WarpStream

WarpStream was founded in mid-2023 (the company's earliest public web presence dates to early August 2023) by Richard Artoul (CEO) and Ryan Worl (CTO), two engineers who had previously worked together at Datadog, where they helped build Husky, a columnar, object-storage-native database that powers Datadog's log search. Frustrated by the cost and operational complexity of running Apache Kafka in the cloud -- particularly the massive inter-availability-zone networking fees created by Kafka's local-disk, triple-replication design, which the founders estimated could represent 70-90% of a large Kafka deployment's infrastructure cost -- Artoul and Worl set out to answer the question, 'What would Kafka look like if redesigned from scratch for a modern cloud environment?' The result, introduced publicly in an August 2023 blog post titled 'Kafka Is Dead, Long Live Kafka,' was a Kafka-protocol-compatible streaming platform built directly on commodity object storage (S3, GCS, Azure Blob), using stateless, JVM-free Go binaries called 'Agents' in place of traditional Kafka brokers, with no ZooKeeper/KRaft, no local disks, and no manual partition rebalancing. WarpStream marketed the product as delivering roughly 5-10x lower total cost of ownership than self-managed or lift-and-shifted Kafka.

On March 14, 2024, WarpStream announced a $20 million funding round led by Greylock and Amplify Partners, with additional participation from data-infrastructure industry angel investors. At the time of the raise the company had just nine employees and was pursuing two go-to-market approaches: a Bring-Your-Own-Cloud (BYOC) deployment model (available immediately) and a fully managed serverless offering. Over the following months the team grew rapidly -- from three people at launch to thirteen by the time of its next major milestone -- achieved general availability, attained SOC 2 Type II certification, expanded into new cloud regions, launched additional products (including Managed Data Pipelines), and won a cohort of high-profile customers such as Grafana Labs, Character.AI, Zomato/Eternal, ShareChat, PostHog, Cursor, Robinhood, Goldsky, Secureworks and Superwall, largely in observability, logging, AI/ML telemetry and blockchain-indexing use cases.

WarpStream's rapid growth and its distinctive BYOC-native, 'zero disk architecture' approach attracted the attention of Confluent (the company founded by Apache Kafka's original creators, including Jay Kreps). Kreps reached out to WarpStream co-founder Richard Artoul directly; the two met in New York and, over the following months, negotiated a path for WarpStream to join Confluent. On September 9, 2024, Confluent announced it had acquired WarpStream, publishing a blog post authored by CEO and co-founder Jay Kreps titled 'Confluent acquires WarpStream.' The financial terms of the deal were not publicly disclosed by either company; press coverage (TechCrunch, Forbes) noted only that WarpStream had raised $20 million from Greylock and Amplify Partners roughly six months earlier. At the time of the acquisition, WarpStream had grown its team to thirteen people. In the deal, Confluent positioned WarpStream (subsequently rebranded 'WarpStream, powered by Confluent' / 'Confluent WarpStream') as a third product pillar alongside the fully managed Confluent Cloud and the self-managed Confluent Platform, filling the 'bring your own cloud' middle ground on the deployment spectrum. WarpStream's own leadership published a companion post, 'WarpStream is Dead, Long Live WarpStream,' emphasizing that the product, brand, and team would continue operating largely as before but with Confluent's engineering resources, enterprise support organization, and customer base behind it.

Business model

B2B infrastructure software (Bring-Your-Own-Cloud data streaming platform), sold via usage-based cloud consumption pricing, direct sign-up console, and cloud marketplaces (AWS Marketplace, GCP Marketplace); also offered a free-tier fully hosted control plane.

Usage-based / consumption pricing for compute (Agent vCPU-hours) and control-plane metadata operations, since customer data itself stays in the customer's own object storage bucket; free credits offered to new signups.