A quick reset before we dive in
We have decided to change the format of News of the Week.
Mastercard Just Built a Payments AI Brain — And It’s Not an LLM
Mastercard dropped a pretty significant announcement yesterday: it’s building a proprietary generative AI foundation model trained not on text, but on billions of anonymized payment transactions — a Large Tabular Model (LTM), not an LLM — with plans to scale to hundreds of billions of transactions using Nvidia and Databricks infrastructure. Here’s the thing — most banks and networks are just bolting ChatGPT wrappers onto their existing pipelines; Mastercard is going a completely different direction, training a domain-specific model that learns to predict transaction characteristics the same way an LLM predicts the next word in a sentence. The first use case is cybersecurity, where early testing already showed the LTM outperforming standard ML techniques — specifically on high-value, low-frequency purchases like wedding rings that trigger false positives constantly. Translation: a model that understands commerce at the structural level, not just the text level. That’s a different game entirely.
Sources: Mastercard
Veriff: The Real Identity Threat Goes Beyond Deepfakes
Veriff’s CTO Hubert Behaghel made a pointed argument in PYMNTS’ “What’s Next in Payments” series yesterday that the entire industry is misdiagnosing AI fraud risk — focusing obsessively on deepfakes while ignoring that deepfakes only attack one layer (computer vision) of a much deeper identity problem. According to Veriff’s 2026 Identity Fraud Report, roughly 1-in-25 online identity verification attempts now involves fraud. The real vulnerability, per Behaghel, is siloed identity architectures that check each signal independently rather than cross-correlating thousands of data points — device intelligence, network behavior, media inputs — together. And here’s the kicker he dropped: KYC data providers themselves are becoming high-value attack targets, because fraudsters know that owning a verification provider’s dataset unlocks a master key to synthetic identity fraud at scale. Basically, the identity infrastructure we built for humans is not ready for the age of AI agents making purchases autonomously — and most banks haven’t started building the accountability chain that agentic commerce will require.
Sources: PYMNTS
Bank of America’s “Build Once” AI Strategy: Discipline or Moat?
Fortune ran a detailed inside look at BofA’s AI architecture yesterday, and the headline is this: while the rest of the industry was running 17 parallel vendor pilots and rebuilding the same LLM plumbing in every division, BofA CTO Hari Gopalkrishnan has been running a “build once, reuse forever” model since 2018 — and just rebuilt it for the gen AI era in 2025. The bank put that discipline to the test, running 15 separate proofs of concept to see whether letting individual units “go rogue” with their own AI stacks would outperform the shared Erica platform — and found they were mostly just reinventing the same infrastructure. The shared platform now supports 60,000 tech staff across consumer, wealth, capital markets, and private banking, all on the same model-agnostic architecture. The tradeoff, as Gopalkrishnan admits, is that you need “institutional patience” — which is a polite way of saying the bank has to say no to a lot of vendors with flashy demos on a regular basis. Whether that plays out as a competitive advantage or an innovation bottleneck is the multi-billion dollar question.
Sources: Fortune
The Fintech Bubble Keeps Growing—Fueled by AI Hype
Forbes came out swinging yesterday with a data-heavy takedown of private fintech valuations — and the numbers are wild. Stripe, which brought in $6.9 billion in net revenue in 2025 (+30% YoY) and processed $1.9 trillion in payments, is currently valued at $159 billion in private markets — nearly 5x its closest public competitor Adyen, which processed similar volumes. Ramp claimed $1 billion in “gross revenue” and fetched a $32 billion valuation — but that gross figure doesn’t subtract the interchange fees and rewards returned to banks and customers, meaning actual net revenue is likely 40%+ lower and the real multiple is closer to 50x, “reminiscent of fintech’s 2021 bubble days” per Forbes. Meanwhile, the top-10 private fintechs’ collective market value jumped 164% over the past 12 months, versus a 2% rise for the top-10 public fintechs. Translation: private fintech is a luxury-goods market right now, where prestige and scarcity are driving prices in ways that have almost nothing to do with underlying economics — and the AI narrative is doing the heavy lifting. The reckoning, when it comes, will be cinematic.
Sources: Forbes
Capgemini: 80% of Bank Execs See No AI Revenue Gains
A Capgemini Research Institute report released Monday — based on 150 senior leaders at corporate and investment banks plus 600 non-bank financial execs — landed with a thud: more than 4-in-5 CIB executives said new product implementations aren’t boosting revenue, and 51% admitted new products didn’t deliver expected cost savings either. Here’s the structural problem: only 29% of annual IT budgets are allocated to transformative technologies, while 43% goes to maintaining legacy systems — which is basically the definition of running to stand still. The customer demand mismatch makes it worse: more than half of CIB customers want real-time responsiveness, 49% want personalized engagement, and 40% want innovative products — but fewer than 20% say banks are actually meeting those expectations. Revenue growth is projected to decelerate from 6.5% (2022–2024) to 5.4% over the next five years as a result. Basically, banks are spending billions on AI while their foundational infrastructure quietly cannibalizes the return. The pilots look great in the boardroom deck; the P&L doesn’t agree.
Sources: Yahoo Finance
Don’t blink—the banking Singularity is accelerating.
If you’re finding these breakdowns valuable, give it a Like and Subscribe so we can keep the real conversations going.
Like this week’s highlights?
Don’t miss the next one.
Ready to Explore?
Subscribe to our newsletter for exclusive insights, transformation strategies, and the latest banking technology updates.
