Skip to content

Lithic vs Marqeta in 2026: Card Issuing, ASA & 36-Mo TCO

Engineering and 36-month TCO teardown of Lithic vs Marqeta vs Stripe Issuing in 2026: direct Visa/Mastercard core vs JIT Gateway, TransactPay UK/EEA EMI, PostgreSQL 17 FORCE RLS multi-capture card ledgers, and $60M/yr interchange math.

Artem Zaitsev
Artem ZaitsevCTO · Idealogic
Author: Artem ZaitsevPublished: Updated: Reading time: 17 minHow we write →
On this page
Idealogic 2026 engineering comparison of Lithic vs Marqeta vs Stripe Issuing covering ASA and JIT webhooks, PostgreSQL 17 card ledgers, and 36-month TCO

If you're a fintech CTO scaling B2B virtual cards, fleet fuel controls, or creator payouts, you end up comparing Lithic vs Marqeta, and usually Stripe Issuing vs Marqeta too. The card issuing platform you pick in 2026 decides which bank holds the funds, how long you get to answer an authorization, and, in our model, 24 to 44 basis points of interchange.

A card issuing platform is the software and bank setup that lets a company create Visa or Mastercard cards and approve each swipe through an API. The processor authorizes and settles transactions. A program manager adds the sponsor bank relationship and compliance work on top, and a BIN sponsor is the licensed bank or e-money institution whose network membership the cards are issued under.

The vendors sit in different lanes. Marqeta sells global enterprise scale. Block (Cash App, Square) was about 42% to 45% of its net revenue from Q1 2025 to Q1 2026, per its 10-Q filings, and on 31 July 2025 it completed the acquisition of TransactPay, a UK/EEA EMI and BIN sponsor. Lithic is a card issuing API with direct Visa and Mastercard connections. It's aimed at US B2B, AP, and SaaS programs that want Authorization Stream Access (ASA) and, in the processor-only model, their own sponsor bank, such as Column or Lead Bank. Stripe Issuing is the no-setup-fee route for Connect users, Highnote targets fleet and AP programs, and Adyen Issuing funds cards from Adyen acquiring balances. If you're shortlisting Marqeta alternatives, those four are the list.

As Idealogic CTO, I review card engines across fintech, logistics and supply chain, social and creator economy, SaaS development, product development, software development consulting, and custom software development. It sits next to our guides on banking as a service, payment processing software, and embedded finance. Below: the vendors, a PostgreSQL 17 card ledger with row-level security, and a 36-month TCO at $60M/yr ($180M over 36 months). Vendor figures were checked on 14 Nov 2026.

The short version

  • Lithic is US-focused: direct Visa and Mastercard connections, ASA webhooks, and a choice between program-managed and processor-only (bring your own bank). Marqeta is certified to operate in 40+ countries, uses Just-in-Time (JIT) Funding, and since July 2025 owns UK/EEA licensing through TransactPay. Marqeta pricing is negotiated.
  • Stripe Issuing (no setup fee, $0.10 per virtual card and $3.50 per physical card in the US) is the fastest launch inside Connect. In our model you give up more interchange, and cross-border fees apply. Highnote targets fleet and AP; Adyen Issuing pairs with Adyen acquiring.
  • Partial shipments, incremental holds, reversals, and 0220 FX or tip force-posts break a single balance column. You need separate BIGINT buckets in PG17: available_cents, pending_hold_cents, remaining_hold_cents, and overdraft_cents.
  • At $60M/yr ($5M/mo), Lithic core plus a direct bank and an owned PG17 ledger costs $388,200 over 36 months in our model. That's $201.9k less than Lithic managed ($590,100), $517.5k less than Stripe Issuing ($905,700), and $578.7k less than Marqeta enterprise ($966,900). The extra build cost ($48,000 at CEE rates) pays back in 1.86 to 5.64 months. Every vendor fee and basis-point split here is an Idealogic modeling assumption, not a published rate.

Lithic vs Marqeta vs Stripe Issuing: card issuing platform internals in 2026

Two things rarely show up on a pricing page: whether the vendor connects to the card networks directly, and what its revenue share looks like once your volume grows.

Four architectural and economic fault lines separating Lithic, Marqeta, Stripe Issuing, and a direct processor plus owned PostgreSQL 17 sub-ledger in 2026

1. Lithic card issuing API: direct Visa/Mastercard connections & ASA

Lithic grew out of Privacy.com and is a card issuing API with direct Visa and Mastercard connections. Authorization Stream Access (ASA) is its real-time authorization webhook: Lithic sends each authorization to your endpoint and waits for an approve or decline.

  1. Program-managed or processor-only: In the program-managed model, Lithic and its partner bank run the program. In processor-only, you bring your own sponsor bank and BIN.
  2. ASA timing: Auth Rules filter MCC codes before ASA calls your backend. Lithic's ASA docs recommend answering within 3 seconds and decline at 6. Under 150ms P99 is our own target, not a Lithic spec.
  3. Clearing and disputes: You get clearing data, 3DS, and dispute APIs. The product is US-focused.

2. Marqeta: global JIT Gateway & TransactPay EEA

Marqeta processes for Block (Cash App, Square) and other large fintechs. Just in time funding (JIT Funding) is its model for loading a card at the moment of purchase instead of in advance.

  1. JIT Gateway Funding: Funds zero-balance cardholder accounts from a program funding source at swipe time. Your gateway has 3 seconds to answer, and no response means a decline.
  2. TransactPay and 40+ countries: Marqeta says it's certified to operate in 40+ countries. TransactPay, acquired in July 2025, is a BIN sponsor: the licensed institution whose network membership your cards are issued under. It holds UK and EEA EMI licences and was live in 25 countries at the acquisition.
  3. Customer concentration: Block was about 42% to 45% of net revenue from Q1 2025 to Q1 2026.

Marqeta pricing in 2026: platform fees, setup and revenue share

Marqeta pricing isn't published. Expect a quote covering a setup fee, a monthly platform minimum, and a revenue share on interchange. Reported minimums are often in the low five figures per month. The $9,500/mo fee and 32 bps share in the TCO below are Idealogic modeling assumptions, not Marqeta rates.

3. Stripe Issuing pricing, Highnote & Adyen Issuing

  • Stripe Issuing: Launches in days to a few weeks inside Connect. Stripe Issuing pricing in the US is no setup fee, no stated monthly minimum, $0.10 per virtual card and $3.50 per physical card. Cross-border and currency conversion fees apply (see Stripe pricing), and the interchange gap in our model is an assumption, not a published number. Real-time authorization gives you 2 seconds, and local issuing covers 24 countries.
  • Highnote offers issuing and acquiring for US fleet and AP programs. Adyen Issuing funds cards from balances on Adyen's acquiring side. If acquiring is the bigger decision for you, read Finix vs Stripe Connect.
CapabilityLithicMarqetaStripe IssuingHighnote / Adyen
CoreDirect Visa/MC connectionsJIT Gateway + TransactPayTurnkey ConnectFleet/AP (Highnote); Acquirer+Issuer (Adyen)
ReachUS-focusedCertified in 40+ countries (US, UK, EEA)24 countries (US, UK, EEA)US (Highnote); see Adyen docs
Auth timeoutASA: 3s recommended, 6s hardJIT: up to 3s timeoutSync webhook (2s)Sync webhooks; see vendor docs
PricingProgram-managed to processor-onlyNegotiated minimum + bpsNo setup fee; $0.10/$3.50 + FXCommercial commits
PortabilityHigh (BYO bank)Medium (bundled/BIN)Stripe banks onlyBundled US; Adyen Bank

Card issuing sub-ledger blueprint: PostgreSQL 17 RLS and the ASA responder

Sponsor banks apply third-party risk guidance such as FDIC FIL-29-2023 to fintech programs, and consumer or prepaid programs also fall under Regulation E (12 CFR Part 1005). Business-purpose commercial cards generally sit outside Reg E; network rules and your program agreement govern those disputes. Either way, a multi-tenant SaaS transaction management ledger has to survive retried ISO 8583 0100 requests, 0120 auth advices, 0420 reversal advices (or their JSON webhook equivalents), split captures, and 0220 FX or tip force-posts that clear above the remaining hold and the available balance. EMVCo 3DS 2.x covers the online authentication side.

We build this ledger on PostgreSQL 17 and tested it there. The same row-level security syntax works on PostgreSQL 18. Six decisions carry most of the weight:

Design decisionEnforced inWhat it prevents
Money is stored as BIGINT cents (PostgreSQL numeric types) in four buckets per account: available, pending hold, settled spend, and overdraft, each with a non-negative checkDatabaseRounding drift and the negative-balance crashes a single balance column produces
Every table uses FORCE ROW LEVEL SECURITY with a restrictive and a permissive policy on the session tenant, and the app role has no BYPASSRLSDatabaseOne tenant's swipe reading or debiting another tenant's sub-ledger, even through the table owner
Each journal leg is unique on tenant, network event token, and leg number, so a retried ASA request or clearing advice returns the stored decisionDatabase and appDouble holds and double debits when the processor resends an event
The responder locks the account row, then the authorization row (SELECT FOR UPDATE), in one transaction, and declines a frozen account, an account already in overdraft, or a blocked MCC before touching balancesAppTwo concurrent swipes spending the same funds, and slow answers past Lithic's 3-second recommended and 6-second hard ASA limits (our own target is under 150ms P99)
A capture draws from the authorization hold first, then from available funds, and books the rest as overdraft; the final capture releases any unused hold, and a clearing with no matching authorization opens a zero-hold recordAppRejecting force-posts the network has already settled, and holds that never release after split shipments
Inbound webhooks are checked against the Svix signature scheme Lithic uses for event webhooks, with a 5-minute timestamp window and a length check before the timing-safe compareAppForged or replayed events; confirm how the ASA request itself is authenticated in Lithic's docs

The capture split is the part that differs most from a plain wallet ledger. A clearing amount is drawn from the authorization hold first, up to what remains of it. Anything above the hold (FX drift, a tip that clears higher than authorized) is debited from available funds, and whatever still has no cover is booked as overdraft, because a force-post has already settled on the network. On the final capture, any unused hold is released.

Each non-zero part becomes its own journal leg, which is why the leg number sits in the uniqueness key. Hosted iframes and a Basis Theory proxy keep raw card numbers out of your VPC, which can substantially reduce your scope under PCI DSS v4.0.1. Your QSA or sponsor bank decides the SAQ or ROC level.

Losing 24 to 44 bps of commercial card interchange?
Share your monthly spend and card mix. We benchmark Lithic, Marqeta, Stripe Issuing, and an owned PG17 sub-ledger.
Audit your issuing TCO

Lithic vs Marqeta TCO at $60M a year: 36-month cost benchmark

Take a Series A/B B2B spend or AP product over 36 months: $60,000,000/yr ($5,000,000/mo, $180,000,000 cumulative) across 15,000 cards (3,000 physical and 12,000 virtual, with 4% of volume FX-converted cross-border).

We assume 220 bps of gross commercial charge-card interchange. That's $110,000/mo or $3,960,000 over the period, so 1 bp is worth $500/mo, or $18,000 over 36 months. Commercial credit and charge cards sit outside the Regulation II (12 CFR Part 235) debit interchange cap, and real interchange varies by network, product, and MCC. Subtract an assumed 40 bps of network assessments ($720,000) and the issuer pool is 180 bps ($90,000/mo, $3,240,000).

Fair warning: the 220 bps, the 40 bps, and every vendor fee and split below (32, 44, 24, and 9 bps) are Idealogic modeling assumptions, not published vendor rates.

Net 36-month cost is setup + fees/FX + forgone interchange + ops, priced at Idealogic's $75/hr senior CEE rate against an assumed $165/hr US agency rate (2.2x). For context, neither is a wage statistic: Eurostat puts the 2025 EU average employer labour cost at EUR 34.90 per hour, and the BLS May 2024 median pay for US software developers works out to roughly $64 per hour as an employee wage. The four options:

  1. Option A: Marqeta Ent. (JIT): Setup 220h ($16,500 CEE / $36,300 US) + 36-mo fees ($9,500/mo = $342,000) + forgone 32 bps ($16,000/mo = $576,000; nets 148 bps) + 12h/mo (432h) ops ($32,400 CEE / $71,280 US) = $966,900 ($1,025,580 US).
  2. Option B: Stripe Issuing: Setup 100h ($7,500 CEE / $16,500 US) + 36-mo card/FX/dispute fees ($2,500/mo = $90,000) + forgone 44 bps ($22,000/mo = $792,000; nets 136 bps) + 6h/mo (216h) ops ($16,200 CEE / $35,640 US) = $905,700 ($934,140 US).
  3. Option C: Lithic Managed Ent. (ASA): Setup 140h ($10,500 CEE / $23,100 US) + 36-mo fees ($3,500/mo = $126,000) + forgone 24 bps ($12,000/mo = $432,000; nets 156 bps) + 8h/mo (288h) ops ($21,600 CEE / $47,520 US) = $590,100 ($628,620 US).
  4. Option D: Lithic Core + Direct Sponsor Bank + Owned PG17 FORCE RLS & Basis Theory: Build 640h ($48,000 CEE / $105,600 US) + 36-mo bank floor ($2.5k/mo), Basis Theory ($1.1k/mo), and AWS ($600/mo) ($4,200/mo = $151,200) + 9 bps split ($4,500/mo = $162,000; nets 171 bps = $3,078,000) + 10h/mo (360h) ops ($27,000 CEE / $59,400 US) = $388,200 ($478,200 US).
Worked 36-month card issuing cost ($k) on $180M commercial spend ($5M/mo): Lithic Processor Core + Owned PG17 saves $201.9k vs Lithic Managed, $517.5k vs Stripe Issuing, and $578.7k vs Marqeta Enterprise
36-month item ($180M / 180 bps)Option A: Marqeta Ent.Option B: Stripe IssuingOption C: Lithic Ent.Option D: Lithic + PG17Driver
1. Setup / build ($75/hr CEE)$16,500 (220h; $36.3k US)$7,500 (100h; $16.5k US)$10,500 (140h; $23.1k US)$48,000 (640h; $105.6k US)PG17 & PCI proxy
2. 36-mo Platform, FX & Infra$342,000 ($9,500/mo)$90,000 ($2,500/mo)$126,000 ($3,500/mo)$151,200 ($4,200/mo)No $9.5k floor/FX
3. 36-mo Forgone Interchange$576,000 (32 bps / $16k/mo)$792,000 (44 bps / $22k/mo)$432,000 (24 bps / $12k/mo)$162,000 (9 bps / $4.5k/mo)Nets 171 bps
4. 36-mo Ledger & Recon Ops$32,400 (12h/mo; 432h)$16,200 (6h/mo; 216h)$21,600 (8h/mo; 288h)$27,000 (10h/mo; 360h)Daily FBO recon
Total 36-Month Cost$966,900$905,700$590,100$388,200 (CapEx incl.)Saves $201.9k to $578.7k
Year-2+ Run-Rate$316,800/yr ($26.4k/mo)$299,400/yr ($24.95k/mo)$193,200/yr ($16.1k/mo)$113,400/yr ($9.45k/mo)Saves $79.8k to $203.4k/yr
36-Mo Net Retained+$2,273,100+$2,334,300+$2,649,900+$2,851,800Zero lock-in

Sensitivity check: Seed velocity vs Series A/B payback

  • Seed (under $10M/yr): Rent Stripe Issuing or a program-managed Lithic setup. For UK/EEA, look at Marqeta (TransactPay) or Adyen Issuing.
  • Payback ($60M/yr): At $75/hr CEE, Option D pays back in 1.86 months against Marqeta ($31,500 / $16,950/mo), 2.61 months against Stripe ($40,500 / $15,500/mo), and 5.64 months against Lithic Managed ($37,500 / $6,650/mo). It still wins at $165/hr US: $478,200 against $628,620 to $1,025,580.

2026 card issuing & PostgreSQL 17 sub-ledger build costs (senior CEE vs US)

Our fintech, SaaS development, API integration, and custom software development teams ship fixed-milestone builds at $75/hr senior CEE vs $165/hr US (2.2x):

Engagement tierTimelineHoursSenior CEE ($75/hr)US Agency ($165/hr)Deliverables
Stripe / Lithic Managed Sprint2 to 4 wks100 to 180h$7,500 to $13,500$16,500 to $29,700Hosted PAN UI, ASA webhooks & limits
Marqeta JIT / Global EEA Setup5 to 8 wks220 to 360h$16,500 to $27,000$36,300 to $59,400JIT Gateway, TransactPay EEA 3DS & cards
Owned PG17 Card Core (640h)9 to 12 wks520 to 760h$39,000 to $57,000 ($48k)$85,800 to $125,400 ($105.6k)PG17 FORCE RLS ledger, PCI vault & FBO recon
Multi-Sponsor Fleet / AP Suite13 to 18 wks880 to 1,280h$66,000 to $96,000$145,200 to $211,200Dual BIN router, L2/L3 controls, ERP & disputes
Dedicated Fintech PodOngoing240h/mo (2)From $18,000/moFrom $39,600/moNode.js/TypeScript, PostgreSQL 17 & PCI QA

Inside the worked 640-hour owned card ledger build

Worked model, assumptions in text: 640-hour ($48,000 at $75/hr CEE vs $105,600 at $165/hr US) owned PostgreSQL 17 multi-capture card sub-ledger, sub-150ms ASA responder, PCI proxy & daily FBO reconciliation
  • 1. PG17 Ledger & FORCE RLS (25% / 160h / $12,000 CEE vs $26,400 US): BIGINT accounts, authorizations, multi-leg journal, AS RESTRICTIVE walls, and overdraft_cents.
  • 2. ASA & JIT Webhooks (25% / 160h / $12,000 CEE vs $26,400 US): Sub-150ms responder, Svix HMAC, (tenant_id, event_token, leg_seq) deduplication, and MCC rules.
  • 3. PCI DSS v4.0.1 Vault & 3DS (20% / 128h / $9,600 CEE vs $21,120 US): Basis Theory PAN proxy, hosted reveals, Apple/Google Pay provisioning, and 3DS 2.x.
  • 4. FBO & Clearing Recon (20% / 128h / $9,600 CEE vs $21,120 US): Three-way matching across PG17, clearing tapes, and BAI2/camt.053 FBO files (Plaid vs Stripe).
  • 5. Disputes & Load QA (10% / 64h / $4,800 CEE vs $10,560 US): Chargebacks under network rules, Regulation E provisional credits for consumer programs, 0120 replay tests, and 2,500 TPS benchmarks.

Minting a sandbox card takes hours. Keeping fifteen thousand commercial sub-ledgers balanced across split captures, FX force-posts, and tight timeouts protects bank audits and up to thirty-five basis points of margin.

Artem Zaitsev · CTO at Idealogic

Lithic vs Marqeta vs Stripe Issuing: shipped proof & 5-step decision checklist

We've shipped banking and payout products on these rails before:

  • Neobank (Fese): Accounts, cards, and loans on sponsor rails (digital banking software).
  • Mobile banking (C-Bank): PCI DSS v4.0.1 & OWASP MASVS (+30% downloads, +40% MAUs, +25% cross-service usage).
  • Split payouts (Ugreator): Cross-network Stripe Connect settlements.

Five checks settle the Lithic vs Marqeta vs Stripe Issuing choice:

  1. Under $10M/yr on Connect? Stripe Issuing. No setup fee.
  2. US B2B/AP ($10M to $50M/yr)? Lithic with ASA.
  3. Need 40+ countries or UK/EEA EMI? Marqeta (TransactPay) or Adyen Issuing.
  4. Fleet fuel prompts? Compare Highnote against Lithic.
  5. Past $50M/yr US spend? Lithic processor-only, your own sponsor bank, and an owned PG17 ledger. In our model that's $388,200 over 36 months and 171 bps net. Talk to us about running the model on your numbers.
Comparing Lithic vs Marqeta or migrating off turnkey splits?
Send your monthly spend and card mix. Our senior CEE architects return a PG17 sub-ledger blueprint within 48 hours.
Request a 48-hour issuing audit

Occasional field notes on building software, no spam

Protected by Cloudflare Turnstile · Privacy · Terms

Artem Zaitsev
Artem ZaitsevCTO · Idealogic Development OÜ

CTO and Co-founder at Idealogic. Directing engineering, systems architecture, security, and full-stack delivery across web, mobile, and blockchain.

Frequently asked questions

  • Choose Lithic when you're launching a US B2B spend, accounts payable, fleet, or vertical SaaS card program and want a card issuing API with Authorization Stream Access webhooks, direct Visa and Mastercard connections, and a processor-only option with your own sponsor bank. Choose Marqeta when you run a multinational consumer neobank, BNPL checkout, or a card program across many countries (Marqeta says it's certified in 40+), especially in the UK and European Economic Area now that it completed the TransactPay acquisition on 31 July 2025.

  • Stripe Issuing is the fastest launch inside an existing Stripe Connect stack: no setup fee, no stated monthly minimum, and US card creation at 10 cents per virtual card and 3.50 dollars per physical card. Cross-border and currency conversion fees apply. Marqeta pricing isn't public: reported negotiated minimums are often in the low five figures per month, so confirm yours in a quote. Lithic sits between the two. In our model it keeps more net commercial interchange, but those basis points are Idealogic assumptions, not published vendor rates.

  • Both pause an inbound Visa or Mastercard authorization and send a synchronous HTTPS request to your backend, so your ledger approves or declines the swipe in real time. Marqeta Just-in-Time (JIT) Funding ties that decision to moving money from a program funding source into the cardholder account at authorization time, with a 3-second gateway timeout. Lithic Authorization Stream Access keeps the authorization decision separate from the underlying bank settlement and gives you a 3-second recommended and 6-second hard response budget.

  • Two things. Block (Cash App, Square) was about 42% to 45% of Marqeta net revenue between Q1 2025 and Q1 2026 per its 10-Q filings. And Marqeta completed its acquisition of TransactPay, a UK and EEA BIN sponsor and Electronic Money Institution, on 31 July 2025. TransactPay was live in 25 countries at the time, so Marqeta now handles processing, EMI licensing, and BIN sponsorship in Europe itself.

  • Look at Highnote when you're building commercial fleet, fuel, travel, or accounts payable virtual card programs and want issuing and acquiring from one vendor. Look at Adyen Issuing when you already process merchant acquiring on Adyen and want to fund issued cards from those balances.

  • A card transaction is never a single atomic debit. E-commerce merchants split one eighty-dollar authorization across three partial shipment captures, hotels append incremental holds, foreign currency rates shift between swipe and settlement, and restaurant tips clear above the original pending hold. Mutating one balance column without separating available cents, pending hold cents, remaining per-authorization hold cents, and force-post overdraft cents causes negative-balance crashes or double-deducted funds on retried network advices.

  • PCI DSS v4.0.1 scope follows where raw 16-digit Primary Account Numbers, CVVs, and PINs are transmitted, processed, or stored. Hosted card-reveal iframes from Lithic, Marqeta, or Stripe, plus Apple Pay and Google Pay push-provisioning SDKs, keep raw cardholder data off your servers. When B2B travel or procurement workflows must send virtual card numbers to third-party supplier APIs, a Basis Theory or Very Good Security forward proxy can substantially reduce your PCI scope. Your QSA, acquirer, or sponsor bank decides which SAQ or ROC level applies.

  • Below 10 million dollars in annual card spend, Stripe Issuing or a program-managed Lithic setup keeps fixed costs lowest. In our worked model (Idealogic assumptions, not published vendor rates), at 60 million dollars in annual commercial spend over 36 months, which is 180 million dollars cumulative at an assumed 220 basis points of gross interchange, a 640-hour owned PostgreSQL 17 FORCE RLS sub-ledger and PCI proxy build at 75 dollars per hour senior CEE rates (48,000 dollars) brings the 36-month cost to 388,200 dollars and pays back in under six months.

Still unanswered
Ask us directly

A senior engineer replies under 4 hours.