What happens when the biggest bank starts thinking like a tech firm? The 3 pillars transforming J.P. Morgan’s banking model

The largest US bank by assets is carving out a differentiated path, defined by structural depth.

With every passing year, J.P. Morgan is taking deliberate steps to rearchitect how its financial infrastructure works — internally, externally, and everywhere in between. It’s doing that through three interlocking shifts across three fronts: AI, embedded finance, and blockchain rails.

These may appear to be separate tracks. But taken together, they point to a coherent and disciplined vision. Through this approach, J.P. Morgan is relocating banking into machine intelligence, into workflows, and onto programmable rails. And that transition is as much about internal control and trust as it is about distribution and speed.

1. AI and Gen AI: The intelligence layer, now being operationalized

In early 2024, J.P. Morgan was still cautiously navigating the advanced AI, generative AI landscape. At the time, Brian Maher, JPM’s Head of Product for AI and ML platforms, characterized the firm’s position as “not yet crawling.”

By now, the bank has transitioned meaningfully into the walk phase, but with its eyes still firmly on the path.

Behind the scenes, J.P. Morgan has spent the last 18 months turning its early AI experiments into foundational capabilities: earnings summarization, advanced internal helpdesk tools, developer enablement, and next-gen risk triage systems. These use cases demonstrate the bank’s larger ambition: to make machine reasoning a native layer in the bank’s internal operations.

More interesting than the tools is the posture. As peers raced to deploy AI in front-office settings, J.P. Morgan leaned into structural questions: 

• How do we supervise these models? 
• How do we align them with regulatory regimes before those regimes fully exist? 
• How do we tie AI to measurable financial outcomes?

CFO Jeremy Barnum outlined this internal accountability framework early on. 

“The firm is not going to be chasing shiny objects in AI. The current focus is on making sure we have a contained, well-chosen list of high-impact use cases and that we’re throwing resources at those in the right way that’s extremely pragmatic and disciplined, and we’re holding ourselves accountable for actual results,” mentioned Barnum early last year. 

That framing remains core to how the bank operates AI today, which isn’t as a product layer, but as a decision logic layer spanning functions and departments.

While many institutions view AI as a channel to customer experience gains or product velocity, J.P. Morgan has always appeared more interested in AI as operational leverage. It’s trying to rewire how work gets done and how decisions get made – at scale, with machine assistance.

J.P. Morgan’s AI strategy has been about internal control, risk calibration, and long-term ROI. That’s likely why the bank’s AI efforts increasingly appear not as a separate function, but as a deeply embedded capability, now more than ever.

2. Embedded Finance: From distribution to design control

While AI represents J.P. Morgan’s inward transformation, embedded finance is how it is projecting outward, without expanding branches left, right, or center or launching consumer-facing apps. The bank is building the invisible banking stack behind platforms and, in the process, changing who gets to own the customer relationship.

When J.P. Morgan Payments integrated its payments suite into Walmart Marketplace earlier this year, it embedded a bank-grade financial nervous system into one of the world’s largest retail platforms, extending beyond basic API connectivity. For sellers on Walmart, this means smoother payouts, capital access, and liquidity management, all without leaving the platform.

But more crucially, it shifts the structural dynamic: the bank is no longer a destination. It’s a layer inside someone else’s interface.

“Traditional channels such as community branches or banker-driven engagement continue to thrive and serve as critical and foundational growth engines for financial institutions. Embedded finance specifically offers banks a unique opportunity to partner with platforms, which they can do in several ways,” Jeff Lin, Head of Industry Product Solutions at J.P. Morgan Payments, defined the shift.

The Walmart deal reflects this thesis in action: not just moving money, but designing infrastructure for high-variance, high-volume ecosystems – from micro-sellers to global brands.

And that’s also where J.P. Morgan’s strategy diverges from a lot of fintechs. Instead of shipping one-size-fits-all tools, the bank is focusing on programmable financial components that platforms can adapt to their own logic: whether it’s instant settlements, cash flow automation, or inventory-based credit offers.

What’s emerging is a new kind of bank-platform partnership. Not transactional. Not vendor-based. But co-designed infrastructure, with the bank acting as a compliance-grade backbone.

In this model, J.P. Morgan is repositioning banking as a form of infrastructure-as-a-service. And the more platforms that adopt it, the more the bank becomes a default layer across industries, without branding itself at all.

In the long term, this enables the bank to scale not just through consumer acquisition but also by embedding itself within marketplaces, ERPs, and vertical SaaS platforms, essentially providing banking without the traditional interface.

3. Blockchain: Rewiring the settlement layer

If AI is intelligence and embedded finance is distribution, then blockchain is J.P. Morgan’s endeavor to recode the pipes of institutional money movement. 

A couple of months ago, the bank launched JPMD, a deposit token, on Base, Coinbase’s Ethereum Layer 2 network. It’s a controlled proof-of-concept, but it makes a strong statement: a legacy institution issuing real bank liabilities on public crypto infrastructure.

Importantly, JPMD isn’t a stablecoin. It’s a tokenized commercial deposit, governed under the same liquidity and risk frameworks as traditional bank deposits. Naveen Mallela, Global Co-Head of Kinexys by J.P. Morgan, emphasized, “JPMD is not a stablecoin; it is a deposit token – a digital representation of a bank deposit that operates on blockchain rails.”

Although still nascent, the move is significant because institutions dealing in tokenized assets, such as money market funds, real-world assets (RWAs), or collateralized lending, have lacked a compliant, on-chain equivalent to cash that aligns with balance sheet requirements. JPMD fills that void.

Choosing Base signals a shift from private blockchains to interoperable, public, composable environments, where institutions, fintechs, and crypto-native players can transact across the same infrastructure layer.

The broader aim is to build a settlement layer for tokenized finance, with round-the-clock liquidity, delivery-versus-payment (DvP) mechanics, and atomic execution. Over time, J.P. Morgan sees JPMD co-existing with its private-chain offering, Kinexys Digital Payments, as part of a flexible toolkit: account-based or token-based, depending on client needs.

But more telling is how J.P. Morgan is shaping the category itself. The firm is not interested in being a crypto bank. The initiative centers on developing an institutional money infrastructure that is blockchain-neutral, aligned with regulatory standards, and built to connect with corporate treasury systems.

Here too, the model is underpinned by a specific mindset: control the infrastructure, let others build on it, and ensure it operates within frameworks trusted by regulators and CFOs alike.