The explosion of AI companies has promoted the demand for computing power to new extremities and companies such as Coreweave, Together AI and Lambda Labs have capitalized from this demand, attracting a huge amount of attention and capital for their ability to provide distributed capacity.
However, most companies continue to store data with the three large cloud providers, AWS, Google Cloud and Microsoft Azure, whose storage systems were built to maintain data close to their own computing resources, not spreading multiple clouds or areas.
“Modern workloads and AI infrastructure select distributed calculation instead of a large cloud,” Ovais Tariq, co -founder and chief executive of Tigris Data, told TechCrunch. “We want to provide the same option for storage, because without storage, the calculation is nothing.”
Tigris, founded by the Uber storage platform, creates a network of local data storage centers that claims to be able to meet the distributed computing needs of modern AI workloads. The Store Store Platform ”moves with your calculation, [allows] data [to] They automatically reproduce where the GPUs are, support billions of small files and provide low delay for training, conclusions and agency loads, “Tariq said.
To do all this, Tigris recently set a round of $ 25 million, led by Spark Capital and saw the participation of existing investors, including Andreessen Horowitz, Techcrunch has learned exclusively. The start is against the established bodies, which Tariq calls a “big cloud”.
Tariq feels that these established bodies offer not only a more accurate data storage service, but also a less effective. AWS, Google Cloud and Microsoft Azure have historical debt output fees (called “cloud tax” in the industry) if a customer wants to emigrate to another cloud provider or download and transfer their data if they want to use a cheaper GPU or train models in different parts of the world. Think about it as if you have to pay your gym extra if you want to stop going there.
According to Batuhan Taskaya, head of engineering at Fal.Ai, one of Tigris’s customers, these expenses once represented the majority of cloud spending.
TechCrunch event
Francisco
|
27-29 October 2025
In addition to the exit fees, Tariq says there is still the problem of delay with larger cloud providers. “Exit fees were just a symptom of a deeper problem: central storage that cannot keep up with a decentralized high -speed AI ecosystem,” he said.
Most of Tigris’ 4,000+ customers are like Fal.Ai: Genetic models of creation, video and voice models, which tend to have large sets of delay sensitive data.
“Imagine talking to an AI agent who makes local sound,” Tariq said. “You want the lowest delay. You want your computer to be local, close, and want your storage to be local.”
Large clouds are not optimized for AI workload, he added. The flow of huge sets of data for training or real -time conclusions in many areas can create a congestion of latent period, slowing the performance of the model. But access to local storage means that the data is recovered faster, which means that developers can perform AI workload reliably and more efficiently using decentralized clouds.
“Tigris allows us to escalate our workload to any cloud by providing access to the same data file system from all these parts without charging,” said Fal’s Taskaya.
There are other reasons why companies want to have data closer to their distributed cloud options. For example, in high regulated fields such as funding and health care, a large barricade for the adoption of AI tools is that businesses need to ensure data security.
Another motivation, says Tariq, is that companies want more and more to possess their data, showing the way Salesforce earlier this year blocked his opponents from using Slack data. “Companies are becoming increasingly aware of how important the data is, how LLMS feeds, how AI supplies,” Tariq said. “They want to be more under control. They don’t want anyone else to control him. ”
With fresh funds, Tigris intends to continue to build data storage centers to support the growing demand – Tariq reports that the start has increased 8 times each year since its founding in November 2021. Tigris already has three database centers in Virginia, In both Europe and Asia, especially in London, Frankfurt and song.
