Confluent
Enterprise SaaS + hybrid subscription: consumption-based Confluent Cloud (fully managed Kafka/Flink) plus term-license Confluent Platform (self-managed), land-and-expand; now operated inside IBM Software
Earnings, margins, COGS & capex
FY2025 total revenue $1,166.7M (+21% YoY), driven by Confluent Cloud $624M (+27%). Subscription revenue $1,119.7M (+21%). Gross margin healthy at 74.3% GAAP; still GAAP operating-loss-making (-$380.1M) but non-GAAP operating margin turned solidly positive (7.4%) and free cash flow was positive (GAAP FCF $38.1M; adjusted FCF $76.0M). Growth had decelerated from the 30%+ era into the low-20s as cloud consumption headwinds and a mid-2025 go-to-market model change bit, but RPO re-accelerated in 2H25. IBM acquired the company at $31/share, closing 2026-03-17.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~26¢ is cost of goods and ~74¢ operating expense, and the remainder is an operating loss.
Revenue trend
Margins
up vs 73.3% FY2024
stable
improving from -43.5% ($419.1M loss) FY2024
expanding from 2.9% ($27.5M) FY2024 — the core margin-inflection story
improving; GAAP FCF $38.1M vs $9.5M FY2024
narrowing loss vs -$345.1M FY2024
COGS structure
COGS is dominated by third-party public-cloud hosting/compute for Confluent Cloud plus customer support and services delivery; as Cloud mix rises, hosting is the key gross-margin swing factor. Non-cloud subscription (Platform licenses) carries higher margin. ~26 cents of every GAAP revenue dollar is cost of revenue.
Capex
Minimal — asset-light; no owned data centers. Capitalized software and modest office/equipment only. Free cash flow is close to operating cash flow because capex is small.
Latest earnings
Q4 subscription revenue $301.6M landed above the guided $295.5-296.5M range; results generally met/beat consensus. No earnings call or forward guidance was issued given the pending IBM acquisition.
Standalone forward guidance is moot — the company was acquired by IBM (closed 2026-03-17) and no longer issues independent guidance. Prior FY2025 framing had targeted ~$1.10-1.11B subscription revenue and ~6% non-GAAP operating margin; actuals came in ahead ($1,119.7M subscription, 7.4% non-GAAP operating margin).
- $100k+ ARR customers
- 1,521 (Q4 2025, +10% YoY)
- $1M+ ARR customers
- ~234 (Q3 2025, +27% YoY)
- Net revenue retention (NRR)
- ~114% (Q3 2025), stabilized
- Gross retention
- ~90% (Q3 2025)
- Total RPO
- $1.42B (Q3 2025, +22% YoY)
- Current RPO
- $754M (Q3 2025, +19% YoY)
- Enterprise customers
- >6,500, ~40% of the Fortune 500 (per IBM close release)
Growth drivers
- Confluent Cloud consumption growth (+27% FY2025) as workloads migrate from self-managed Kafka to fully managed cloud
- Flink stream-processing (managed Apache Flink) expanding the platform beyond Kafka messaging into real-time processing
- Real-time data for AI/agents — positioning Kafka as the 'nervous system' feeding LLMs and agentic workflows fresh, trusted data (core IBM acquisition thesis)
- Land-and-expand: $100k+ ARR customers 1,521 (+10% YoY, Q4 2025) and $1M+ ARR customers ~234 (Q3 2025, +27%)
- Consumption/GTM model change in 2025 shifting to a consumption-forward motion; RPO re-accelerated for four consecutive quarters into 2H25
Bull & bear
The bull case was effectively realized: Confluent owned the streaming category with a margin/FCF inflection intact, and IBM paid a ~24-28% premium ($31/share, ~$11B) to acquire the real-time data layer for enterprise AI — closing the equity thesis at a clean cash outcome.
- Only independent, category-leading pure-play in data streaming — a scarce strategic asset, which is why IBM (and reportedly other suitors) bid
- FY2025 proved the model can compound at ~20%+ while flipping to positive non-GAAP operating margin (7.4%) and positive free cash flow
- AI wave structurally increases demand for real-time, streaming data pipelines — a secular tailwind Confluent sits directly under
- $31 all-cash at ~9-10x forward revenue delivered a certain, premium exit versus carrying execution and multiple-compression risk
The bear case was the reason a sale made sense: decelerating growth, persistent GAAP losses, heavy SBC, and encroaching hyperscaler/Kafka-compatible competition capping the standalone multiple — and the $31 price, while a premium, is well below CFLT's post-IPO highs.
- Growth had roughly halved from its 30%+ peak; consumption softness and a bungled 2025 GTM change showed the model's sensitivity
- Free, ubiquitous open-source Kafka plus cheaper Kafka-compatible rivals (Redpanda, WarpStream) and bundled hyperscaler services pressure pricing power
- Never reached GAAP profitability as an independent company (-$380.1M GAAP operating loss FY2025); value rested on non-GAAP metrics and stock-comp-heavy economics
- The $31 take-out crystallized a modest outcome relative to the 2021 IPO-era valuation — a premium to a depressed price, not a triumphant one
- Integration/execution risk under IBM could slow the product velocity that made Confluent the leader
What it is worth
Settled acquisition value (not a forward-looking model) — cross-checked against revenue multiple
$31.00
even the bear must mark to the closed cash price; the bear critique is that $31 crystallized a modest exit versus the IPO-era valuation, not that the price is at further downside risk (there is none, the deal is done).
$31.00 all-cash
final — deal closed, consideration fixed and paid.
$31.00 (the realized cash price)
the bull outcome was the take-out itself; ~9-10x forward revenue for the category leader.
Valuation is no longer a market question: IBM acquired Confluent for $31.00/share all-cash, ~$11B enterprise value (equity value roughly $10.5-11B depending on dilution), closing 2026-03-17. That equates to roughly ~9-10x FY2025 revenue ($1,166.7M) / ~10x forward revenue — a premium (~24-28% to the pre-announcement price) but well below CFLT's 2021 IPO-era peak multiples. CFLT no longer trades; there is no independent public equity to value.
SWOT
Strengths
- Category-defining position in data streaming as the commercial steward of Apache Kafka, the de facto open-source streaming standard
- 74%+ gross margins with a proven non-GAAP operating-margin and free-cash-flow inflection (FY2025 7.4% non-GAAP op margin / positive FCF)
- Strong enterprise install base: >6,500 customers, ~40% of Fortune 500, expanding $1M+ ARR cohort
- Platform breadth beyond messaging — managed Flink processing, connectors, governance — creating a fuller real-time data stack
- IBM ownership brings enterprise distribution, on-prem/hybrid reach, and balance-sheet stability
Weaknesses
- Growth decelerated from 30%+ to low-20s as cloud consumption slowed and a 2025 GTM/consumption model change disrupted the motion
- Still GAAP operating-loss-making (-$380.1M FY2025) with heavy stock-based compensation driving the GAAP/non-GAAP gap
- Open-source Kafka is free — Confluent must continually justify a paid managed premium versus self-hosting or Kafka-compatible rivals
- Cloud gross margin is exposed to hyperscaler hosting costs (its COGS runs on AWS/Azure/GCP, who also compete with it)
- Loss of independent capital-markets identity/optionality now inside IBM
Opportunities
- Real-time data as the substrate for enterprise AI/agents — the explicit IBM thesis; every agent needs fresh, trusted, streaming data
- Cross-sell into IBM's large enterprise and hybrid-cloud base (watsonx, Red Hat, consulting)
- Managed Flink and stream governance monetization expanding wallet share beyond core Kafka
- Displacing legacy ETL/batch pipelines with streaming-first architectures
Threats
- Hyperscaler substitutes — Amazon MSK/Kinesis, Azure Event Hubs, Google Pub/Sub bundled cheaply into cloud contracts
- Kafka-compatible challengers (Redpanda, WarpStream) and Pulsar (StreamNative) undercutting on cost/simplicity
- Data-platform adjacents (Snowflake, Databricks) building native streaming/ingest that reduces need for a standalone streaming vendor
- Integration risk / talent attrition inside IBM; potential deprioritization of the standalone product roadmap
Moats, dependencies & bottlenecks
Moats
Deep mindshare and de facto standard status, but the standard itself is open source, so the moat is expertise/managed-experience, not IP lock
Streaming pipelines become mission-critical central nervous systems; ripping out Confluent means re-plumbing real-time data flows across the org
Fuller stack raises the bar for point-solution rivals, but each layer faces specialized competitors
Undercut by the fact that it rents infrastructure from hyperscalers who are also competitors
Dependencies
Microsoft Azure, Google Cloud (AMZN, MSFT, GOOGL) Infrastructure supplier & marketplace channel — and direct competitor Confluent Cloud runs on these hyperscalers (hosting is its main COGS) while MSK/Event Hubs/Pub-Sub compete with it — supplier and rival simultaneously
Core technology foundation Product is built on OSS Confluent stewards; a fork or a stronger Kafka-compatible standard would erode differentiation
Owner / capital / distribution Roadmap, investment, and go-to-market priorities now set by IBM Software; upside is distribution, risk is deprioritization/integration drag
Consumption model ties revenue to customer usage, which flexes with macro and cloud-optimization cycles
Advantages
- Undisputed category leadership and Kafka provenance
- Broadest managed streaming + processing platform in one vendor
- Large, sticky Fortune-500-heavy enterprise base with expanding high-ARR cohorts
- Proven path to positive non-GAAP margins and free cash flow at scale
- Now backed by IBM's enterprise sales, hybrid-cloud footprint, and balance sheet
Weaknesses
- Persistent GAAP losses and high stock-based compensation (standalone)
- Growth deceleration and 2025 GTM-transition self-inflicted disruption
- Structural pricing pressure from free OSS Kafka and cheaper compatible rivals
- Infrastructure dependence on competitors (hyperscalers)
- Loss of independent strategic control post-IBM
Bottlenecks
- Cloud gross margin capped by hyperscaler hosting costs it does not control
- Consumption-based revenue makes near-term growth sensitive to customer cost-optimization and usage throttling
- Monetizing against free open-source Kafka requires continuous premium justification
- Post-acquisition talent retention and roadmap continuity inside a much larger, slower-moving parent
Top signals & trends
Top signals
Resolved / neutral for CFLT holders · Equity thesis terminated at a fixed premium price; no further public trading
Forward-booked demand rebuilt after the mid-2025 consumption/GTM disruption
Durable profitability inflection — a key reason IBM could underwrite the price
Expansion moderated from the 120%+ era but stabilized after consumption headwinds
Maturation plus competitive/consumption pressure; part of the rationale for selling rather than going it alone
Trends
Core secular tailwind and the explicit strategic logic of IBM's purchase — agents need fresh streaming data
Structurally expands the addressable market for Kafka/Flink-based platforms
Commoditization/pricing pressure on standalone streaming vendors
Undercut on infra cost and operational simplicity for cost-conscious buyers
Both a partner ecosystem and an encroachment risk on the ingest layer
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 cloud hosting for Confluent Cloud; also a competitor
Cloud hosting + marketplace; also a competitor
Cloud hosting + marketplace; also a competitor
Open-source technology foundation Confluent commercializes and stewards
Broad enterprise base across financial services, retail, tech, telecom; land-and-expand consumption model
Large financial-services and retail enterprises Real-time fraud, event-driven commerce, and operational data pipelines are anchor use cases
Managed Kafka (MSK) and Kinesis bundled into AWS; both a hosting supplier and a direct substitute
Kafka-protocol-compatible Event Hubs bundled into Azure enterprise agreements
Managed messaging/streaming bundled into GCP; competes on price and integration
Data + AI platform adding streaming ingest (Spark Structured Streaming, DLT); competes for the real-time data budget
Snowpipe Streaming and native ingest reduce need for a separate streaming layer for some workloads
Kafka-API-compatible, C++-based, positioned on lower cost and operational simplicity
Pulsar-based streaming alternative to Kafka
Previously offered MQ/Event Streams (Kafka on Red Hat); resolved the overlap by acquiring Confluent outright