Is Wasabi Fire Key to Enabling the Next Major AI-Era Storage Platform?

Wasabi says its 100,000+ customer mix has shifted, now including large-scale AI data lakes. In our recent conversations with company leadership, they shared examples including a computer vision provider storing 30 billion images and petabytes of training data, and a multimedia AI platform storing raw video footage and derived artifacts. In these cases, Wasabi serves as the persistent storage layer, allowing organizations to scale their data without being locked into a single hyperscaler's billing model. Let this resonate: Wasabi now operates at multi-exabyte scale with a global footprint spanning sixteen regions, placing it in the upper tier of cloud object storage platforms outside the major hyperscalers.

The timing of Fire’s arrival also lines up with a structural shift in AI infrastructure buying behavior. Enterprise teams are beginning to separate storage decisions from compute decisions as GPU scarcity forces them into multi-cloud and hybrid deployments. In this environment, object storage has to act as the durable backbone across increasingly distributed pipelines. Fire positions Wasabi to benefit directly from this decoupling trend, serving as a vendor-neutral performance tier that can follow the data wherever compute must land.

Seen through that lens, the Wasabi Fire service is less of a departure and more of an extension. Until now, Wasabi’s storage platform has primarily relied on HDD-based capacity, which works well for long-term retention and many production workloads, but is not always ideal for the front end of AI pipelines. The early stages of model development, especially for vision and multimodal systems, can be dominated by small-object reads, random access to frames inside video files, and repeated passes through raw and lightly processed datasets. Those patterns benefit from lower latency and higher throughput. Fire is designed to move Wasabi into those stages of the pipeline by placing data on SSD/NVMe, while preserving the S3 compatibility and flat-rate billing that customers already understand.

In our view, the decision to launch Fire first in a San Jose region inside an IBM Cloud data center is strategically consistent with how Wasabi has approached the market. The company has never tried to build its own compute service. Instead, it is colocating storage with partners that already run compute and GPU infrastructure and negotiating economics that remove friction. In San Jose, customers can place Wasabi Fire storage and IBM Cloud compute in the same facility and move data between them without egress fees, connected through cross-connects in the Meet-Me-Room. For AI teams that have grown weary of paying to move their own data inside and out of hyperscale platforms, this is significant.

The company clearly expects a step-function performance improvement over its HDD-backed tier, but during the controlled introduction it is holding back from publishing throughput, IOPS or latency guarantees. High-performance object storage is a category where expectations are high and the workloads are varied. Taking time to tune behavior with a small set of early customers before translating those lessons into public commitments is a disciplined approach.

From our perspective, the Fire launch and the San Jose region indicate that Wasabi is stepping more directly into the AI infrastructure conversation in a way that fits its identity. The company is not trying to match the breadth of AWS, Azure or Google Cloud. Instead, it is sharpening its role as a storage foundation that can sit under AI data lakes, lakehouse pipelines and vector databases while keeping cost and complexity in check. Fire gives Wasabi a way to support more of the AI lifecycle, from long-term retention of raw datasets to performance-sensitive preparation and embedding generation.