Tigris Data Challenges Big Cloud With Decentralized Storage for AI

The surge in artificial intelligence startups has skyrocketed demand for powerful computing resources. Distributed compute providers like CoreWeave, Together AI, and Lambda Labs have stepped up to meet this need. Yet, for data storage, most businesses remain tied to major cloud providers—AWS, Google Cloud, and Microsoft Azure—whose infrastructure is designed to keep data close to their own compute, not distributed globally.

Enter Tigris Data, a startup founded by the developers behind Uber’s storage platform. Tigris aims to disrupt the status quo by offering a distributed, AI-native storage network that can keep up with the unique needs of modern AI workloads.

What Makes Tigris Different?

According to Ovais Tariq, co-founder and CEO of Tigris Data, the current cloud giants aren’t optimized for the flexibility and speed that AI workloads demand. "Modern AI workloads and AI infrastructure are choosing distributed computing instead of big cloud," he said.

• Storage That Moves With Compute: Tigris automatically replicates data to where GPUs are available, minimizing latency for training and inference tasks.
• Support for Billions of Small Files: Essential for generative AI startups working with image, video, and voice models.
• No Egress Fees: Unlike major cloud providers that charge users to move data out (sometimes called a "cloud tax"), Tigris eliminates these costs, making scaling across multiple clouds financially viable.
• Low Latency: Localized storage centers ensure rapid data access, essential for real-time AI agent applications.

Real-World Impact

For companies like Fal.ai, the benefits are tangible. Batuhan Taskaya, Fal’s head of engineering, shared that egress fees with traditional clouds were once their largest expense. Now, Tigris allows them to scale workloads across different clouds without those additional costs, while providing access to the same data filesystem everywhere.

Tariq also highlights a major pain point with centralized storage: latency. "Egress fees were just one symptom of a deeper problem: centralized storage that can’t keep up with a decentralized, high-speed AI ecosystem." For AI models relying on rapid, local access—for example, real-time audio agents—every millisecond counts.

Data Control and Compliance

Beyond performance and cost, data control is becoming a top priority for businesses. High-profile incidents, such as Salesforce blocking rivals from using Slack data, have made companies more conscious about owning and securing their data. This is especially critical in regulated sectors like finance and healthcare, where data locality and compliance are non-negotiable.

Growth and Future Plans

To fuel its expansion, Tigris recently raised a $25 million Series A led by Spark Capital, with participation from Andreessen Horowitz and others. Since launching in late 2021, Tigris has seen impressive growth—expanding 8x annually and serving over 4,000 customers. The company already operates data centers in Virginia, Chicago, and San Jose and plans to expand further in the US, Europe (London, Frankfurt), and Asia (Singapore).

As AI workloads become ever more distributed, solutions like Tigris could reshape how businesses think about both data storage and sovereignty, offering greater efficiency, security, and control.

References

• “Modern AI workloads and AI infrastructure are choosing distributed computing instead of big cloud,” Ovais Tariq, co-founder and CEO of Tigris Data, told TechCrunch.
• “Egress fees were just one symptom of a deeper problem: centralized storage that can’t keep up with a decentralized, high-speed AI ecosystem.”