Decentralized cloud-based storage company Storj Labs Inc. said today it has acquired the graphics processing unit cloud provider Valdi Labs PBC.

The deal means its customers will be able to access on-demand GPU capacity to power their artificial intelligence workloads whenever they require it.

Storj is pioneering the concept of more affordable and secure decentralized storage. The company doesn’t own any server hardware, but instead rents spare storage capacity from individuals and organizations across the world, before making this space available to its customers.

Through this model, Storj reckons it can provide access to highly distributed cloud storage that’s up to 90% more affordable than renting storage capacity from traditional cloud infrastructure providers such as Amazon Web Services Inc. and Microsoft Corp. What’s more, it offers the same availability guarantees and promises its performance is at least as good, if not better than those rivals.

Storj, which is headed by its Chief Executive Officer Ben Golub, minimizes the obvious risks associated with distributed and decentralized cloud storage by encrypting its customers data before sending it to its rented servers. Moreover, it breaks each piece of data into about 80 different fragments, which it calls “pieces,” which are spread across various different servers and storage drives around the world.

This ensures iron-clad security, as any would-be hacker is required to hack into numerous different computer systems spread across the globe to gather all of the pieces required to decrypt customer’s data and make sense of it. Storj says at least 30 pieces are required to put customer’s files back together.

That not only makes the system highly secure, but also solves the problem of accessibility. As many as 50 of the 80 underlying storage systems used to store the data could go offline, as the data would still be accessible, so long as the other 30 are still up and running.

The company says it’s seeing big demand for its decentralized storage services due to the rapid growth of the AI industry, which has an insatiable thirst for training data. AI training is done using hundreds of powerful GPUs, but it also requires a powerful and reliable storage platform to store the data that’s used to train the AI models. As such, it Storj believes it makes sense to provide both GPUs and storage in a single, reliable platform.

That’s why it has decided to acquire Valdi, the creator of a distributed GPU cloud platform. Valdi makes its 16,000-strong network of GPUs available on-demand and at massive scale, and claims to offer some of the most competitive rates in the industry.

By acquiring Valdi, Storj says, it will be able to offer an integrated cloud storage and GPU service. It says the combination is ideal because most AI workloads demand significant GPU compute, storage and movement of large datasets. With Valdi, customers can access GPUs instantly, with no contracts required.

“It expands the value of our distributed solutions for the enterprise, enabling a full-stack cloud offering for our wide range of global customers,” Golub said.

Holger Mueller of Constellation Research Inc. said Storj is following the tried and tested strategy of bringing data to the AI applications, which makes sense given it controls access to massive amounts of customer’s data. “Proximity to, and availability of GPUs is becoming more important than ever when it comes to the criteria enterprises use to choose a storage platform,” Mueller explained. “For companies building generative AI applications, having integrated GPUs and storage means a lot less complexity.”

Storj claims that its Amazon S3-compatible distributed object storage and Validi’s high-performance GPUs will be “infinitely scalable and rapidly available” in any location in the world. Its customers will have the flexibility to spin up as many GPUs as they require, or scale down during periods of lower usage.

Valdi CEO Nikhil Jain said the company’s customers have substantial yet highly flexible storage needs that they sometimes struggle to meet. “So uniting our organizations delivers benefits to customers while empowering us to grow faster,” he said.

Image: SiliconANGLE/Microsoft Designer