: threadreaderapp.com ↗

American Express is building out a full small-business banking stack, according to a ThreadReader thread. The company is grouping its Business Checking and Business Savings accounts under a single operating account for small businesses, rather than treating them as separate products. Jamie Dimon’s shareholder letter flags data access and third-party rails as the key battlegrounds here.

Amex is moving from being a card issuer to a primary bank account, which is a much stickier relationship. The competitive dynamics in payments infrastructure are shifting: whoever owns the operating account owns the customer’s cash position, and that is a very hard relationship to leave. For engineers, this means Amex is now building the kind of ledger, reconciliation, and data-access infrastructure that Stripe and Mercury have been running on for years.

Mastercard and Trip.com Go Agentic in Travel

Source: finextra.com ↗

Mastercard and Trip.com are targeting agentic commerce in the travel sector, per Finextra. The idea is that AI agents will book travel on your behalf, which means payments infrastructure has to handle agent-initiated transactions, authentication, and disputes without a human at the keyboard. That is a very different threat model than card-present or card-not-present.

For engineers, agentic commerce raises a question the industry has not answered yet: how do you authenticate an AI agent, and who is liable when it books the wrong flight? Mastercard is trying to be the layer that answers those questions before someone else does. Trip.com gives it a distribution channel with real volume.

Tabby's $6.5B Valuation and the BNPL Endgame

Source: reuters.com ↗

Saudi Arabia’s Tabby raised $233 million in a Series F led by Blue Pool Capital, per Reuters. The round values the BNPL fintech at $6.5 billion, up from $4.5 billion a year ago. Tabby is processing more than $18 billion in annualized volume and expanding beyond BNPL into larger consumer financing, working capital, and cash-account products.

The interesting part is the expansion: BNPL was the wedge, but Tabby is moving upmarket into what looks like a full consumer banking stack. That is the same playbook Klarna ran in Europe, and it works when you have the volume and the data. For engineers, the takeaway is that BNPL is no longer a payments product - it is a customer acquisition cost for a balance sheet.

Radical Ventures' $1B AI Fund

Source: betakit.com ↗

Radical Ventures launched Canada’s largest AI fund with a first close of over $1 billion, per BetaKit. The Radical Breakouts Fund targets late-stage AI companies in Canada and globally, a push to give Canadian AI scaleups a path to stay home instead of moving to the US.

The fund’s size matters because AI is capital-hungry at the scaleup stage, and Canada has historically lost its best AI companies to US investors who can write bigger checks. Whether this moves the needle depends on execution, but the first close number is real.

Grab Buys Atome for $1.5B

Source: wsj.com ↗

Grab is acquiring a majority stake in SoftBank-backed Atome for about $1.5 billion, per the Wall Street Journal. Atome is a BNPL player in Southeast Asia, and the deal gives Grab a bigger share of the region’s consumer credit market. It is a consolidation play in a market where the cost of capital is forcing smaller players to sell.

Wells Fargo Brings International Transfers to Mobile

Source: finextra.com ↗

Wells Fargo added international money transfers to its mobile app, per Finextra. It is a table-stakes feature for a major bank, but the fact that it took this long shows how legacy infrastructure slows down even the obvious stuff.

The Barbell: Why AI Kills the Middle of Fintech

Source: primary.vc ↗

A Primary VC piece on enterprise AI argues AI will erode traditional software moats by making products cheaper to replicate and migrations easier, pushing the industry toward a barbell structure. At one end, a small number of dominant platforms own entire buying centers; at the other, tiny niche software businesses flourish. The middle - mid-sized point solutions - gets squeezed out.

A companion piece from Matt Verna applies Ben Thompson’s newspaper barbell analogy to software: top players like The New York Times grew from 1.1M to 13.4M subscribers while regional papers like the Dallas Morning News dropped 88%. Amazon built its moat through 7,500+ days of relentless execution, not initial defensibility. The lesson for fintech engineers: moats come from compounding execution, switching costs, and network effects - not just technical barriers.

The AI Accounting Stack Is Coming

Source: threadreaderapp.com ↗

A ThreadReader thread on AI accounting highlights two products attacking the space. One lets users ask plain-language questions about spending changes and cash position without spreadsheets, with support for bringing your own bookkeeper or partners like Pilot, Fondo, and Ridgewood. The other is Stack, an AI operating system for accounting firms that learns a firm’s processes, runs the close, and posts journals - with design partners closing books in half the time.

The close is the last bastion of manual accounting work, and these tools are coming for it. For engineers, the interesting problem is trust: an AI that posts journals can also post garbage journals, and the accountability model for that is still being written.