Wasabi and Megaport integrate S3 storage and private network for AI

Wasabi Technologies will provide the object storage layer of Megaport Storage as part of a new partnership focused on AI infrastructure, analytics, and hybrid cloud. The proposal will combine S3-compatible storage, private connectivity, and, through Latitude.sh, computational capacity with CPU and GPU within Megaport’s catalog.

The key points of the Wasabi and Megaport partnership in 20 seconds

• Wasabi will supply the object storage for Megaport Storage.
• Megaport will add high-speed private connections to Wasabi’s 16 regions.
• Latitude.sh provides the compute layer with CPU and GPU.
• The offering targets AI data, analytics, backups, and disaster recovery.
• Not all regions, configurations, or SLA details have been disclosed yet.

The service will be available through Megaport’s platform in supported regions. Although the announcement from 07/21/2026 does not list initial locations nor specify when all 16 of Wasabi’s regions will be connected, the companies indicate ongoing development of technical integrations and new joint offerings.

This collaboration is not starting from scratch. Wasabi and Megaport already allowed for private connections between data centers, cloud environments, and Wasabi’s storage regions. The change now is that this relationship extends beyond connectivity: Wasabi becomes part of the storage solutions offered under Megaport Storage.

A platform with computing, networking, and storage

Megaport began as a network-as-a-service provider, offering private connections that can be configured via their portal or API. With the acquisition of Latitude.sh, completed on 11/27/2025, they added dedicated servers, virtual machines, and on-demand GPU capacity. Megaport Storage, launched in June 2026, completes the trio: storage directly connected to this network.

The partnership with Wasabi helps define who will provide part of this layer. The announcement specifically refers to cloud object storage, not Megaport’s entire storage portfolio.

Currently, Megaport offers standard and high-performance object storage, along with block and file options. Wasabi’s statement does not specify that its technology will support all these modalities. Instead, Wasabi is presented as the general object layer for AI data, data lakes, content, cloud applications, backups, and disaster resilience.

This distinction matters because object storage does not replace parallel file systems or local NVMe storage installed next to GPUs. It’s suitable for preserving datasets, documents, videos, results, backups, and model artifacts. However, certain training or inference phases require low latency and access patterns better served by block, high-performance file storage, or local drives.

The proposed architecture allows objects to remain in Wasabi and be transferred over Megaport’s private network to Latitude.sh servers or other platforms. The Australian company claims its storage connections can reach speeds of up to 100 Gbit/sec, though actual speeds depend on location, contracted ports, and service limits at each end.

Wasabi operates 16 regions across North America, Europe, and Asia-Pacific. Megaport spans over 1,100 data centers in 31 countries, while Latitude.sh offers compute in over 20 locations. Their coverage is not identical, so each deployment will need to verify where storage, GPU capacity, and private connectivity coincide.

A neo-cloud composed of specialized services

Wasabi and Megaport present this alliance as part of a neo-cloud infrastructure. The term describes providers building services specifically geared toward AI workloads, typically with access to GPUs, high-speed networking, and storage systems different from those of general-purpose clouds.

This isn’t about building a new hyperscale from scratch, but about assembling specialized services:

• Megaport automates the connections.
• Latitude.sh provides compute capacity.
• Wasabi manages object storage.
• The customer can run other applications on their data center or across multiple cloud providers.

This approach addresses one of the common challenges in AI architectures: the model can run in one location while data is distributed across private facilities, SaaS platforms, backup repositories, and various clouds. Moving these volumes over the public internet can impose bandwidth limits, variable latency, and unpredictable costs.

A private connection offers a more controlled route but does not automatically unify all services into a single platform. Different accounts, permissions, contracts, locations, and support cycles remain. The integration will need to demonstrate how well Megaport can mask this complexity for the customer.

The announcement speaks of a “single automated experience,” although technical documentation available before launch still describes a multi-step process. Customers must provision object storage through the Latitude.sh portal and then create a private virtual connection on Megaport to that destination. The integration of Wasabi may alter that process, but updated technical guides have not yet been published.

Megaport currently separates storage and network billing. Their documentation states that the virtual connection, known as VXC, is billed separately from capacity consumption. The commercial page shows standard object storage starting at $8.49 per TB per month, with no egress, API, or retrieval fees; however, connection charges depend on market, speed, and contract length.

Therefore, “no egress charges” does not mean data can be moved freely at no cost. While storage may not incur egress fees, the customer still pays for private connectivity, ports, contracted capacity, and infrastructure for data transfer.

AI, backups, and resilience will share the same layer

The partnership isn’t limited to training models. Wasabi and Megaport mention data lakes, analytics, multimedia repositories, cloud applications, disaster recovery, and backups.

This set of uses makes sense because AI data doesn’t always need to stay in the most expensive storage tier. A company might keep historical archives and less-accessed datasets in object storage, transfer some to high-performance layers during processing, and then return results to the main repository.

S3-compatible storage can also serve as a destination for backup tools. Megaport announces object locking features to prevent modification or deletion during retention periods. This control can help against ransomware and accidental deletions but doesn’t replace comprehensive backup policies, credential separation, and restore testing.

Private networking doesn’t eliminate all risks. Traffic no longer traverses the internet route, but data still depends on identity configurations, bucket permissions, keys, encryption, and access policies. Excessively privileged credentials can cause issues even if the network is isolated.

Before adopting this setup, organizations should understand at least:

• Supported regions and physical data locations.
• Availability and connectivity SLAs.
• Contracted durability and uptime guarantees.
• Options for geographic replication.
• Behavior during private connection outages.
• Total cost of storage, ports, and transfer.
• Support for object locking and regulatory retention.
• Identity management across Megaport, Latitude.sh, and Wasabi.
• Effective bandwidth between storage and GPU.
• Procedures for data egress or provider change.

This proposal is especially appealing to companies that prefer not to concentrate network, compute, and storage in a single hyperscale. However, working with multiple specialists shifts dependency toward integration capabilities and data mobility when needs evolve.

Wasabi and Megaport are crafting a modular alternative for heavy data workloads. Compute can be activated on demand, the network links environments, and storage grows independently. Ultimately, the success of the platform depends on whether this combination is managed as a unified service and whether its overall cost remains predictable, as both companies promise.

Frequently Asked Questions

What will Wasabi contribute to Megaport Storage?

Wasabi will provide the cloud object storage layer, used for AI data, analytics, applications, content, backups, and other S3-compatible repositories.

Does Megaport also offer GPU-enabled servers?

Yes. Megaport added CPU and GPU compute capabilities through the acquisition of Latitude.sh, completed in November 2025.

Will all Megaport storage use Wasabi?

The announcement confirms Wasabi’s participation only in the object storage layer. It does not indicate support for high-performance block or file modalities.

Are data transfers free?

The offering states no egress, API, or retrieval fees for object storage. However, Megaport’s private connection may incur costs depending on location, bandwidth, and contract duration.

via: wasabi