Purpose Designed For Scale: How We Built It | Lithic

Purpose Designed For Scale: How We Built It

October 14, 2025

Jason Kruse
Chief Technology Officer

We founded Lithic because we recognized that ambitious financial technology companies deserve infrastructure partners who can match their innovation and scale. We are mission critical infrastructure and that is not a responsibility we take lightly. We agonize over millisecond latency, engineer redundancy and resiliency, and obsess over accuracy– because in payment processing, uptime matters and correctness is non-negotiable.

Our modern card issuing platform was designed from the ground up to eliminate the limitations we experienced with legacy processors when scaling our own fintech, Privacy.com. Today, we process billions in monthly volume with 99.999%+ authorization uptime, supporting hundreds of programs with always-on systems that don’t require scheduled maintenance or downtime windows.

‘Real’ Infrastructure Designed for High Availability

In modern cloud computing, provisioning containers with a single click has become standard practice. However, payment processing demands a different approach. We operate a hybrid cloud infrastructure that combines horizontal scalability from cloud computing with the deterministic performance guarantees of traditional datacenter infrastructure.

Our infrastructure consists of single-provider fiber circuits connecting directly from our datacenters to payment networks and the Federal Reserve. This approach eliminates the variability and unreliability inherent in multi-hop internet routing and allows us to achieve a higher uptime SLA and lower transaction latency than our competitors.

Close But Not Too Close

We maintain datacenters in Northern California and Oregon adjacent to AWS’ us-west-1 and us-west-2. This geographic separation provides:

src)

This infrastructure represents several years of engineering effort and millions of dollars in capital expenditure. The investment reflects the critical nature of payment processing and our commitment to customers who’ve entrusted us with their financial lives. If our systems are unavailable, there are real people who cannot pay rent or buy groceries. It is our commitment to these cardholders that drives our continued mission to maintain the most reliable processing stack available.

Physical Direct Connectivity

A critical differentiator in payment processing infrastructure is the connection into the payment networks. We maintain physical, direct connections to the payment networks rather than routing through third-party gateways or even the internet. That is, we have multiple single-provider, private circuits into Visa, Mastercard, and the Federal Reserve – the entire internet could be down, and we will continue to route and process transactions.

When evaluating payment processors, key questions to ask are:

Multiple redundant private circuits provide stable and predictable routing, bandwidth, and latency. This can be an advantage relative to public internet routed connectivity. Although the public internet is often fairly resilient, this type of connectivity can be subject to changes in routes, number of hops (and thus end-to-end latency), and congestion outside of the control of the payment processor or network.

Redundant From the Top Down

We design our system to achieve at least N+1 redundancy – our redundancy extends beyond simple backup connections. We maintain four diverse private circuits- two from each geographically separated datacenter. This N+3 redundancy ensures continued operation even with multiple simultaneous failures.

The redundancy stack includes:

This defense-in-depth approach to redundancy eliminates single points of failure across power, network, and compute infrastructure ensuring our customers’ cards work when they most need them, always.

Software That Scales

We implement much of our critical path, core authorization processing logic in Rust, a strongly-typed systems programming language with zero-cost abstractions and no garbage collection that provides deterministic, low latency performance. Its strict compiler and ownership model ensures predictable runtime performance and safe multithreaded execution, scaling to thousands of transactions per second (TPS).

For data persistence, we use DynamoDB for its full ACID semantics and guaranteed O(log n) query performance. Unlike traditional relational databases that use a declarative language like SQL where query performance can degrade unpredictably based on query complexity and data distribution, DynamoDB provides consistent single-digit millisecond latency at any scale through a composite hash partitioning index and B-tree indexing. This architectural choice trades flexibility in query patterns for horizontal scalability and guaranteed performance bounds essential to the performance, throughput, and uptime demands of transaction processing.

Building for the Next Decade, Not the Last One

Next gen payment infrastructure isn’t just about uptime and speed, it’s also about enabling product experiences that weren’t possible before. Our architecture powers features that fundamentally change what developers can build, from real-time transaction intelligence to flexible ledger systems that support novel account types. While other providers are constrained by their underlying legacy systems, our infrastructure enables:

Lithic’s product development ethos has always prioritized giving our clients the flexibility, control, and transparency to build better payment products that change peoples’ lives. That thesis has never been more critical. Frankly, if you're going to build payment infrastructure from scratch in 2025, it should perform like infrastructure built in 2025. That means real-time processing, enterprise-grade reliability, and the architectural flexibility to support product innovations we haven't even thought of yet.

The choice isn't just about features or pricing—it's about whether you're building on the network topology of the future or getting trapped by the limitations of the past.