Sacra estimates that Redis reached at least $300M in annual recurring revenue (ARR) in January 2026, up from $297M at the end of 2025. Redis disclosed the $300M+ milestone alongside 12,000+ paying customers and 50+ accounts spending over $1M annually, with the number of million-dollar accounts growing more than 20% year over year.rowth was fastest in 2021 and 2022 as Redis added cloud and enterprise customers, then moderated through 2023-2024 amid infrastructure spending optimization and the licensing transition.

Redis generates revenue from Redis Cloud, a managed, consumption-oriented database service priced by memory, throughput, topology, and enterprise features across AWS, Google Cloud, and Azure, and Redis Software, which sells annual enterprise licenses and support for on-premises, private-cloud, and hybrid deployments. Premium support, professional services, cloud marketplace transactions, and channel partners provide additional revenue. Cloud is the fastest-growing segment and likely accounts for roughly 60% of current revenue, with self-managed enterprise contributing most of the remainder.

Redis was valued at over $2B after raising a $110M Series G in April 2021 led by Tiger Global Management, with participation from SoftBank Vision Fund 2, TCV, Bain Capital Ventures, Francisco Partners, Goldman Sachs Growth, Viola Ventures, and Dell Technologies Capital.

Founded in 2009 and headquartered in San Francisco, Redis raised earlier rounds from investors including Viola Ventures, Dell Technologies Capital, Bain Capital Ventures, Francisco Partners, and Goldman Sachs Growth. Its $100M Series F in August 2020 valued the company above $1B for the first time.

Redis has raised approximately $347M in total funding across all rounds.

Redis is an in-memory database that keeps frequently accessed data close to the application layer, allowing reads and writes in microseconds rather than the milliseconds typical of disk-based databases. A traditional database functions like a warehouse in another building, while Redis acts as a well-organized refrigerator next to the kitchen.

Developers connect applications to Redis using client libraries for Python, Java, JavaScript, Go, or .NET. The application checks Redis first for the data it needs. If the item is there, Redis returns it immediately. If not, the application queries the primary database, returns the result, and writes a copy to Redis with an expiration time that removes stale data. This pattern offloads repeated reads from the system of record, allowing a single relational or document database to support a larger workload.

Beyond caching, Redis provides programmable data structures that can increment counters atomically, maintain ranked leaderboards through sorted sets, publish messages to subscribers, queue tasks in streams, store and query JSON documents, run full-text and geospatial searches, and compare vector embeddings for semantic similarity. Teams can use these capabilities in place of separate caching, queueing, counter, leaderboard, and lightweight search systems.

Redis 8, released in 2025, integrated formerly separate modules such as Search, JSON, TimeSeries, and Bloom filters into the core distribution. Developers installing Redis now get vector search, hybrid text-and-vector queries, and probabilistic data structures without adding extensions.

For AI applications, Redis serves as a vector store for retrieval-augmented generation, a semantic cache for LLM responses through LangCache, and a persistent memory layer for agents through Agent Memory. Redis Iris, introduced in May 2026, packages these capabilities into higher-level managed services, including Context Retriever, which gives agents governed, tool-based access to operational business data without raw database credentials.

Deployment options include Redis Cloud for fully managed infrastructure, Redis Software for self-managed enterprise environments, and a bring-your-own-cloud model in which Redis operates the software within the customer's AWS account. Redis Flex tiers hot data in RAM and warm data on NVMe flash, allowing larger datasets to avoid all-RAM pricing.

Redis uses an open-core model in which free open-source software creates a developer funnel, while commercial products monetize production deployments. Its B2B go-to-market motion has a product-led component: developers adopt Redis through open-source packages, free cloud tiers, client libraries, and framework integrations. Enterprises then purchase managed cloud or self-managed software for availability, security, support, and governance.

Redis Cloud charges for provisioned capacity rather than purely per-command usage. Customers select dataset size, throughput, replication topology, and cloud region, then pay hourly or through discounted annual commitments. This model provides more predictable revenue than serverless pricing and gives customers reserved resources and consistent performance. Revenue expands as applications store more data, require higher throughput, add availability zones, deploy Active-Active multi-region databases, or adopt search, vector, and AI context services.

Redis Software is sold through sales-led enterprise contracts for self-managed deployments. Because customers provide the infrastructure, this product has structurally higher gross margins than hosted cloud, offset by longer sales cycles and more complex support environments. Under the BYOC model, Redis operates the software while customers pay cloud providers directly for compute and storage. This reduces Redis's infrastructure cost exposure while retaining a managed-service relationship.

Licensing determines how Redis converts open-source adoption into commercial revenue. In March 2024, Redis moved newer code to RSALv2 and SSPLv1 to prevent hyperscalers from offering competing managed services without commercial arrangements, prompting the Valkey fork. In May 2025, Redis added AGPLv3 as an OSI-approved option for Redis 8, restoring open-source status while retaining copyleft obligations for hosted or modified network deployments. Redis Cloud and Redis Software customers operate under separate commercial terms.

Expansion starts when Redis enters as a cache or session store and teams adopt its counters, streams, search, and vector capabilities, consolidating more workloads and moving to larger configurations. AI context services such as Agent Memory, LangCache, and Context Retriever add another consumption path: agent interactions generate memory, cached responses, and retrieval activity, increasing cloud usage above the base database layer.

Redis competes across managed caching, Redis-compatible open-source engines, and real-time AI data infrastructure. The primary competitive dynamic is protocol commoditization: Redis compatibility has become an acquisition channel for competitors rather than a moat.

AWS is Redis's most direct competitor because it controls infrastructure procurement, billing, and customer relationships in the largest public cloud. ElastiCache for Valkey prices up to 20-33% below Redis OSS tiers and supports in-place migration with minimal application disruption. AWS can also bundle Valkey into existing committed-spend agreements, making it a default shortlist option for AWS-centric organizations that primarily need caching.

Google Cloud Memorystore uses a similar model, managing Valkey and Redis through a common cloud-native experience with native GCP networking, monitoring, and billing. Microsoft takes a different approach: Azure Managed Redis is a first-party service built on Redis Enterprise technology, creating a distribution partnership rather than a competitive fork. Microsoft is retiring legacy Azure Cache for Redis tiers through 2027-2028 and directing Azure customers toward Redis Enterprise, though Microsoft retains control of the customer relationship and pricing surface.

Valkey is the strongest ecosystem-level threat. Governed by the Linux Foundation and backed by AWS, Google, Oracle, and others, it preserves Redis protocol and client compatibility under permissive BSD licensing. Valkey 9 added built-in full-text and hybrid search, narrowing a functional gap with Redis Stack. Organizations that prefer permissive licensing or deep hyperscaler integration can adopt Valkey with low switching friction.

Dragonfly targets performance-sensitive workloads with a multithreaded, thread-per-core architecture that claims up to 25x Redis throughput in vendor benchmarks. A single large Dragonfly instance can replace a sharded Redis cluster, reducing infrastructure costs for high-throughput deployments. Dragonfly does not need to displace Redis broadly: winning a small number of very large clusters could weaken Redis's expansion within its highest-value accounts.

The broader database market is converging around Redis's newer capabilities. MongoDB, Couchbase, and DataStax compete for operational database budgets, while Postgres continues absorbing simpler caching and real-time workloads as its performance improves. Pinecone, Weaviate, Qdrant, and Turbopuffer compete for vector search and AI retrieval use cases, while SingleStore, ClickHouse, and Firebolt overlap with Redis in real-time analytics.

Redis Iris moves the company into direct competition with agent-memory services, AI gateway providers, and data integration platforms such as Confluent and Redpanda. Neo4j addresses graph workloads that Redis does not natively serve. Redis risks broadening its product scope before establishing differentiation outside caching, while specialists focus on narrower adjacent workloads.

Redis is expanding from its historic role as a caching and session-store layer into a real-time data platform for AI agents, production ML, and enterprise applications.

The largest expansion opportunity is serving as the default memory and context layer for production AI agents. Redis Iris packages agent memory, semantic caching, governed context retrieval, and data synchronization into managed services above the database. Multi-step agents that operate across hours and business systems generate more memory, retrieval, and checkpoint activity per task than simple chatbot interactions, increasing Redis consumption even as model inference costs decline.

RedisVL's nearly one million December 2025 downloads and reported use by 42% of surveyed developers for agent-memory storage indicate early traction. Redis aims to shift from a low-level infrastructure purchase to a higher-value AI platform relationship in which each new agent interaction generates incremental cloud consumption.

Redis is expanding upstream from serving data into moving and transforming it. Redis Data Integration, which reached general availability on AWS in June 2026, uses change-data capture to keep Redis synchronized with operational databases and warehouses. The September 2025 acquisition of Decodable added real-time streaming pipelines and connectors.

Redis Feature Form, launched in preview in April 2026 following the Featureform acquisition, extends Redis into ML feature orchestration, including definition, versioning, lineage, and sub-millisecond online serving. This adds a TAM adjacent to generative AI, with production use cases in fraud detection, credit scoring, recommendations, and personalization. These products let Redis capture spending across the path from source data to low-latency application or model access.

Redis serves 58% of the Fortune 50 and has over 50 customers spending more than $1M annually, but the implied average ARR per customer of roughly $25,000 suggests room for expansion within the existing base. Active-Active multi-region deployment, Flex storage tiering, BYOC, and Iris services provide upsell paths as datasets grow and customers consolidate workloads.

Redis Cloud operates across AWS, Google Cloud, and Azure regions spanning North America, Europe, Asia-Pacific, Latin America, the Middle East, and Africa. Partnerships with Adfinis across EMEA and APAC and Fierce Software for U.S. federal and defense environments extend distribution into regulated and sovereignty-sensitive markets, where Valkey's permissive licensing alone does not address procurement, compliance, and support requirements.

Protocol commoditization: Valkey's Linux Foundation governance, permissive BSD licensing, and backing from AWS, Google, and other hyperscalers are turning Redis-compatible caching into a commodity utility, threatening Redis's ability to monetize basic workloads and potentially detaching the Redis developer ecosystem from Redis the company.

Hyperscaler dependence: Cloud providers are Redis's most important distribution channels and strongest competitors, giving them control over default service placement, pricing, committed-spend budgets, and customer relationships, while AWS and Google promote lower-priced Valkey alternatives and Microsoft controls go-to-market for Azure Managed Redis.

Platform overreach: Redis's expansion across open-source database, managed cloud, self-managed enterprise software, BYOC, Azure co-delivery, Flex storage, Search, RDI, Feature Form, Decodable streaming, and the Iris agent-context suite, combined with the July 2026 reduction of approximately 200 roles, raises the risk that it cannot integrate acquisitions and mature multiple preview-stage AI products while advancing the core database against well-funded competitors in each adjacent market.