Virtual vs Physical Cards: The Real Fulfillment and Cost Tradeoff
A physical card pulls an entire separate PCI security standard into scope. A virtual card sits on the other side of the fraud liability framework instead.

- Physical card issuance brings an entire separate PCI standard into scope: the Card Production and Provisioning Security Requirements, covering manufacturing, personalization, fulfillment, and PIN mailing. A virtual-only program never has to assess any of it.
- That standard's compliance is managed directly by the card networks, not just by PCI itself. A program choosing physical cards has to confirm its own card vendor is actually validated against it, not assume the vendor relationship alone covers the requirement.
- Physical and virtual cards sit on different sides of the fraud-liability framework entirely. Visa Core Rules Section 11.1.2 ties the EMV liability shift specifically to chip-initiated, card-present transactions, while a card-not-present dispute under Dispute Condition 10.4 is invalid only when 3-D Secure authentication was actually used, a protection with nothing to do with a physical chip.
- The real tradeoff is not fulfillment cost against printing savings. It is choosing which fraud-liability framework a program's transactions actually fall under, and which operational burden, card production security diligence or 3-D Secure implementation, the program is actually equipped to carry well.
Virtual versus physical cards comes down to a tradeoff most program teams underestimate on both sides. It is not simply that a physical card costs more to produce and ship. Choosing physical pulls an entirely separate PCI security standard into scope, one a virtual-only program never has to touch. Choosing virtual means relying on a different fraud-liability framework altogether, not just skipping the printing bill.
The security standard that only exists because of the physical card
PCI’s Card Production and Provisioning Security Requirements address an entire category of risk that simply does not exist for a virtual card. The standard covers card manufacturing, chip embedding, data preparation, pre-personalization, card personalization, chip personalization, fulfillment, packaging, storage, mailing, shipping, and PIN printing and mailing. A card that never gets physically manufactured, embossed, or mailed has no material to secure under any of these categories, because none of the activities the standard governs ever actually happen.
This is not an old, retired standard. PCI confirmed in 2022 that the current version, 3.0.1, remains the active requirement set that card vendors are assessed against, with earlier versions formally retired. A program choosing physical cards is choosing to operate, through its card vendor, inside a security framework built specifically around the physical risks of manufacturing, personalizing, and mailing a card. A program choosing virtual-only never enters that framework at all, not because it was exempted, but because the risk the framework addresses simply is not present.
Why this is the program’s problem, not just the vendor’s
It is tempting to treat this as something a card vendor handles entirely on its own, since the vendor is the one actually running the manufacturing line. Compliance with this standard is directly managed by the payment card networks, not generically by PCI, which means a program’s own card network relationship is actually on the hook for confirming its vendor meets the requirement. A program that assumes “our vendor handles PCI stuff” without specifically confirming current Card Production and Provisioning validation is assuming a diligence step nobody has actually completed.
This matters at the moment a program chooses between vendors, not after launch. A vendor quote for physical card fulfillment that does not clearly confirm current certification against this standard is a quote missing a real piece of the actual cost and risk picture. A program comparing two physical card vendors on price alone, without checking this specific validation, is comparing numbers that do not reflect the same underlying risk.
The fraud framework that splits along the same line
The cost tradeoff does not stop at production and fulfillment. Physical and virtual cards sit on genuinely different sides of the fraud-liability framework that determines who absorbs a fraudulent transaction. Visa Core Rules Section 11.1.2 defines the EMV liability shift specifically around chip-initiated, card-present transactions. In the US Region, that shift covers domestic and interregional counterfeit point-of-sale and ATM transactions. A physical chip card, read properly at a chip-enabled terminal, earns this protection automatically. A virtual card, which has no physical chip and is never read at a point-of-sale terminal in that sense, cannot earn this specific protection at all, because the fraud category it addresses does not apply to how a virtual card is used.
A virtual card instead sits inside the card-not-present fraud framework. Dispute Condition 10.4, covering other fraud in a card-absent environment, is invalid specifically when the transaction was processed as a Secure Electronic Commerce Transaction, authenticated through Visa Secure using EMV 3-D Secure, with the resulting authentication value carried into the authorization request. A virtual card program that implements 3-D Secure correctly earns a real, rule-based fraud protection of its own. One that skips this step is not simply missing a nice-to-have feature. It has given up the one structural fraud protection actually available to a card-not-present product, without gaining the chip-based protection that was never available to it in the first place.
What this actually means for program economics
Put together, the real tradeoff is not fulfillment cost against printing savings. It is a choice between two different operational and compliance burdens, each attached to a different fraud-liability framework. A physical card program takes on card production and provisioning diligence, real fulfillment and mailing costs, and the logistics of physical delivery, in exchange for EMV liability shift protection on card-present fraud. A virtual card program avoids all of that production overhead and launches faster, in exchange for taking on the specific discipline of implementing 3-D Secure correctly, since that is the only structural protection available to it.
Neither path is free of risk or cost. A physical program that cuts corners on vendor validation is carrying unassessed card production risk. A virtual program that treats 3-D Secure as optional friction to minimize for conversion is giving up its only real fraud protection to save a few seconds at checkout. The actual economics of each path depend entirely on whether the program is equipped to do the specific work each one demands well, not on which one looks cheaper on a per-card cost sheet.
The operational load that shows up after launch, not before it
The fulfillment comparison most teams run before launch, physical printing and mailing cost against virtual’s near-zero marginal cost, only captures the one-time setup difference. The ongoing operational load that follows launch looks different, and it does not favor virtual automatically.
A physical card program’s post-launch load centers on replacement: a lost or stolen card needs a new one manufactured, personalized, and mailed, which takes days and carries its own per-card cost every time it happens. A virtual card program avoids that cycle entirely, since a compromised virtual card can be deactivated and reissued instantly with no physical object to track down or replace. But a virtual program’s post-launch load shifts somewhere else: every card-not-present dispute that arrives has to be defended with the specific 3-D Secure authentication evidence Dispute Condition 10.4 actually requires, which means the operations team handling disputes needs access to authentication logs and CAVV data, not just a transaction record.
Neither load is inherently lighter. A program that only modeled the upfront fulfillment cost difference, without accounting for which ongoing operational muscle its dispute and support team actually needs to build, is likely to discover the real cost gap well after launch, in exactly the function, disputes and card replacement, that a program’s own cost model should have weighed from the start.
A concrete example: two programs, two different real costs
Picture two expense management platforms launching card programs at the same time. The first chooses physical cards exclusively, because its target customers expect a tangible corporate card. It selects a card vendor on fulfillment pricing alone, without separately confirming the vendor’s current Card Production and Provisioning validation, assuming the program manager relationship already covers it. The second platform launches virtual-only, prioritizing speed to market, and treats 3-D Secure as an optional configuration to revisit after launch because the integration guide marks it as non-mandatory.
The first platform’s real exposure is a card production risk nobody specifically verified, invisible until a card network’s own review asks for the vendor’s current certification status and the program has no clean answer. The second platform’s real exposure is a fraud-liability gap: every card-not-present fraud dispute it faces is fully contestable, because the one protection available to a virtual product was never actually implemented. Both platforms believed they had made a reasonable, cost-conscious choice. Neither had actually priced in the specific risk its chosen path carried.
What to actually calculate before choosing
A program evaluating virtual against physical, or planning to run both, should answer three questions concretely rather than defaulting to intuition about what customers expect. Does the use case genuinely require card-present acceptance, since that is the only scenario where EMV liability shift protection is even relevant to begin with. Can the team realistically confirm and maintain a card vendor’s current Card Production and Provisioning validation, or would the program rather avoid that diligence surface entirely by staying virtual. And does the checkout or payment flow actually implement 3-D Secure correctly and verifiably, since a virtual program that cannot answer this with confidence has not actually secured the protection it is counting on.
Running this calculation accurately, against a specific program’s actual use case and operational capacity rather than a generic assumption about physical versus virtual, is exactly the work behind Midcore’s card-as-a-service consulting engagements. Where the same program sits on top of a broader banking relationship, banking-as-a-service consulting is where that foundational structure gets evaluated alongside the card decision. And for a platform issuing cards specifically to fund disbursements or advances, lending-as-a-service consulting is where the card choice gets weighed against the broader lending program it actually serves.
Frequently Asked Questions
Does choosing virtual-only cards actually avoid PCI compliance work entirely?
It avoids one specific category entirely: the Card Production and Provisioning Security Requirements, which cover manufacturing, personalization, fulfillment, and PIN mailing for physical cards. A virtual-only program never triggers that standard. It does not avoid PCI DSS generally, since cardholder data handling, transaction processing, and the underlying payment infrastructure are still in scope regardless of whether a card is plastic or a number on a screen.
Why does a physical card get a fraud liability protection that a virtual card does not?
Because the protections are built around different fraud categories. Visa Core Rules Section 11.1.2 defines the EMV liability shift for chip-initiated, card-present transactions specifically, which is the fraud pattern a physical chip card actually prevents. A virtual card is inherently card-not-present, so it relies on a different protection, 3-D Secure authentication, which makes a card-not-present fraud dispute under Dispute Condition 10.4 invalid when the transaction carried that authentication. Each protection matches the fraud risk specific to how that card type is actually used.
What should a program actually calculate before choosing virtual, physical, or both?
Whether the program’s actual use case needs card-present acceptance at all, since that determines whether EMV liability shift protection is even relevant. Whether the team launching the program can realistically validate a card vendor against Card Production and Provisioning Security Requirements, or would rather avoid that vendor diligence entirely. And whether the program’s checkout flow can implement 3-D Secure correctly, since a virtual card program that skips this loses its own primary fraud protection without gaining a physical chip’s protection to replace it.
Sources: PCI Security Standards Council, press release on Card Production Security Requirements and PCI Security Standards Council, Card Production and Provisioning bulletin confirming current version 3.0.1, and Visa Core Rules and Visa Product and Service Rules, Sections 11.1.2 and 11.7.5.3 (Dispute Condition 10.4).