Tigris Data

Tigris Data Inc. runs a globally distributed object storage service that is compatible with the Amazon S3 API, allowing applications written for S3 (boto3, aws-cli, rclone, PyTorch and similar tooling) to use Tigris by changing the endpoint. The service markets bottomless capacity with no per-account or per-region limits, zero egress fees, and per-bucket placement options across single-region, multi-region or global configurations. Platform capabilities described by the company include standard, infrequent-access and archive storage tiers with lifecycle rules, zero-copy copy-on-write bucket forks, whole-bucket point-in-time snapshots and restore, object-event webhooks, fine-grained S3-compatible IAM with per-bucket/per-prefix/per-key policies, custom domains with managed TLS, and presigned URLs. An agent-oriented SDK (@tigrisdata/agent-kit) exposes forks, ephemeral per-agent workspaces with TTL and scoped credentials, checkpoints, and webhook-based coordination for multi-agent pipelines.

The company positions the service against the object storage offerings of AWS, Google Cloud and Microsoft Azure, which its CEO refers to collectively as "Big Cloud" and criticizes for egress fees (a "cloud tax") and centralized architectures. Tigris says its system replicates data to where GPUs are located, supports billions of small files such as embeddings, and delivers low-latency access for training, inference and agentic workloads. Reported latency techniques include caching based on a log-structured merge-tree data structure that keeps data in adjacent storage segments, and automatic relocation or cached replication of frequently accessed repositories closer to the requesting region. A migration feature called "shadow buckets" incrementally copies an application's most frequently used objects from another platform rather than requiring a bulk dataset move.

At an earlier stage the company described Tigris differently: a 2023 interview presented it as a developer data platform combining a NoSQL database for structured and unstructured data with an integrated full-text and vector search engine, with event streaming and caching planned.

Founding story

The three co-founders worked together for roughly six years on Uber's global storage platform, where they built and operated systems including Docstore, Herb and DBEvents supporting the Uber Rides and Uber Eats apps. Ovais Tariq led Uber's storage/database infrastructure teams and has said he spent 17 years in the data infrastructure space before Tigris; he grew up in Karachi, studied computer science at the University of Karachi, worked at a local IT firm and remotely for a Singapore-based database company before joining Uber in the United States. The founders' stated motivation was to make the kind of managed storage platform Uber's internal developers enjoyed available to any developer, under the mission of simplifying storage for developers.

Business model

Tigris sells usage-based cloud object storage directly to developers and companies. Its service agreement states that fees for the Tigris Object Storage Service are billed monthly in advance and payable within 30 days of invoice, that users register accounts (optionally via GitHub or Google identity providers), and that access may also be purchased through authorized resellers. Enterprise users can invite end users under their account. During the 2023 public beta on Fly.io the service was not charged for usage, with any eventual charges intended to appear on the customer's Fly.io bill.

Usage-based fees for the object storage service, invoiced monthly in advance, with storage priced by tier (standard, infrequent access, archive) and no charge for egress bandwidth. During the Fly.io public beta usage was free and billing was intended to flow through the Fly.io bill.

Traction

Reported metrics include more than 4,000 customers as of October 2025, most of them AI startups, and CEO-stated 8x annual growth since founding. The company's website cites 1M+ buckets, 20PB+ of storage, 10B+ objects and 99.99% availability, and a case study claiming 85% storage cost savings. fal.ai is named as a customer, with its head of engineering stating that egress fees previously made up the majority of the company's cloud spend.

Latest developments

In October 2025 Tigris Data announced a $25 million Series A led by Spark Capital with participation from Andreessen Horowitz, to be used to extend its hardware footprint to London, Frankfurt and Singapore. Current website positioning centers on storage for AI training, inference and agent workloads, including an agent-kit SDK, checkpoints, forks and workspaces.