Redis
Open-core commercial open source (AGPLv3/SSPL/RSAL tri-license) with usage-based DBaaS (Redis Cloud on AWS/Azure/GCP), self-managed enterprise subscriptions (Redis Software), and marketplace-led sales
Total raised: $347M net per company / $355M gross per Tracxn across 8 rounds (earlier rounds: Series A-E 2012-2019 incl. $44M Series D led by Goldman Sachs PCI in 2017 and $60M Series E led by Francisco Partners in 2019 - valuations for those rounds were not officially disclosed, so they are omitted from the trail). No primary round since Apr 2021. Secondary-market marks (Hiive $16.17/share on 2026-07-09 with 0 live orders; Nasdaq Private Market $18.85/share estimate on 2026-06-25) cannot be converted to a reliable implied cap without a public share count, so no post-2021 valuation point is plotted.
Earnings, margins, COGS & capex
Redis passed $300M ARR in January 2026, up from roughly $200M when Rowan Trollope became CEO in February 2023 - solid but unspectacular growth (~14% implied CAGR) for the vintage, now reaccelerating on AI workloads. 12,000 paying customers, a third of the Fortune 100, and 50+ customers spending over $1M/year (up 20%+ YoY). Profitability, gross margin, and cash burn are not disclosed; the company has not raised primary capital since its $110M Series G (Apr 2021, led by Tiger Global with SoftBank Vision Fund 2 and TCV, $2B+ valuation), which suggests either near-breakeven operations or reliance on the 2021 war chest.
Revenue trend
Margins
likely pressured by hyperscaler infrastructure COGS on Redis Cloud; enterprise self-managed licenses carry software-like margins
no primary raise since 2021 hints at discipline, but unverified
COGS structure
Not disclosed. Structurally: Redis Cloud COGS is dominated by compute/memory rented from AWS, Azure, and GCP (RAM-heavy workloads are expensive to host); the Speedb storage-engine acquisition (Mar 2024) targets exactly this - tiering data to flash to cut the memory bill. Self-managed Redis Software revenue carries minimal COGS.
Capex
Not disclosed; asset-light model - infrastructure is opex via hyperscalers, not owned data centers.
Latest earnings
not applicable
None issued; company narrative targets AI/agentic workloads as the next growth phase
- ARR
- $300M+ (Jan 2026)
- Paying customers
- 12,000 (Jan 2026)
- $1M+ ARR customers
- 50+, up 20%+ YoY (Jan 2026)
- Fortune 100 penetration
- ~one-third (Jan 2026)
- RedisVL downloads
- Nearly 1M in Dec 2025, 10x vs Dec 2024
- Total equity raised
- $347M net ($355M gross per Tracxn) across 8 rounds; last: $110M Series G (Apr 2021, Tiger Global, SoftBank Vision Fund 2, TCV)
Growth drivers
- AI agent memory and context — Redis positioned as the short-term memory / context engine for agentic systems (named customers OpenAI, Uber, Lovable)
- Vector search adoption — RedisVL vector library downloads reached nearly 1M in Dec 2025, 10x vs Dec 2024 and 3x vs Sep 2025; Vector Sets data type announced Apr 2025, beta in Redis 8 (May 2025)
- LangCache semantic caching (announced Apr 2025, public preview Sep 2025) - cuts customers' LLM inference spend, a budget line growing everywhere
- Return to open source (AGPLv3 with Redis 8, May 1, 2025) rebuilding developer goodwill and funnel after the 2024 license backlash
- Enterprise expansion: 50+ $1M+ customers growing 20%+ YoY; land-and-expand within a third of the Fortune 100
- Acquisitions filling product gaps — Speedb (Mar 2024, flash-tiered storage engine) and Decodable (announced Sep 2025, real-time data pipelines feeding context into Redis)
Bull & bear
Redis is the default real-time data layer of the internet quietly becoming the default memory layer of AI agents - $300M+ ARR, Fortune-100 penetration, a reunified open-source funnel under AGPL, and a product surface (vector search, semantic caching, streaming context) that maps directly onto where AI infrastructure spend is going. A 2021 valuation mark of $2B looks cheap against current infra-software ARR multiples.
- $300M+ ARR with 50+ $1M+ customers growing 20%+ YoY shows real enterprise pull, not just OSS popularity
- AI tailwind is concrete, not narrative: OpenAI is a named customer; RedisVL downloads up 10x YoY to nearly 1M in Dec 2025; LangCache monetizes LLM cost pain
- AGPL return plus Redis 8's consolidated feature set (JSON, query engine, vector) restores the developer funnel Valkey threatened
- Speedb flash-tiering attacks the main COGS and pricing objection (RAM cost), expanding the addressable dataset size
- At 8-12x ARR - normal for growing infra software - Redis would be worth $2.4B-$3.6B+, above its last mark; IPO or strategic acquisition (it is a scarce asset in AI data infrastructure) offer multiple exit paths
Redis is a ~14%-growth 2021 unicorn whose crown-jewel open-source project was successfully forked by the richest companies on earth. The hyperscalers now ship a free, competitively-benchmarking, cheaper substitute inside their own consoles, the AI positioning is contested by every database vendor alive, and nothing about margins or burn is disclosed.
- Valkey is permanent: AWS, Google, and Oracle back a drop-in BSD alternative priced 20% below Redis OSS node-based tiers (and ~33% below on serverless) on ElastiCache - structural, well-capitalized price pressure at the exact point of customer choice
- Growth from ~$200M (Feb 2023) to $300M+ (Jan 2026) is roughly 14% CAGR - below the typical infra-software IPO bar and below the multiple its $2B mark assumes
- Zero margin disclosure: RAM-intensive cloud hosting suggests below-peer gross margins, and profitability is unproven
- AI memory layer is a knife fight: hyperscaler-native stores, Postgres+pgvector, MongoDB, and standalone vector DBs all claim the same workload; Redis has no exclusive position
- License reversals (BSD to SSPL to AGPL in 14 months) signal strategic zig-zag and permanently ceded community leadership to a neutral foundation
- Secondary marks ($16-19/share, incl. a Hiive mark with zero live orders) with no primary round since Apr 2021 leave price discovery weak; late-2021-vintage privates routinely re-price down at IPO
What it is worth
Last priced round + ARR-multiple triangulation (no public financials; all multiples are analyst framing, not disclosed data)
~$1.5B-$2B (5-6.5x ARR)
flat-to-down vs 2021 if growth stays around the mid-teens and Valkey pressure compounds; secondary illiquidity persists
~$2.4B-$3B (8-10x ARR)
modestly above the 2021 mark, consistent with a disciplined IPO or strategic sale
~$3.6B-$4.5B (12-15x ARR)
if AI-driven reacceleration into 25%+ growth is demonstrated into an IPO
Last priced: $2B+ post-money (Series G, $110M, led by Tiger Global with SoftBank Vision Fund 2 and TCV, Apr 2021; accompanied by a ~$200M secondary). No primary round since. Secondary platforms mark shares at $16-19 (Jun-Jul 2026) with thin order flow and no reliable public share count to convert into an implied cap. Triangulation off the one hard metric - $300M+ ARR (Jan 2026): private infra-software comps trade roughly 6-15x ARR depending on growth; Redis's ~14% growth argues for the low-to-mid end, its AI positioning and Fortune-100 base for a premium.
SWOT
Strengths
- Category-defining brand — Redis is among the most-loved and most-deployed databases in developer surveys; enormous installed base
- Sub-millisecond in-memory performance is a genuine technical moat for caching, session, leaderboard, and now agent-memory workloads
- 12,000 paying customers incl. a third of the Fortune 100; 50+ $1M+ accounts
- Multi-cloud DBaaS (Redis Cloud) plus self-managed offering covers both cloud-native and regulated buyers
- Redis 8 consolidated JSON, time series, probabilistic types, and the query/vector engine into core - a broader platform than the cache label suggests
Weaknesses
- Valkey fork (Linux Foundation, backed by AWS, Google, Oracle, Ericsson, Snap) is a free, drop-in, BSD-licensed substitute actively marketed by the hyperscalers at lower price points
- License whiplash (BSD to SSPL Mar 2024 to AGPL May 2025) burned community trust; the AGPL return does not dissolve Valkey
- Growth (~$200M to $300M ARR over ~3 years, ~14% CAGR) is modest for a company priced at 2021 multiples
- No disclosed profitability or margins; RAM-heavy cloud COGS is structurally expensive
- Stale $2B valuation mark from Apr 2021 creates down-round / flat-IPO overhang for employees and late investors
Opportunities
- Agentic AI memory layer — every agent needs fast state, session, and semantic cache - Redis's core competency repackaged for the biggest budget line in software
- LLM cost optimization via LangCache semantic caching - sells savings, not just speed
- Flash/SSD tiering (Speedb) can cut customers' memory bills and expand Redis into larger, cheaper datasets
- Real-time data pipeline integration (Decodable) turns Redis from a cache into a context platform
- IPO window: at $300M+ ARR Redis is IPO-scale; a listing would provide currency and liquidity (no IPO filed as of Jul 2026)
- Vector search consolidation — bundled vector capability can undercut standalone vector DBs (Pinecone, Weaviate, Qdrant)
Threats
- Hyperscalers steering workloads to Valkey-based services (AWS ElastiCache/MemoryDB, Google Memorystore) that they monetize themselves
- Performance-focused challengers — DragonflyDB claims multi-fold throughput on modern hardware; Valkey benchmarks competitively vs Redis OSS at 20% lower node-based and ~33% lower serverless AWS pricing
- General-purpose databases (MongoDB, PostgreSQL ecosystem, Elastic) absorbing caching and vector use cases
- If agentic AI standardizes on hyperscaler-native or model-vendor-native memory stores, Redis's AI narrative weakens
- Down-round or extended illiquidity if private-market conditions stay tight for 2021-vintage unicorns
Moats, dependencies & bottlenecks
Moats
Redis protocol/API is a de facto standard; millions of deployments. Eroding at the margin because Valkey speaks the same protocol.
Caches are semi-stateless and Valkey is drop-in compatible (forked from Redis 7.2.4), so switching cost is lower than for systems of record; enterprise features (Active-Active CRDT geo-replication, RDI) are stickier.
Active-Active replication) Real but contested - Dragonfly and Valkey benchmark competitively on raw throughput; Redis 8 claims up to 87% faster commands and 2x throughput vs prior OSS.
AGPL keeps hyperscalers from re-hosting new Redis versions without reciprocity; but Valkey exists precisely because this lever was pulled.
Ecosystem integrations (client libraries, LangChain, frameworks) help but replicate quickly for Valkey.
Dependencies
supplier + channel + competitor Redis Cloud runs on and is sold through the hyperscalers' marketplaces while they simultaneously sell competing Valkey-based services.
The freemium funnel depends on developers choosing Redis over Valkey at project start; the AGPL move (May 2025) was damage control.
In-memory economics dominate COGS and customer TCO; Speedb flash-tiering is the mitigation.
AI-workload growth rides on being the default memory/cache integration in agent frameworks.
No primary raise since Apr 2021; burn profile undisclosed. An IPO or new round is the liquidity path.
Advantages
- De facto standard protocol and brand for in-memory data
- Broadest feature surface in its category post-Redis 8 (JSON, query, vector, time series, probabilistic)
- Named AI marquee customers (OpenAI, Uber, Lovable) validating the agent-memory thesis
- Multi-cloud neutrality vs single-cloud native caches
- Enterprise-grade differentiators: Active-Active geo-replication, RBAC, on-prem/hybrid deployment
Weaknesses
- Drop-in free substitute (Valkey) backed by AWS, Google, Oracle, Ericsson, and Snap
- Community trust damaged by two license changes in 14 months (SSPL Mar 2024, AGPL May 2025)
- ~14% growth at a 2021 unicorn price
- No disclosed profitability; RAM-heavy COGS
- Founder-project split history (creator Salvatore Sanfilippo left leadership in 2020; returned Nov 2024 as a developer evangelist, not an executive)
Bottlenecks
- Hyperscaler channel conflict: the biggest distribution channels are also the best-funded competitors
- RAM-bound cost structure limits price competitiveness for large datasets until flash tiering (Speedb) fully lands
- Undisclosed margins/burn make the equity hard to underwrite from outside
- Growth rate (~14% CAGR) must reaccelerate on AI workloads to justify a mark above $2B at IPO
Top signals & trends
Top signals
Voluntary disclosure of a round-number milestone often precedes IPO positioning; no IPO filing as of Jul 2026.
Leading indicator of AI-workload funnel.
Suggests cash discipline or strong balance; terms undisclosed.
Thin, inconsistent price discovery; no official re-price since Apr 2021.
The fork is compounding, not fading; the discount is permanent, not promotional.
Community-repair actions with measurable goodwill effect, though Valkey persists and community reception was mixed.
Trends
Directly matches Redis's latency profile; the core growth thesis.
LangCache sells cost reduction - resilient even in tight budgets.
Valkey is the canonical case study; margin and funnel pressure is permanent.
Helps Redis vs standalone vector DBs; hurts pricing power for vector as a feature.
pgvector plus in-Postgres caching patterns absorb some Redis use cases at the low end.
Provides an exit path but likely at disciplined multiples for ~15% growers.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Primary hosting substrate for Redis Cloud
Hosting + Azure Managed Redis co-sell partnership
Hosting and marketplace channel
Named customer - AI/agent workloads (private)
Named customer - real-time infrastructure
Named customer - AI app-builder startup (private)
Per Jan 2026 company disclosure
Largest channel and largest competitor; ships Valkey 20% cheaper node-based and ~33% cheaper serverless than Redis OSS tiers.
Valkey backer; native cache service on GCP.
Partner-competitor; Azure Managed Redis is a Redis Inc partnership, making Microsoft the friendliest hyperscaler.
Valkey backer; OCI cache services.
General-purpose document DB absorbing caching/vector/AI-memory workloads; Atlas competes for the same AI-app stack.
Search/vector overlap for RAG workloads; similar open-source-to-AGPL license arc.
Private; Redis-API-compatible engine claiming large multi-core throughput advantages.
Not a company but the single biggest competitive force - BSD-licensed drop-in fork of Redis 7.2.4 with hyperscaler funding (AWS, Google, Oracle, Ericsson, Snap).
Caching + document platform competing in enterprise accounts.
Private; flash-optimized real-time NoSQL competing at large-dataset, low-latency workloads.
Private standalone vector DBs contesting the AI retrieval layer Redis wants to bundle.