In 1996 the Palm Pilot felt like the future. For those unfamiliar (or not yet born), the Palm Pilot was a pocket computer that organized contacts, calendars and tasks in a way that was fast, simple and portable. Developers created apps allowing users access to the internet. It wasn’t the first Personal Digital Assistant (PDA), but it was the first that worked for the mainstream.
Or those clever enough to master the Graffiti script that made it work. I never really did. HotSync made a handful of apps that were as real-time as real time got in 1996.
For a few years, the Palm Pilot was indispensable. The initial release sold one million copies in 18 months. Hip and happening executives carried it everywhere. Developers built more and apps. Amazon was one of its early commerce touchpoints. It defined a moment.
That, at the time, seemed to define the future.
Then along came Blackberry with a real keyboard that made it easier to use and more attractive to more users, though not so much to developers. Then the iPhone after that ditched keyboards for a touchscreen, an App Store and lots of apps. In 2009 Uber wrote its app for iOS, a milestone that got most hardcore Blackberry users (like me) to finally make the switch. Blackberry slowly bit the dust.
The Palm Pilot, in just 11 short years, became a historical artifact. Or maybe, more appropriately, the bridge to a different way of accessing real-time content, finding and transacting with businesses, and communicating in new ways. Not the be-all-and-end-all, but an important stop along the way.
It’s an outcome that even Palm Pilot’s developer knew was likely. The smartphone as we know today, he remarked, was his real vision and intention.
As shopping moves from online and apps to agents and agentic commerce, digital wallets could become the Palm Pilot of payments.
Not the endpoint, but an important bridge to connect commerce to secure credentials in an always-on, always-connected world.
A waypoint along the journey to a new era that will redefine the shopping and payment experience for consumers and businesses.
As AI and agents deliver a sea change in how consumers search for and find what they want to buy. And most importantly, how they want to pay.
The Digital Wallet’s Identity Crisis
Digital wallets earned their stripes by solving for fear and friction as eCommerce scaled. Early on, consumers were reluctant to type card numbers into unfamiliar sites. Digital wallets also accelerated payment to sellers. PayPal’s most notable achievement when it was launched in 1998 was eliminating the need for eBay buyers and sellers to wait, quite literally, for the check to clear before a transaction could settle and goods would ship.
Wallets became the “buy button” that sat between the consumer, her credentials and the merchant. They made checkout feel safer. PayPal more or less cornered the market for the better half of two-plus decades.
As the web matured, the wallet value proposition shifted from safety to speed as wallet providers sought merchant acceptance. PYMNTS Intelligence measured checkout conversion for 5 years, from 2016 to 2020. We learned that buy buttons improved conversion by reducing tens of keystrokes on the way to checkout and payment to a few, even one. More consumers who wanted to pay that way caused more merchants to want to add them to their checkout pages.
Wallet issuers used higher conversion as their raison d’être. Soon, checkout pages became Nascar-ized with buy buttons galore.
Yet, even today, some 27 years after the launch of PayPal, and 11 after the launch of Apple Pay, wallets remain a small fraction of volume on merchant sites and apps. Even less in the store. Merchant acceptance took longer and cards on file and guest checkout got easier. Outside of Amazon’s massive, closed-loop ecosystem, where stored credentials are king and queen, wallets still represent a small fraction of digital transactions.
According to PYMNTS Intelligence, in May 2025, 9.6% of consumers used wallets to pay online. However, they used debit four times more frequently (36%) and they used credit 3.3 times more frequently (32%).
When considering that Amazon accounts for around half of online retail sales, and stored credentials are the method of payment on that marketplace, the total addressable market for wallets in 2025 and projected in 2026 and beyond is roughly $78 billion. Sort of a drop in the bucket, all things considered, based on current projections of wallet usage and adoption online.
Today, digital wallets are still a thing because online checkout has remained largely unchanged, just snazzier and faster, for nearly three decades. When traditional checkout disappears, the wallet’s center of gravity will go with it.
The No-Checkout Checkout
And that’s the big headline and digital shopping paradigm shift: the rise of agents and Zero Checkout.
Here’s what that may look like.
In a world where agents shop and pay, they will not fill out forms. Consumers will not go to a checkout page. Agents will interpret prompts, orchestrate product and payments options, and execute end to end.
The consumer will specify outcomes and constraints. Price points and delivery windows. Preference for sustainable materials. Sizes and colors. Favorite brands. Desire to redeem points or apply store or loyalty credits. Options for shorter- and longer-term payback periods or installments at certain thresholds.
The agent will translate those instructions into a sequence of machine calls across sellers, apply loyalty and financing logic, and return a confirmed order with a short explanation of the tradeoffs and the decision that drove the outcome.
In that world, the consumer prompt replaces the not-that-always-relevant guesswork that wallets once called personalization.
The consumer and her instructions are the personalization engine. Her wish is the agent’s command.
That makes the big decision in the agentic commerce world how to pay. Not where to pay at checkout on a page. One-Click will become Zero-Click.
Over time the consumer’s agent will get smarter and learn because it will see how the consumer engages across all the businesses they shop and the payments options that are most appealing. The consumer will gain more trust. The agent will learn when to conserve cash and pay later, when to redeem rewards instead of earning more, when to split a payment, and when to lean on a specific card to trigger a reward or cashback tier. The action moves beneath the surface, and it will happen in milliseconds.
Like the Palm Pilot in 1996 as the first PDA, digital wallets introduced people to a better and more secure way of paying for things as eCommerce took hold.
But in a world where agents are the deciders, the infrastructure that matters most is the quality and security of the credentials they can use.
That is where networks and issuers may already have a decisive lead.
Tokenized Credentials and the Digital Wallet Watershed
Over the last decade Visa, Mastercard, Amex and issuing banks built an invisible tokenization layer that spans merchants, wallets, subscriptions and account-on-file ecosystems. Today, the Visa network has issued 13.7 billion tokens which accounts for nearly 50% of eCommerce transactions globally. Meanwhile Mastercard says that more than 30% of their transactions are tokenized.
Tokens are not static substitutes for plastic. They are dynamic, updating when cards are replaced, adaptable to security and rewards, and embedded into recurring payments and installments.
Networks and issuers are layering intelligence on top of tokens so the right credential can be chosen for each transaction, with the right terms at the right moment. Flexible or dynamic credentials are a single account that simplifies payment options from multiple cards to one, with multiple ways to settle that transaction. Buy and pay now. Buy and pay a little later. Buy and pay a lot later.
That is the foundation for agentic commerce. A pool of smart, tokenized accounts that agents can use to deliver the consumer’s outcome each time they make a purchase.
In lockstep is the effort to make credentials carry more than payment authority.
A smarter credential can transport verified identity, risk/reward preferences, and attributes that improve acceptance and the shopping experience: age or business status for restricted goods, address trust and delivery preferences, dispute history and claims behavior, verified KYB for B2B transactions.
When identity travels with the credential in a privacy-preserving way, agents encounter fewer roadblocks and sellers accept them with greater confidence. The value migrates from the container called a wallet with multiple options to a single credential that can pay, authenticate and present the right payment options at the right time.
Enter Google and PayPal
Google and PayPal’s multiyear partnership, announced last week, reflects the scramble happening across many platforms as each tries to jump on the agentic commerce bandwagon. PayPal’s infrastructure will be embedded deeper into Google’s products. Google Cloud will become PayPal’s AI backbone. Together they say they’ll “revolutionize” agent-driven shopping. So they say.
It will be one of several experiments taking place with agentic commerce. In truth, it is very early days. Perplexity partnered with PayPal for commerce in AI search. OpenAI is working with Shopify to connect prompts to merchant catalogs. firmly connects Perplexity to millions of merchants for app checkout.
PayPal’s approach centers on bringing wallets and accounts to AI models. The theory: Consumers interact with agents through payment platforms, making wallets the bridge between intent and transaction.
The distinction matters because it shows how the wallet-centric worldview is already being challenged by infrastructure that doesn’t need wallets at all. And how those with a vested interest in them are already doubling down.
The Account Base Illusion
The Google/PayPal partnership logic assumes, in part, that PayPal’s hundreds of millions of accounts matter in agentic commerce. You know what they say when you assume. Consumers don’t start shopping on payment sites. They start with discovery. Amazon became a commerce force by capturing that beginning and making discovery and checkout one-click without leaving that ecosystem. Today more than half (54%) of product searches start on Amazon.
Google sees a come-from-behind opportunity. It has billions of users across Gmail, YouTube, Chrome and Android. If Google can capture product searches before they migrate elsewhere, Google believes it can finally monetize commerce at scale.
But they may be betting on the wrong horse.
PayPal logins don’t create intent. What matters is where consumers begin their journey. Increasingly, PYMNTS Intelligence finds that conversational AI platforms can orchestrate complex transactions that go beyond keyword searches without routing through wallet infrastructure at all.
In an agentic world, the advantage goes to whoever owns the intent signal. Google’s challenge isn’t building a wallet. It’s defending the starting point before Gen AI-native platforms pull it away.
The Infrastructure That Actually Matters
While platforms experiment with partnerships, networks and issuers have built something more fundamental over the last decade: intelligent credentials that make payment decisions autonomously. These tokenized accounts don’t just substitute for plastic cards. They update dynamically and embed into any system that needs payment capabilities.
When a consumer prompts an AI about what to buy, personalization happens in the prompt and merchant response. When that consumer needs to pay, the decision shifts to how to pay. Which credential maximizes rewards? Which card hits a spending threshold? Which account optimizes cash flow?
Issuers are trying their hand. BNPL providers provide a single credential that offers multiple ways for consumers to settle transactions that are also linked to offers and promotional financing. Global technology providers enable connectivity between existing debit cards and those flexible credentials. Credit card issuers are adapting existing credit lines into shorter- and longer-term installments. Others are turning debit into credit options to offer payment flexibility using a single card credential. Each have their own incentives to explore options outside of digital wallets, even as they all provision their cards to make them work inside of them.
In all cases, the aim is to build the foundation for credentials smart enough to make those decisions without human intervention and without toggling between multiple card options. Payment choice becomes automated based on rules the consumer sets once, not selections they have to make and think about at checkout every time.
This is the foundation for agentic commerce. Agents that don’t need wallets when tokenized credentials offer the same payment optionality without the acceptance hassle.
Playing the Short Game
For PayPal, partnerships like Google read as a bid to stay relevant. PayPal was the default when merchant checkout felt risky. That advantage eroded as Apple Pay, Amazon one-click, Shopify with Shop Pay, and embedded payments scaled. PayPal’s online checkout share has eroded over the last five years from 9.9% in 2021 to its May 2025 value of 7.2%.
The wallet market got crowded while networks and issuers invested billions in tokenization that made them indispensable. At the physical point of sale, cards still dominate, even as just about everyone with a smartphone has a wallet to use if they wanted to. And just about every merchant has a point-of-sale device capable of enabling a wallet transaction.
Google offers PayPal distribution. PayPal gives Google payment capabilities for commerce experiments and probably a pretty good deal on payment processing. Both can buy time to understand how agent-driven transactions develop. But they’re building on the assumption that wallets can remain central when agents shop.
That assumption looks increasingly outdated.
Google’s Real Game: Monetizing Agentic Commerce
Here’s what’s actually interesting about Google. It’s not clear that Google wants to be in the wallet business, despite more than a decade of attempts to create one that people use. Its DNA is advertising. In Q2 2025, Alphabet reported $96.5 billion in ad revenue. That was about 75% of its revenue, with Cloud representing most of the rest. Commerce has always been Google’s weak link. Consumers may search for what products to buy on Google, but they buy on Amazon.
That gap has haunted Google for two decades. It started Google Checkout in 2006, and then discontinued it in 2013. It rebranded Google Wallet to Google Pay in 2018 and then back again in 2022. Funny enough, it eliminated the ability for PayPal wallets to be connected to Google Wallet in June of 2025. But none of these efforts dented Amazon’s dominance as the product discovery/buy search alternative. Google captures the query, then hands consumers off to complete purchases elsewhere.
But Google’s real advantage is intent. Billions of times daily, people tell Google what they want, when and where. That’s why their ad business is so profitable. Google monetizes intent by essentially auctioning access to it.
With agents, that moment of intent becomes transactional.
With agentic commerce, Google believes it finally has a way to close that gap. When consumers tell AI systems what to buy, that intent signal can trigger orchestrated transactions across merchants, payment methods and fulfillment. Google doesn’t need to have a wallet, it has stored GPay credentials in its browser, or connects to PayPal’s, to finally monetize commerce at scale. They do need a great agent and payment agent protocol. Something they also announced last week.
The Faster Disruption
Digital wallets spent nearly three decades trying to scale. The problem isn’t lack of user access; it is a lack of use and adoption.
Now wallets are being challenged by something moving much faster and, seemingly, of much greater interest to consumers: AI agents that personalize through prompts and execute through intelligent credentials.
Consumers become the DIY personalization engines by stating exactly what they do and do not want, in a conversational style with the agent, until they perfect the prompt. Agents become smart sales assistants that deliver those outcomes. Smart flexible credentials become the payment foundation.
The Google-PayPal partnership, like similar deals, reads more like experimentation than strategy. Useful for learning, probably not the one big thing that will drive change. In agentic commerce, the critical shift is from what to buy to how to pay. The real differentiation comes in payment optimization. The credential that maximizes rewards, preserves cash flow, leverages installments most effectively, redeems points at the right moment.
The future of payments in an agentic world isn’t a faster button or smarter wallet. It’s no button at all. Zero Checkout. Intelligent credentials making optimal payment decisions invisible to consumers. Agents shopping on their behalf and executing exactly what the consumer wanted without them having to think about it.
Digital wallets, like the Palm Pilot, proved the concept. They may not survive the transition to what comes next. And it might not take 11 years.
