IREN scores a $9.7B partnership with Microsoft to fuel global AI expansion

Microsoft taps IREN to unlock fresh AI power

Microsoft has just signed a five-year cloud capacity agreement with Australian data center operator IRE, worth approximately $9.7 billion, giving it priority access to NVIDIA GB300-class AI accelerators.

The goal is simple yet huge, easing a computational crunch that has been slowing Microsoft’s ability to fully monetize the demand for generative AI across Copilot and Azure.

Markets loved it, sending IREN shares sharply higher as investors priced in years of contracted hyperscale revenue.

IREN reinvents itself as an AI cloud heavyweight

If IREN sounds new to you, that may be because it previously focused on bitcoin mining and is now transforming into a GPU-based AI cloud infrastructure provider.

The company now operates approximately 2,900 megawatts of largely renewable-powered, grid-connected capacity across North America, purpose-built for dense GPU clusters.

That footprint and power access are exactly what hyperscalers crave. This Microsoft win crowns IREN’s transformation from niche miner to mainstream infrastructure player and instantly makes Redmond its largest single customer.

The partnership is built around the next wave of Nvidia chips

At the heart of the deal is guaranteed access to Nvidia’s 300 GB GPUs, the successors to the H series chips that currently dominate AI training.

IREN will procure the processors and supporting hardware through a separate $5.8 billion agreement with Dell, then expose that capacity to Microsoft as a GPU cloud.

For Microsoft, this is akin to reserving priority seating on the next-generation AI hardware train, rather than scrambling for leftovers on the open market.

This deal helps Microsoft dodge the data center bottleneck

Building hyperscale AI campuses from scratch takes years and massive power permits. By leveraging IREN’s existing sites, Microsoft can expand its AI capacity without waiting for new land, power hookups, and construction.

It also avoids front-loading even more chip capex onto its own balance sheet at a time when quarterly capital spending is already approaching $35 billion.

Instead, Microsoft treats much of this as contracted cloud spend, smoothing both timelines and accounting optics.

Childress, Texas becomes a new hub for frontier models

Most of the new Nvidia hardware will be located at IREN’s Childress campus in Texas, a 750-megawatt site engineered for AI and liquid cooling.

The company plans a phased deployment of the GB three hundred systems through 2026, with new liquid-cooled halls delivering roughly 200 megawatts of critical IT load.

Dell quietly supplies the hardware muscle behind the scenes

This story is not just about Microsoft and IREN. Dell plays a crucial third role as the integrator, supplying Nvidia’s GeForce 300 platforms and supporting gear.

IREN’s separate $5.8 billion purchase from Dell effectively pipelines high-end servers straight into Microsoft’s hands. That structure enables Microsoft to tap into cutting-edge systems without having to negotiate directly with every supplier.

At the same time, Dell enjoys a multi-year revenue stream tied to one of the most visible AI capacity expansions on the market.

Prepayments and timelines define the financial backbone

The contract spans five years and includes a twenty percent prepayment from Microsoft, credited back against later usage. That early cash helps fund IREN’s Dell purchase and buildout, but it comes with teeth.

If IREN misses delivery milestones, Microsoft can terminate the deal, a reminder that execution risk is a genuine concern.

Once fully ramped, analysts estimate the agreement could generate nearly $1.9 billion in annualized revenue for IREN, dramatically reshaping its financial profile.

Neocloud providers reshape who controls AI compute

IREN joins names like CoreWeave, Nebius, and Lambda in a new class of AI-focused cloud providers, often referred to as neoclouds.

Instead of competing with Microsoft or Amazon at the software layer, they specialize in power, land, and GPU density, then wholesale capacity back to hyperscalers.

Microsoft has already made significant GPU commitments with Nvidia and AMD, as well, spreading its bets across multiple partners. IREN’s win demonstrates the growing centrality of this outsourced model in hyperscale AI planning.

IREN capacity and renewable power become a strategic asset

One underrated angle here is power. IREN has secured nearly three gigawatts of grid-connected, largely renewable energy across its sites, a feat that even tech giants struggle to achieve quickly.

That combination of clean power, rural land, and transmission access turns into strategic leverage when AI demand is soaring.

For Microsoft, plugging into that portfolio accelerates its own sustainability and capacity goals. For IREN, it proves that the energy first site selection was precisely the right long-term bet.

Microsoft stretches its capital further by renting power

This deal reflects one example of Microsoft leveraging external capacity contracts alongside its own infrastructure build-out. That approach converts some capex into operating style commitments while still guaranteeing access to scarce GPUs.

It also reduces the risk of being stuck with older hardware once Nvidia’s next generation arrives. In effect, Microsoft is renting time on someone else’s fast-depreciating assets while focusing its own spending where it gains the most strategic leverage.

What this partnership could mean for everyday users

For most of us, this deal will not show up as a new button on screen, but it underpins everything from smoother Copilot sessions to more reliable chatbots embedded in everyday apps.

Extra compute means fewer slowdowns during peak usage, faster rollout of new AI features, and more room for experimental products that were previously too resource-intensive.

In a way, the Childress campus might quietly power the next wave of tools you rely on at work and at home.

Learn how security experts uncovered new risks tied to Microsoft’s AI tools in Microsoft Copilot Studio agents, which were exploited to steal tokens, experts warn.

IREN, Nebius, and Lambda.

Step back, and the pattern becomes clear. Tech giants are no longer just building their own data centers; they are stitching together a global network of specialized partners, such as IREN, Nebius, and Lambda.

Power access, chip supply, and regulatory permissions are now as strategic as software features. This latest Microsoft IREN pact is not the end of the story, but a strong hint that the real winners in AI may be the companies that master infrastructure on a planetary scale.

Discover how Microsoft’s latest $20 billion partnership is reshaping its cloud strategy as the company enters a cloud partnership with Nebius worth nearly $20 billion.

What do you think about IREN’s partnerships with Microsoft to expand AI globally? Please share your thoughts and drop a comment.

Read More From This Brand:

• Microsoft OpenAI Partnership Under Review
• Meta’s partnership with Scale AI shows early signs of strain
• Broadcom soars after partnership with OpenAI to build new AI chip

Don’t forget to follow us for more exclusive content on MSN.

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one.

This slideshow was made with AI assistance and human editing.