this distributed data storage startup wants to take on big cloud

The Rise of Distributed Compute and a New Approach to Data Storage

The rapid growth of Artificial Intelligence (AI) companies is creating unprecedented demands on computing resources. Companies such as CoreWeave, Together AI, and Lambda Labs are successfully meeting this demand, attracting significant investment for their distributed compute capabilities.

The Limitations of Traditional Cloud Storage

Despite this trend, the majority of organizations continue to rely on the established “big three” cloud providers – AWS, Google Cloud, and Microsoft Azure – for data storage. These systems were originally designed to maintain data proximity to their own computing infrastructure, rather than across diverse cloud environments or geographical locations.

Ovais Tariq, co-founder and CEO of Tigris Data, explains, “Contemporary AI applications and infrastructure are increasingly favoring distributed computing over traditional big cloud solutions. We aim to offer a comparable storage solution, recognizing that compute power is ineffective without readily available data.”

Introducing Tigris Data: AI-Native Storage

Tigris Data, founded by the team behind Uber’s storage platform, is developing a network of strategically located data storage centers. This network is specifically engineered to address the distributed compute requirements of modern AI workloads.

The startup’s AI-native storage platform is designed to “move with your compute,” automatically replicating data to GPU locations, supporting vast numbers of small files, and delivering low-latency access for training, inference, and agentic applications, according to Tariq.

Securing Series A Funding

To facilitate this development, Tigris recently secured $25 million in Series A funding, led by Spark Capital, with participation from existing investors including Andreessen Horowitz.

Tariq positions Tigris as a challenger to the established “Big Cloud” providers, asserting that their services are both more costly and less efficient.

The Issue of Egress Fees

Historically, AWS, Google Cloud, and Microsoft Azure have imposed egress fees – often referred to as “cloud tax” – when customers migrate data to alternative cloud providers or download it for use with cheaper GPUs or geographically diverse training processes.

Batuhan Taskaya, head of engineering at Fal.ai, a Tigris customer, notes that these fees previously constituted the largest portion of Fal’s cloud expenditure.

Addressing Latency Concerns

Beyond egress fees, Tariq highlights the issue of latency inherent in larger cloud providers. He states, “Egress fees were merely a symptom of a more fundamental problem: centralized storage unable to keep pace with a decentralized, high-velocity AI ecosystem.”

Serving the Needs of Generative AI Startups

The majority of Tigris’ 4,000+ customers, like Fal.ai, are generative AI startups focused on developing image, video, and voice models, which require large, latency-sensitive datasets.

Tariq illustrates this need: “Consider an AI agent processing local audio. Minimal latency is crucial. Both compute and storage should be localized for optimal performance.”

Optimizing for AI Workloads

He adds that existing large cloud infrastructures are not optimally configured for AI workloads. Streaming substantial datasets for training or real-time inference across multiple regions can introduce latency bottlenecks, hindering model performance.

However, access to localized storage enables faster data retrieval, allowing developers to execute AI workloads reliably and cost-effectively within decentralized cloud environments.

Taskaya of Fal.ai confirms, “Tigris enables us to scale our workloads across any cloud by providing access to the same data filesystem from all locations without incurring egress charges.”

Data Security and Ownership

Companies are also motivated to locate data closer to their distributed cloud options due to concerns surrounding data security, particularly within highly regulated industries like finance and healthcare.

Tariq further emphasizes the growing desire for data ownership, referencing Salesforce’s recent decision to restrict AI competitors from utilizing Slack data. He explains, “Organizations are increasingly recognizing the importance of data as the driving force behind LLMs and AI. They seek greater control and independence.”

Future Expansion Plans

With the new funding, Tigris plans to expand its data storage center network to meet increasing demand. The company has experienced 8x growth annually since its inception in November 2021.

Currently operating data centers in Virginia, Chicago, and San Jose, Tigris intends to broaden its presence in the U.S., as well as in Europe and Asia, with planned expansions in London, Frankfurt, and Singapore.